1 • What are “human-data interactions”? Another short-lived buzzword?

Human-Data Interaction is topic which aims to put the human at the centre of the data driven industry.

It is not a buzzword, on the contrary. More than a technological advancement, Human-Data interactions address long lasting issues caused by a new social and cultural context.

The digital age has brought along a situation of information overload, or infobesity. Our screens are increasingly flooded with data about our work, our processes and ourselves.

Human-Data interactions refer to connected interfaces that everyone can easily engage with, in order to better understand, to inquire, share and work from data.

This is dramatically changing the way we learn, the way we apprehend our environment, the way we govern our cities, the way we interact with others and even the way we know ourselves.

2 • Why did you choose to specialize in Human-data interactions?

We perceive there is a need to radically transform the way organizations — either public institutions or corporations — understand the value attached to data, and how they implement solutions to unleash their potential.

Data has become a very valuable currency in the digital economy, thus making it vital to have interfaces allowing to understand, process and share data.

The need for comprehension remains widely unsatisfied, for two main reasons. First of all, data does not interpret itself, which results in everyone not being able to fully process what the message is. Besides, while this mass of data could reveal extremely useful to better understand our environment, data too often remains in silos, scattered in different repositories which make it difficult to access and connect.

This lead us to a strong conviction that unveiling the value and the information inside massive data set is a great challenge to tackle.

3 • What added value does your work bring to a client?

Dataveyes’ approach does not consist in setting up yet another data analysis tool for the companies it works with.
In reality, Dataveyes strives to help their clients better extract value from their data, preparing them to the new data ecosystem.

We design and produce advanced human-data interaction solutions, bridging the widening gap between abstract data and the various needs around data : from better comprehension to better innovation.

This long term initiative not only requires the advanced interfaces we design; it needs our ability to involve the client all along a project.

This can be done by carefully listening to their issues, understanding their business purpose, analyzing their processes and organization structures, conceiving interfaces that efficiently answer their needs and testing the most relevant information experiences

4 • How are you different from business intelligence tool providers?

We believe that the true potential lies at the fingertips of end users, rather than within data centers.

We perceive our mission to be at the very end of the data value chain, in close touch with users : this is where data-driven strategies are implemented, and where organization changes can be witnessed.

• We always keep the final user in mind. Our ambition is to design bespoke interfaces, which will greatly help users grasp, process and understand the data.
• We favour short, agile work cycles. They allow the creation of more mature tools on an ongoing basis, rather than a long tunnel of development which can divert us from the intricacies of the data.
• We spend a great part of our time creating interfaces meeting non-standardized user needs. This is how we guarantee long term relevance. We also develop our own proprietary technological components, to capitalize on our research works.
• We will allow the whole organization to improve its data literacy. Along our creation and production process, we want to ensure that all stakeholders feel more comfortable understanding the data, not only the people in charge of the tool’s administration.

5 • What kind of company are you at the end?

Our way of working consists in developing operational interfaces.

But we do not define ourselves as a design studio, nor as a software publisher.

Our clients can rely on our vision and strategy, but we do not define ourselves as a consulting or as a communication agency.

Even if interfaces are the result of our work, they are not the only asset we can bring to our clients. We perceive each of our deliverables as a step towards the construction of a robust ecosystem which will reveal insight behind the data.

The data deluge is challenging the way organizations define their process and their value. It therefore makes it tricky for a company like ours to tick traditional description boxes.

Our job is a work of mediation between data and users: by mobilizing the right mix of skills, we think of interfaces as information experiences, capable, through their interaction strategy, of making into reality new uses of data.

Do you have any other questions for us? Write to us at contact@dataveyes.com or consult our projects on our site.

What are you doing? What do you sell? Some questions you may have for us was originally published in Dataveyes Stories on Medium, where people are continuing the conversation by highlighting and responding to this story.

In 2015, we created a generative art application for Société Générale. Realized in collaboration with the Magic Garden agency, the project gave birth to 24,000 original creations. It answered a challenge: using algorithms as raw materials for a public communication operation.

USING ALGORITHMS AS CREATIVE RAW MATERIALS

Generative art is a branch of digital art in which algorithms are used as creative raw materials. Thus, in a work of generative art, a computer program transforms data and signals into aesthetic patterns. A distinctive feature of this program is that it includes a “random” part: signals are never translated the same way. In this feature lies its poetry: generative art adds a measure of irrationality and uncontrollability to the usual mathematical coldness of data.

This is probably why this artistic trend is now more resonant than ever: in a world where we’re inundated with data on a daily basis, generative art grants our wish to still see fortuity and uncertainty have their place.

The “Créa-Carte” operation, carried out in 2015 for Société Générale, gave us the opportunity to validate this intuition, and to demonstrate that the general public would warmly welcome this generative aesthetic.

Créa-Carte is a collaborative project on generative art: it invites visitors to give an input gesture, from which a 3D aesthetic form will be generated. This individual creation was then added as a collaboration to the flow of all contributions recorded from the start of the operation.

The application contains an algorithm that transforms the participants’ gesture data into harmonious 3D objects, and adds a bit of randomness to the generation process.

This “algorithm-assisted creation” experience was conceived for Société Générale’s 2015 contest to create credit card visuals. It is the result of six months of work involving the entire Dataveyes team. It has allowed us to develop new technologies and to define efficient arbitrations between a non-expert public and the data world.

The present case study describes how we mobilized the aesthetics of data to offer a wide-audience, interactive experience.

THE METHOD: HOW DID GENERATIVE ART ENABLE US TO MEET THE REQUIREMENTS FOR A WIDE-AUDIENCE OPERATION?

At the origin of this project is the intuition of Magic Garden agency, in charge of organizing the competition of credit card visuals creation for Société Générale: the latter saw in generative art a promising creative concept, capable of giving renewed momentum to the third edition of this competition, targeting a very wide audience.

The subtle blend of art and algorithms appeared as a solution to reach several goals:

• Allowing non-experts to create aesthetic and original visuals;
• Encouraging numerous contributions, each with a unique result;
• Reaching a wide audience with a stimulating experience;
• Adding a collective dimension to the individual creation process;
• Allowing creation on any device.

ALLOWING NON-EXPERTS TO CREATE AESTHETIC AND ORIGINAL VISUALS

The main challenge of this competition was to reach a wide audience, and go beyond the small circle of professional designers and creators. To do this, it was important to allow any user to feel like an artist… and therefore to propose a creative application capable of generating aesthetic and original visuals, whatever the level of inspiration of the participant or their skill on digital media.

Upon launch of the project, we devised several generative approaches to meet the objectives of the project.

The second approach, featuring a “revolution of matter” concept, was retained, because it created a direct link between the gesture drawn by the participant and the aesthetic rendering wrapped around this initial layout.

The generative part offered an efficient solution to obtain consistently aesthetic results. Indeed, generated visuals are not made by the participant only: an algorithm intervenes to transform the initial gesture of each contributor into a 3D object.

Once the “revolution” approach was validated, our mission was then to calibrate the algorithm so that it could transform any gesture, even the most absurd one, into a fine-looking shape.

To that end, a designer and a developer worked hand in hand for several days, combining the projection capabilities of a 3D design tool like C4D, and the building capabilities of WebGL. By synthesizing their views and sensitivities, our lead designer and lead creative coder put together a computer program that was sufficiently powerful and poetic to generate ever-harmonious 3D shapes.

ENCOURAGING NUMEROUS CONTRIBUTIONS, EACH WITH A UNIQUE RESULT

The 3D shape generation program had to produce results that were sufficiently varied from one participation to the next, to give contributors the impression they were the authors of a unique creation, and to encourage them to participate several times.

To that end, we introduced a strong random ingredient in the algorithm that would generate 3D shapes. Two parameters were involved in this algorithm, which was used to wrap the participant’s gesture in matter: the angle of revolution, and the height of the translation between the revolution’s starting and ending points. These two parameters were chosen randomly within ranges of values we had previously defined, so that the result would always appear aesthetic and unique even when the user performs an identical gesture.

Then, the 3D object’s gradient colors and the background color are randomly set within a palette specifically composed for the operation. The eighteen colors composing this palette were designed to blend harmoniously whatever the quartet randomly chosen.

Finally, once the 3D object was generated by the algorithm, the user could then customize their creation by modifying its colors, material and transparency.

Given the 18 x 18 x 18 x 18 color combinations, the three available materials and the ten selectable levels of transparency, more than three million unique customization combinations were possible, in addition to the already infinite number of possible shapes generated by the algorithm’s random part. In other words… it was impossible to get the same result twice!

REACHING A WIDE AUDIENCE WITH A STIMULATING EXPERIENCE

This was probably the greatest challenge in the project: get a wide audience interested in generative art and thus turn a topic that may seem esoteric into a playful experience. To achieve this, we focused on two objectives:

• create an experience that is attractive enough to be self-sufficient;
• eliminate all obstacles to user interaction and progression in the experience.

First, we centered user path on the creation area: all navigation options in the application converge towards it. Within this creation area, we encouraged creativity by allowing users to do, undo, redo, test, customize, and so on. We thus made it possible for the user to get enjoy a stimulating experience: seeing a simple gesture turn into something tangible and fine-looking.

In order for this experience to reach as wide an audience as possible, we hunted down everything that could hinder interactions. We ran usability tests on the application to identify and correct small frictions that may discourage users.

These tests were carried out using the “voice aloud” formative test method (see detailed presentation of our test methodology in this previous article). In this method, a candid user is confronted with concrete situations, in order to get them to perform tasks on a prototype of the application. The tester is asked to verbalize what they are doing and what they see on the application.

These tests enabled us to correct specific problems of usability and ergonomics.

ADDING A COLLECTIVE DIMENSION TO THE INDIVIDUAL CREATION PROCESS

To echo Société Générale’s brand signature — “Team spirit” — it was decided that the operation would not be limited to an individual competition for the creation of credit card visuals, but would also include a collaborative dimension.

We thus transformed individual gestures into acts of co-creation: after having created and customized their 3D particle, each contributor was invited to add this piece to a collaborative work composed of the particles of all previous participants.

Here we also mobilized our revolution algorithm, as a cement between individual and collective levels of the application. To plot the collective work, we calculated the mean gesture of all participants, and used it as an input for the algorithm, in order to obtain a 3D skeleton that embodied the community, skeleton which was then populated by the miniatures of individual creations.

As a result, each new participant was likely to influence the outline of the collective form by their gesture. This was, of course, particularly true for the first contributors, when any additional plot had a proportionally stronger impact. Then it was less and less visible once the number of participants increased, and individual creations only marginally influenced the collective average.

In conclusion, each new individual creation contributed to the collective work in two ways:

• by bringing an extra piece into the mesh making up the collective work,
• and by influencing the average trace determining the 3D form of this collaborative work.

This close intertwining of individual and collective levels embodied very well the notion of “team spirit” promoted by Société Générale, but required the implementation of advanced technical solutions.

Indeed, by postulating that the competition would attract tens of thousands of participants (which it did), displaying all 3D creations side by side within the collective work presented several difficulties. First, this could create visual noise and interfere with the readability of the interface. Secondly, this posed a serious performance problem: our browsers are far from being able to display so many complex 3D objects at the same time, all the while retaining the ability to interact with them.

We therefore decided to draw the collective work from simplified forms: identical red dots, each symbolizing an individual creation and its precise positioning within the collective work. When the user zooms on the collective work, the closest red dots fade and the original creations are revealed.

Still for performance reasons, we could not display these creations as 3D objects: the collective mesh actually featured their reflection in 2D. We thus had to produce a 2D image of each individual creation.

To ensure integrity of the image database, we could not render 2D images in the browser. So we used Atom Electron to generate them server-side. This technology allowed us to embed a programmatically manipulated browser on a rendering server, thus enabling us to automate the transformation of WebGL code from 3D creations into 2D images.

To guarantee the robustness of this solution, we supplemented it with a queue management system, RabbitMQ, which ensures that the image renderings are done one at a time, with no risk of server overload or data loss.

Finally, given the resources mobilized server-side by this image generation function, and the database size, we opted for a technical architecture articulating three servers:

• a front-end server, to manage the application to which the user connects;
• a database server, to manage the contributions database;
• an image rendering server, equipped with a graphics card.

This robust and virtuous architecture had the advantage of being scalable, and could overcome potential load increases: RabbitMQ supports multiple communication channels, it was thus possible to set up new image rendering servers, to manage an increased inflow of participations.

ALLOWING CREATION ON ANY DEVICE

In order to reach the largest number of users, the application required compatibility with a wide range of devices: computers, tablets and smartphones.

Visuals and user path automatically adapted to both the platform’s screen resolution and performance.

3D rendering was developed in WebGL and worked smoothly on computers, but also recent tablets and smartphones. For older smartphones, a Canvas version was able to display a slightly less aesthetic, but less resource-heavy, 3D rendering that was thus compatible with a greater number of devices.

Finally, the experience was also made available in the physical universe, as a demo version for terminals in Société Générale agencies. A dozen of them, located in city centers, overlooking a busy street, displayed the latest creations of the contest’s participants. From the street, passers-by could interact with these 3D shapes thanks to a webcam that detected their movements.

TAKEAWAYS: WHAT DID WE LEARN FROM 24 000 USERS ?

Review of the operation

The operation was launched on May 19, 2015. The participants then had 3 weeks to submit their most beautiful generative art visuals to the credit card contest. Once the winners were announced on June 9, 2015, the experience was perpetuated for a year: visitors to Société Générale’s “Concours creacarte” website could still create and personalize their 3D particle and submit it to the collective work , but also visualize works created by all previous contributors.

More than 13,500 participants were registered during the first 3 weeks of the competition, which represented a significant increase in the number of entries compared to previous competitions.

The project was rewarded with an Awwwards on June 4th and an FWA on July 21st. These two design awards helped revive visibility of the project, and thus traffic on the dedicated website.

A study of the numerous tweets and Facebook posts related to the project shows that the experience has reached a wider audience than the only circle of digital design lovers.

One year after its launch, the website counted more than 24,000 particles.

User preferences

Contributions in themselves constitute an interesting data set describing the gestures made by the participants as well as their color choices.

Red, black and yellow were among the preferred colors of contributors, who, on the other hand, shunned the pastel colors. The fact that red and black are Société Générale’s signature colors may have something to do with this particular result.

The gestures traced on computer, tablet and mobile present a very wide range of shapes: very minimalist traits, signatures, figurative drawings, etc.

Following is a ranking by shape similarity for the first 2000 participants. Please see: http://sg.sorting.dataveyes.com.

CONCLUSION: WHY A BANK BRAND SHOULD CONSIDER GENERATIVE ART

This experience of generative art carried out alongside Société Générale has proved that it is possible to reach a wide audience by proposing an interactive experience where data aesthetics comes first. In the background, this original project allowed Société Générale to demonstrate to the general public that algorithms and data can be domesticated and put to the service of the user. And such a demonstration holds strong value for the banking sector.

WATCH THE VIDEO

Want to know more? Get yourselves ready for our next articles by subscribing here on Medium.

You can also follow us on Twitter or discover our last use cases on our website.

A multi-media 3D experience to raise general awareness of generative art was originally published in Dataveyes Stories on Medium, where people are continuing the conversation by highlighting and responding to this story.

During the summer of 2014, we worked with a team of researchers and designers at EDF R&D on a project around an educational simulator. This project was intended to introduce the concept of energy flexibility and explain how it can potentially make our energy consumption more sustainable.

TURNING CONSUMERS INTO CONSCIOUS ENERGY MANAGEMENT STAKEHOLDERS

Energy is a fast-changing sector. The development of renewable energies, along with the development of smart grids, is changing our traditional model. We are moving away from a centralized, one-way structure, towards a more local energy management system, down to the individual scale and based on decentralized production modes.

In this new model, the consumer, either a citizen or an economic player, is invited to play a much more active role. Learning thermostats (such as Nest) and smart meters, help optimize electricity consumption on a daily basis. Besides, domestic batteries, such as the prominent Tesla PowerWall, enable to store energy that consumers can use when they please.

Those new technologies have brought along a variety of expectations, such as being able to compare home appliances, lower bills while not compromising on comfort, reduce load curves or soften environment impact… However, fulfilling those expectations requires a robust technical knowledge. For the general public, the subject is all the more difficult to understand as electricity is a phenomenon which is by nature invisible, and tricky to conceptualize. Not to mention that the relevant information is mostly available in a data form, which is difficult to decipher.

Nonetheless, the sustainability of our electricity system depends on our ability to better control the demand side. Sustainability thus requires a broader awareness of those new devices, and of the level of flexibility they can bring to networks.

In this context, how can citizens be empowered to be involved in these new smart systems? How can they gain the capacity to better understand, and thus better manage their electricity consumption?

The need for a better mediation between data and users is particularly anticipated. To ensure an educational approach, and make new intelligent devices easier to use, people need interfaces that extract meaning from the data and make that data both usable and useful.

This was the beginning of our data visualization project, conducted in collaboration with EDF R&D. The below use case shows how we built a visual and interactive application to place users into a situation where they can manage energy flexibility.

BEYOND AWARENESS, AN ACTIVE UNDERSTANDING OF THE DATA

The concept of energy flexibility refers to the ability of our domestic electrical installations to modulate their consumption during the day. This way, they can lower their consumption during peak hours or smooth out consumption peaks to put less pressure on the global network. (A more detailed explanation of this concept will be provided later on in this case study.)

This concept of energy flexibility is particularly interesting in light of the above introduction, precisely because it allows citizens to make better decisions and control their appliances, which leads to a more sustainable consumption. In other words, energy flexibility is one way to empower consumers.

Our contacts at EDF R&D wanted to propose an educational demonstration of this concept, based on simulation data and a model of the electricity consumption across houses.

From the beginning, it was anticipated that the application could be integrated into a very large screen in EDF showrooms, and appeal to a a variety of visitors, from energy professionals, decision makers and opinion leaders.

We analyzed how such an application could mobilize the understanding of the showroom visitors, in a way that encourages them to take action, and does not simply raise awareness. To do this, we executed a data visualization application that allows visitors to understand the concept of flexibility for various home appliances and configurations. Users can compare the effects of flexibility through values such as level of electricity consumption, temperature or cost of bills.

Within this application, we created an experience articulated on three axes, to place the user in an active position:

1. Active understanding, with an « explorable explanation » approach in mind.
2. Enhanced interactivity via a visualization operable from a smartphone.
3. Playful dimension, based on a collaboration mode

1 — FEELING, RATHER THAN ASSIMILATING: THE EXPLORABLE EXPLANATIONS

The concept of « explorable explanation » was introduced by Bret Victor to describe a series of experiments around active reading and reactive texts.

According to Bret Victor, an active player possesses the ability to interrogate the text, consider various assumptions, question the coherence and reliability of the content, rather than passively assimilate information.

Victor observes that the typical format of explanatory content — articles, books, reports, etc. — rarely encourages such critical thinking, and often merely presents the arguments of the author.

Bret Victor expresses the wish that those content types do not convey information for mere consumption, but rather prove thought-provoking, and lead readers to think further, thus placing them in an active position.

« A reactive document allows the reader to play with the author’s assumptions and analyses, and see the consequences. »

Explorable Explanations / Bret Victor / March 10, 2011

Bret Victor presents various examples where interactivity, coupled with modeling and data, leads the reader to a more stimulating and intuitive understanding of physical, economic, or mathematical phenomenon.
Bret Victor’s « explorable examples » aim to bring readers to experience first hand complex models by playing with sliders, altering key figures or observing recurring patterns.

« An explorable example makes the abstract concrete, and allows the reader to develop an intuition for how a system works. »

Explorable Explanations / Bret Victor / March 10, 2011

Other people, in addition to Bret Victor recently embraced a similar approach, in order to create highly compelling educational scenarios. Here is an example of an « explorable explanation », conducted by Bret Victor.

To conceive an educational application about energy flexibility, we experienced the same need to mix text, interactivity features and data visualization to allow users to understand a complicated concept.

In the section below, we attempt to reproduce a short part of the demonstration implemented in the final simulator.

Let us start with a very simple principle: it is complicated to store electricity on a large scale.
Therefore, it is necessary to balance at all times the supply and demand on the French power grid, nationally and locally.

This balance is tricky to obtain:

• From the demand side, we collectively tend to consume electricity at the same times: in the morning before going to work, in the evening as we come home, on those very cold winter days, etc. The demand for electricity in France is thus characterized by strong variations which follow the seasons and hours of the day.
• From the supply side, however, it is very costly to vary the production capacity in such proportions. Nuclear plants cannot easily adjust their pace of production, and renewable energy (solar, wind, etc.) are by nature irregular.

Electricity consumption levels in France, every 30 minutes, in 2013:

The difficulty to instantly adjust supply and demand creates episodes of stress on the global electricity network. Those episodes are a source of risk and extra cost to all parts involved in the electricity system, producers and consumers alike.

For a long time, only the supply side was expected to adjust to the consumption peaks. Today, new technologies introduce a possibility for the demand side to be flexible as well. This can help reach a more stable balance and a smarter control of the national energy network as a whole.

It is very simple to schedule the running time of a washing-machine or a dishwasher, so that they consume electricity outside of the peak hours. However, it is not as trivial for heaters, air conditioning systems or water heaters, as interrupting them for too long would impact the comfort of a home.

The concept of energy flexibility can be better understood here: it describes flexible electrical systems able to regulate consumption levels depending on the times of the day, without ever altering comfort.
Flexible electrical systems thus have the ability to erase consumption during certain hours of the day, (i.e. by postponing consumption, or smoothing it out over time) without inhabitants perceiving any difference in the room or the water temperature.

All the heaters and air conditioning systems do not have the same flexibility capacity, as it depends on their underlying technology. A heat pump or micro-storage heater will likely be more flexible than a convection heater, in most homes.

The objective is to facilitate a better balance between production and consumption of power across the entire network. From the perspective of a consumer citizen, the goal is simple: run the heater or air conditioning device in the way that limits consumption during the peak hours as much as possible, without altering comfort.

Now, we can only imagine how things could look like if this adjustment mechanism could be automatically managed by those smart regulators bound to be increasingly present in our homes… This scenario helps perceive the interest of the energy flexibility concept, and its formidable potential to enable a better balance of the electrical network.

2 — COUPLE THE MOBILE PHONE WITH THE BIG SCREEN FOR MORE TANGIBLE INTERACTIONS

We tailored this demonstration to fit the specific context of the project executed for EDF R&D:

• The project was meant to be accessible as part of a showroom, where the visual attention of visitors would already be stretched. This would result in visitors having little time to read lengthy explanations.
• The project was meant to provide support for oral presentations to an audience of experts, decision makers and opinion leaders.

We therefore strengthened the visual and interactive dimension of this demonstration, to turn it into a captivating and entertaining application:

• We rolled up the load curves to make the main graphic look like a clock.

• We refined the illustrations to give a familiar feel to the application.

We chose to move the interaction points away from the big screen, onto the visitors’ or the presenter’s smartphones. Thanks to a pairing technology, users can take control over the big screen with the help of their smartphones.The smartphone interface thus acts a remote to select items or interact with the data visualizations displayed on the big screen. The interface on the big screen allows a large number of users to observe the visual representations and enjoy the explanations.

Visitors of the EDF R&D showroom can alter the information displayed on the bigscreen by moving the slider on the smartphone.

The pairing system is based on the WebSocket technology that enables bi-directional communications between a browser and a webin server. An node.js server orchestrates synchronization between the phone and the big screen. In order for them to communicate together, all devices are connected to the same wifi network.
This contactless interaction technology entices visitors to strongly engage with the installation, and makes it a highlight of their showroom visit.

3 — FROM GAMIFICATION TO EMPOWERMENT

A final lever enabled us to further establish the role of the user as an active player: the collaborative game.

The application includes a multiplayer mode in which several participants interact on the big screen. Together, using their smartphone, they take control over a small village and attempt to improve its electricity consumption.
The playful feel of this section, combined with the fact that it leads people to work collaboratively, reinforces the educational role of the project.

Each player can make their selection vary on their phone, to update the information displayed on the big screen.

After four months of work involving researchers, designers and developers, this application demonstrates that through three levers: the game, the visualization and the physical interaction, a large audience can understand complicated phenomenons and become players making better energy choices.

WATCH THE VIDEO

Want to know more? Get yourselves ready for our next articles by subscribing here on Medium.

You can also follow us on Twitter or discover our last use cases on our website.

An educational simulator to understand the concept of energy flexibility was originally published in Dataveyes Stories on Medium, where people are continuing the conversation by highlighting and responding to this story.

The new identity of Dataveyes originates from the relations between its members, literally and figuratively. It is dynamic, set in motion by our activity data, which best reveals how we all work together.

WHY A NEW VISUAL IDENTITY?

We recently unveiled a whole new approach to our identity: new website, new logo, new way of describing our business, and more generally, new visuals and terminology across all our media platforms. This project was ongoing for over a year before we finally found the tenacity — and time — to finalize it between November 2013 and January 2014.

This long moment of hesitation probably speaks of the difficulty we experienced to define our business. Big data, open data, data journalism, infographics, etc.: we were often associated with these terms, although we felt that none of them captures what we actually do.

Our work takes different forms: we create data visualization applications as well as ambient experiences, connected objects, or business dashboards. Our approach is as much about technology as it is about design. We put these two dimensions at the service of our users, with the technical ability to process large volumes of data. We are neither a software publisher, an agency, a consulting firm nor a research laboratory. We lie at the crossroads of them all.

How to visually embody so many paradoxes?

While working on a new way of describing our business, we laid down the foundations of a new visual identity.

Our old logo had done its time. At the beginning of Dataveyes in 2010, the logo allowed us to gain a foothold in our market by alluding to the Pie-Pacman reference. It was simple, light and friendly: the image of our young impetuosity to implement data visualization in France

Three years later, we felt ready to embark on a logo that reflects a more mature, explicit vision of Human-Data Interactions.

A DYNAMIC IDENTITY

After initially exploring graphics, we unanimously headed towards the option of a generative logo. That is to say, a logo whose shape is not settled, but is self-generated by an algorithm and operates by creating a pattern. It may be because we had a hard time envisioning our identity as something static, considering how we keep on learning new things on a daily basis.

Our field, data visualization, is recent and still largely unformatted: methods, techniques, approaches are constantly re-invented. The market we play in is expanding very quickly, and is still in the process of understanding its needs.

How to express this dynamic situation without making our identify confusing? The idea of ​​a generative logo brought an answer: it allowed us to create a coherent symbol, a frame, while leaving room for endless variations. Generative art is also a field that is not so distant from ours. Where we create interfaces fueled by data, generative art bases its creative approach on algorithms.

However, despite its captivating aesthetics, we were not comfortable relying on the randomness of generative art. We are used to leveraging data at the service of a better understanding of information; generative art seemed comparatively pretty poor on the informational level.

Our job is to tell stories with data. What’s more natural for us than trying to tell our own identity from data?

Beyond generative identity, we were looking for a dynamic identity, linked to real data.

AN IDENTITY DRIVEN BY DATA

Although many of our team members are “quantified self”-savvy, we quickly abandoned the idea of ​​monitoring personal data. We wanted to shift the identity focus on how we work as a team.

This is one of the most characteristic features of Dataveyes. Our team is composed of very different profiles: developers, designers, statisticians, information architects, user experience specialists or data storytelling. Throughout our client projects, we combine our different skills to master the complexity of data and invent relevant and compelling user experiences.

We reached the conclusion that the best way to visualize Dataveyes would be to highlight the density of interactions within the team.

Good data to describe interactions was available, right before our eyes. Since April 2013, each team member has been logging their daily tasks (client name, nature of task, duration, resources etc) on Toggl, a time tracking tool.

From that point, we had to explore these datasets and make them talk. Time-tracking data would then shed light on the richness of our collaboration.

AN INDICATOR OF OUR ORGANIZATION

Through this prism of visualization, data describing our daily work takes a more symbolic dimension. It becomes a better way to reconcile the reality of our organization and its potential of representation.

Our time-tracking tool allows us to export our data in csv format. We conducted a preliminary exploration to update links between team members, as they appear in the tasks inside the tool.

A first exploration workshop allowed us to compare our interpretations of the data by designing early sketches.

The circular shape seemed naturally obvious: it is the visual translation of the unity and cohesion of the group.

The circle served as our framework for organizing our data. Through multiple variations, it gave birth to a strong identity of a company revolving around the interactions of its team.

We defined a methodology to visualize the concept of cooperation from our data.

1. Tasks: over the course of a week, each team member would carry out a number of tasks.
2. Projects: each task corresponds to a project, which is then highlighted in a specific color.
3. Interactions: We then connect the tasks performed by different team members for a common project. Interactions within the team indicate the fact of working together or at the same time on the same project: design group, pair programming, design duo, discussions, etc. Each project takes on a unique shape; it becomes a key element which gathers the relevant skills and elicits collaboration.
4. Patterns: repeating the previous steps for each project helps build a pattern. All projects show complex shapes that intersect, split and merge as if to reveal the activity in a neural network.

The complexity of the forms obtained helped us identify the limits of classic visual design tools. Illustrator (or any other vector drawing software) proved too cumbersome to handle such dense patterns (approximately 500 dots for tasks spanning 2 weeks, and 250,000 possible connections between them)

Our test dataset covering a week’s worth of collaboration also had limitations. We needed to address the data more broadly, to discover the good and bad surprises that our dynamic logo had in store.

We needed a dynamic prototype to keep moving forward.

AT THE CROSSROADS OF DESIGN AND DEVELOPMENT

Like with many of our projects, close collaboration between interaction designers and creative developers enabled us to build a prototype suited to give life to the data.

Through successive iterations between design and development, we created a customized tool that freed us from the constraints of traditional tools and nourished our creative process.

The major advantage of this prototype is its method of rendering SVG and Canvas:

• The SVG format is ideal for handling visualization both technically (through code) and graphically (using a vector drawing tool): a true common language between designer and developer.
• Canvas allows us to visualize more data with fewer limitations on the display performance.

We managed to tame our data to give it shape and meaning. But we feared that our emerging logo would begin to look too cold and technical.

A HUMAN LOGO

We returned to working on the graphics. During this last phase, we were particularly sensitive to metaphors evoking humans. We believe our mission is to inject a human dimension into the data. This is why we refer to the concept of Human-Data interfaces.

Mastering complexity to humanize it: this is our paramount goal, for this logo as well as for each of our projects.

In order to face the “data deluge”, we looked for a familiar and reassuring visual like the iris of an eye. A symbol that resists the variations in data, and is recognizable over time.

This design introduces a subtle dialectic between the whole team and its components. It encompasses every team member, every task, every project in the same symbolic universe.

A PATTERN-LOGO

The prototype has continued to evolve by integrating the latest graphical iterations. Dynamically connected to the Toggl data, it allows us to generate and refresh a logo for Dataveyes at any time.

We created a logo that is flexible and changing. The overall pattern remains the same, but the fine details are different each time.

This is a pattern-logo: it remains coherent over time and changes with our activities.

It does not command a unique vision, but instead articulates a multitude of daily fragments to form a coherent representation. In our quickly changing environment, this elasticity creates strength, like the Reed in La Fontaine’s fable.

Now it is up to you to explore this visualization on your own

A LIVING LOGO

To make this logo more expressive, we animated it several times over.

The pattern-logo becomes organic, it evokes how nature thrives on apparent complexity to structure life.

By inhaling and exhaling, the visualization breathes in unison with the collaboration of the team.

On our website, we visually translate the new definition of our business. The construction of the logo becomes a relevant metaphor of the link between Humans and Data, by turning abstract particles into a symbolic eye iris.

CONCLUSION: FROM THE LOGO TO OUR GLOBAL IDENTITY

We managed to visualize Dataveyes through our own methodology, our vision of data and their potential.

The logo is a tribute to the shape and unity of Dataveyes, by materializing the multitude of interactions that make up the team.

The logo and its graphics system now apply to every communication material produced by Dataveyes, such as business cards, client deliverables, or large posters.

Want to know more? Get yourselves ready for our next articles by subscribing here on Medium.

You can also follow us on Twitter or discover our last use cases on our website.

A new identity based on the interactions between team members was originally published in Dataveyes Stories on Medium, where people are continuing the conversation by highlighting and responding to this story.

At the beginning of summer 2013, the Paris Science Museum, also known as “La Cité des Sciences” came to ask us to create an application that would complement its next big exhibition.

VISUALISING NEWS IN REAL TIME

The proto-project is called “Actualités”. We executed this project with Antidot, a French company specialized in rich data treatments and semantic analysis.

The project was about visualizing, in real time, current French news about the videogame scene. Upon initially hearing about their brief, we almost felt like replying: “how about you re-use the NewsMap concept by Marumushi?”

After all, we’ve always believed this visualization really worked well, because it reached two of the main goals of dataviz:

• compute and manage large amounts of data/ defeating infobesity
• reveal hidden patterns and insights

With that in mind, we could have questioned the point of reinventing the wheel and coming up with a new type of data visualization.

Had we done that, our client could have rightfully replied that the Newsmap doesn’t include Facebook nor Twitter.

Come to think of it, the newsmap was initially published in 2004 — ten years ago! The informational landscape changed dramatically since then, and requires new forms of visual representation.
News are abundant, and travel faster than ever. Current news are the result of multiple sources with different timescales, which makes it very difficult to sum up and represent in a two dimensional space. Besides, news can switch directions at any second through the effect of buzz and the great energy of social networks, which literally shatters the neat little boxes of the Newsmap.
Being able to faithfully replicate this lively dimension was a good first reason for us to try and find new ways to address what already sounded like an exciting project.

Besides, another good reason to jump into this project was about its context. We were tasked with integrating this visualization within an exhibition entirely dedicated to the theme of videogames. This would mean dozens of other interactive interfaces across a room, calling on every visitor’s attention at every moment, asking them to come play.
If we wanted our visualization to be visible and appreciated in the midst of this noise of numerous interactions… we had to create something quiet and relaxing.

1. Therefore we took on the challenge to create an application with the things we liked from the Newsmap:

Reflect on our world by adding meaning.

Help manage information overload and focus our attention to what matters.

2. While taking three objectives into consideration :

Make it quiet.

Make it live.

Set it free.

MAKE IT QUIET

One of the things we really liked about this “Actualités” project was the opportunity to create something without “voluntary interactions”. That was a really radical and inspiring change from our usual projects.
The application would be projected on a big screen, people would pass infront of it but would not be able to interact with it. This reminded us of the whole concept around “ambiant computing”: untangible but everywhere present. That really got us interested. We had in mind this barely futurisitc idea about a world where computing systems would seem as natural and free as the air we breathe.

This is what we wanted for this project. We wanted to design an interface which would be:

• discrete and gracefully blends into the environment of the exhibition
• elegant and polite, in other words that does not jump into your face as you come in
• that does not demand to be at the center of your attention span, and is effortless to access, for instance just by walking in front of it.

That would make a quiet interface.

• In our context of data deluge, a quiet interface is not only an interface that is aesthetic and moves slowly.
  Creating a quiet application is creating an application that helps us manage, and defeat the information overload and does not overwhelm our ability to focus. This is exactly what allowed us to get into the project. We conceived it as an interface with several levels of engagement, which can either call on the centre or the periphery of our attention.

FIRST — “AESTHETICS” LEVEL

This is the level that requires no concentration whatsoever. The application was thought out to be aesthetically pleasing and appeasing. The interface helps cancel the surrounding noises. By doing so, it plants a few bits of high level information, that one would digest without even trying.

SECOND — “DESCRIPTIVE” LEVEL

This level only requires from users to pay attention to the application, and understand what each element symbolizes and describes. In other words, by entering that level of comprehension one realizes that:

• each planet represents a topic.
• a planet is made of a hard core whose size depends on all the related articles published over the past week.
• there is a halo around each planet, symbolizing the echo that those related articles have had on social networks, based on the number of shares or discussed on FB and Twitter.
• also, by that point users will have realized that for each new article, post or tweet, a small particle will appear at the top of the screen and will slowly come orbit around its home planet. It will orbit round and round and will end up being absorbed after a few hours, contributing to making the planet bigger.

By reaching this level of understanding, users will be able to figure out what the main topics are on the videogames news sphere, and which are the key ones. This effectively replaces a press review, in a much more engaging and intuitive way.

THIRD — “ANALYTICAL” LEVEL

Reaching this level means that your curiosity has been tickled and made you stay for a few minutes in front of the application to fully grasp:

• how each planet is formed and what is the life of a cluster
• what makes it that some topics go viral on social networks and others generate an impressive quantity of news articles
• why some topics have a lifespan exceeding weeks, while others seem to be shorter phenomenons

This is by far the most interesting level. This application becomes a genuine media source, with stories to share on a daily basis. The interface feeds from the video game universe to explain it with all the subtle complexity it deserves. After all, the topic itself is quite fascinating: video games do have a social dimension, by raising issues that often make it to the general debate.

To go further, we have decided to create an interface that is “alive”, not only in the sense of real-time, but also in the sense of giving the application some autonomy, by inventing a system that would track data over time, before even knowing how data would evolve.

MAKE IT LIVE

There are of course many different ways of making information easier to understand and telling stories through data
The most commonly used is space-based information: that consists in visually displaying quantitative values in a very intuitive way, sparing the brain some significant amount of effort.
The size of a planet’s core is proportionate to the number of articles related to a topic. The size of the surrounding halo also depends on the number of mentions on social networks.

Animations and interactions also allow to display the information progressively, and break down complexity into several steps, separate scenarios and users paths that make understanding the data easier.

• real time insight is provided through animations and meters
• small particles are labels when they land, and the label then vanishes in order to not overcrowd the screen
• similarly, the details for each cluster appear one by one

We’d also like to present a dimension we really care about at Dataveyes : using visual metaphors that transform intangible data into seizable matter, something you can touch, manipulate, understand.
This is very useful because it gives users the impression that the data actually exists and means something, that the datasets is not some sort of impenetrable black box or undecipherable algorithms.Data looks tamed, less intimidating and easier to approach. This way, data is given a shape, it puts users in the driving seat as they realize they are able to do something with the data.
Taking a step back, this is something we strived to accomplish on previous projects:

1. In this project for the city of Rennes, we created matter through small bubbles that users can distribute across the screen as if they were playing with marbles.
2. In our Metropolitain project, we used data to show a 3D map, piling up the data in mountains of insight that you wouldn’t see from ground level.

So how does one exactly bring matter to data?

HOW WE TRANSFORMED UNSTRUCTURED DATA FEEDS INTO USABLE ASSETS

When we began working on this project, we started by putting together the mechanism that would allow us to automatically retrieve the relevant data. That was done through a partnership with another Paris-based startup called Antidot, which specialized in semantic data processing.

Our primary material was feeds of articles from about 30 sources:

• RSS feed from specialist blogs
• general information sites
• Google news, Yahoo news, etc.

This allowed us to gather a solid pool of articles, updated in real time, which we would use as the core ingredients for this exciting recipe. The processing phase was lead by our partner Antidot, who used technological components of their own to assess lexical proximity between articles and identify relevant topic clusters.
This allowed us to address the biggest challenge of this project: transform bits of scattered news into a topic, without any human intervention. That is a similar technology to Google News actually: starting from a pile of raw, unstructured data, and give them a shape, refine this raw material to extract the most relevant value afterwards.
To this effect, data is enriched and linked through statisticial and semantic technologies. Content feeds go through a series of filters than progressively gives them a new dimension and makes it usable. The path is essentially: crawl > cleaning > extraction > classification > clustering > indexing.

The exhibition was planned to last 8 months. By the middle of this period, this hungry algorithm had ingested over 20,000 articles, and identified more that 1,000 different topics over the weeks.
After the clustering phase, we gathered the data from the very clean API provided by Antidot. And it’s on the back of those refined, structured datasets that we built the visualization interface.

HOW WE MANAGED TO WEAVE SUBSTANCE FROM THE DATASETS

Designers and developers from our team worked hand in hand in a very iterative fashion along the course of the project. They constantly checked and worked from development and design tools : developers built shapes in WebGL, which designers refined to bring out relevance, precision and aesthetics within Illustrator. Developers then replicated those variations in the prototype, added some new parameters, and gave it back to designers and so on…until we reached a more polished deliverable.

The first version of our prototype was very rough. A grid-shaped cage, with animated little lights that travel from the back of the screen, glittering all over. We were quite anxious to receive thousands of articles at the same time, we firstly wanted to make sure we had performance issues under control before switching to more elaborate visual forms.

At this point in the project, we tested a great number of variables, with very different visual outcomes. We were playing around, trying to find the best way to embody the materiality of that data.

In the end it was pretty close to the process of designing actual video games, where developers test animations on simplified skeletons before adding the flesh, modelling matter, fabric and environment.
Finally we went for a fairly “low poly” approach. While arguably a trendy approach, we leaned towards it for its ability to process matter, for how it reverberates light on different sides of a shape. Through this approach, shapes are simplified. With a great emphasis on smooth surfaces, the spatial presence of an object matters more than its texture or anything else.

Fun fact, we also picked the “low poly” approach as a tribute to the world of video games. Back in the early days of gaming, where technologies were not what there are today, forms and shapes were simplified, not for aesthetics, but for performance reason: to make games lighter and easier to animate.

Creating matter is one thing…
bringing life to it is another

BRINGING LIFE TO THE MATTER

From the beginning of the project on, we had very organic metaphors in mind : seaweeds, trees, flowers, stem, helium balloons, planets, northern lights, etc. This is why we created animations, and sought to bring movement inside the application from the beginning of the prototyping work. At the end, we conceived the animated mode as the default state of the application. This is a different approach from most of our other projects, where the animation comes in later on to reinforce the purpose of an interaction. With this project, the animation is the purpose of the application.

SET IT FREE

To make this happen, we chose to work using 3D, to best highlight those very organic metaphors that we picked. As a consequence, this introduced really strong links between elements of the interface. Elements attract or repel each other according to physics law that replicate reality, like gravity.
However, it was impossible to predict all the potential configurations of this living matter: we did not know how the data would behave over time.

This could create some problems:

• What would happen if a massive topic came up and a giant planet took up the entirety of the screen?
• What would happen if topic categorization was not finely tuned, resulting in one planet attracting all the others?

That is why we had to imagine visualization mechanims that were autonomous. For instance, we implemented non linear scales and behavioral rules to avoid that a brutal or abnormal event completely overwhelms this universe. In a nutshell, we put together a series of saving and self-preservation mechanisms.We tried to anticipate all the situations that could come and disturb our ecosystem, so that the interface would not be at risk of letting us down.

For instance :

• The application cannot freeze. If the interface ceases to receive data for any reason, planets will continue their course quietly, and they will maintain their shape and size until the connection to data resumes
• When connection does resume, the catch-up effect is avoided: the interface doesn’t suddenly become crowded with a flood of articles published during the cut-off. The catchup phase is spaced out so it looks natural.
• In case of a major server issue, a local version of the interface is on standy by and ready to take over to play a sequence of simulated data. The transition back to the real-time interface, once server issues are addressed, is undetectable to a user’s eye.

With the help of all those self-preservation mechanisms, the interface has become an autonomous, lively organism, that ingests data and recycles it to feed an entire, dynamic ecosystem. The ecosystem is able to reflect at any point in time the state of news in the video game space, without any human intervention.
This is how this system reflects the reality, by gathering descriptive signals and representing them with an added value, with a layer of poetry and aesthetics that does justice to the beauty of data.
All in all, this project is also a tribute to the mysterious energy of data, the energy to transcend reality, connect insight to users and power planets. Datavisualization is an amazing medium to translate the cold data of our information landscape into a poetic, yet meaningful universe.
This is where the greatness of data storytelling lies: transform raw data into tangible matter, to build a quiet, playful, and priceless experience… that, at the end, escapes human control.

WATCH THE VIDEO

Want to know more? Get yourselves ready for our next articles by subscribing here on Medium.

You can also follow us on Twitter or discover our last use cases on our website.

A living landscape to translate current news related to video games was originally published in Dataveyes Stories on Medium, where people are continuing the conversation by highlighting and responding to this story.

We wanted to liven up the traditional maps of the Parisian underground system, in a way that better matches the reality of commuting.

PARIS NEVER SLEEPS

Millions of commuters travel through the city — one of the world’s top tourist destinations — on a daily basis. Therefore it is no wonder that the city’s backbone consists of its metro system, known for its density and round the clock affluence.

As a result, the official metro map conditions the very way commuters approach time, and space, as they tend to select their journeys based on the perceived smallest distance between two points.

A team of loyal Metropolitain travellers, the Dataveyes members were keen to visually represent the underground system and its dynamics in an alternative way, which would better show the true reality of a commute, through data.

This is why we created metropolitain.io

Metropolitain.io is led by Dataveyes’ very own R&D lab. The project attempts to visualize and make sense out of Paris urban data. The interface enables users to interrogate datasets around journey time between metro stations, at any hour of the day, as well as the number of people touching in at each station.

The data around crowd turnouts was made openly available by the RATP (Autonomous Operator of Parisian Transports). Other datasets about time between stations were provided by the start-up Isokron (who realized the Locomote App).

A series of distinct elements got us to work on this project:

WHY WE DID IT

Our passion for smart cities

We are fascinated by anything that revolves around the ‘connected city’. In other words, how data can help shape and challenge the visions that we have of the areas we live in.
Visualization precisely goes beyond the abstraction of data. It helps uncover actual meaning on our activities, our habits, our environments etc. Data visualization can connect the dots, make you comprehend information that sometimes you would not see from ground level.
We have the unique opportunity to create innovative interfaces that transform the data that matters to our city lives into interfaces. Those can help us understand our metro network, identify hot and cold spots and ultimately save time during our daily commutes.

In a nutshell, unleash the power of abstract data to serve the public interest.

Our passion in User Experience

Besides we are keen to provide our audience with a genuine, innovative user experience. Since Dataveyes was created in 2010 we have done a lot of thinking around data in general and how it can be best represented.
This visualization offers to challenge the way we traditionally view our 2D metro maps.

Métropolitain takes on an unexpected gamble: using cold, abstract figures to take the pulse of a hectic and feverish metropolis.

The metro map is no longer arbitrarily dictated by the spatial distance between two points. By playing around with two extra variables — time and crowds — users can transform the map, view it in 3D and unveil the true reality behind their daily commute.

Our eagerness to challenge representations

This project was carried out by our R&D lab, which gave us the ability to “play around” with a level of flexibility we would not necessarily have when working on a client brief. Looking back, we seized that opportunity, by trying out representation modes we were new to, such as map distortions or 3D.
The challenge for us was clearly to make sense out of those new representation modes, by reflecting on how they would help the user understand the information better. We see Metropolitain.io as an experiment, an opportunity to learn new approaches along the way, see whether people engage with it, show interest or want to emulate/ improve our existing project.

The first version of Metropolitain.io raised a number of questions:

• Are those new representation modes efficients?
• How will the general public approach this 3D interface?
• What will be the most used functionalities?

We created a testing plan which allowed users to freely discover the application so we could assess how usable the interface was. Here are the results:

1. The interface generated curiosity among participants.
2. Users immediately seek to use the interface as a tool to answer practical issues such as finding out the time it takes to get from one specific station to another.
3. After a quick learning period, participants understood the purpose of the application and expressed a need for more detailed analysis functionalities.
4. Many users would have liked to merge the Crowd View and the Time Map, in order to see how crowd levels evolved during the day.
5. 3D often generated positive reactions. However, most users kept the 2D mode in order to easily navigate on the map.

OUR APPROACH

Prototype First

Contrary to the way we usually approach datasets, we started by prototyping straight away, without any thorough wireframing, and even before we knew what the datasets exactly contained.
We directly developed the interface in WebGL as we felt it was the most suited for our purposes. Using WebGL (through a Three.js library) was an exciting task. While we had used WebGL for a few internal projects in the past, this was the first time we built an interface out of it for the wider public to use. We feel that this technology really fulfils the potential of graphical maps to dynamically calculate complex 3D objects. Overall, WebGL enabled us to break free from the performance issues.
We felt we had to prototype first, in order to determine how the datasets would carry information across and challenge our representations. The prototype really gave us the key to understanding the data and what we could do with it.

From data-centered focus to user-centered focus

From that point onwards, the challenge was no longer for us to understand the data but to help others see it too. The design phase enabled us to make that data come to life, make it usable, in a way that is both exciting and detailed. This translates into a variety of interaction possibilities, buttons, info-boxes etc. We tried to find the right balance between aesthetics and relevance, to design a visualization that would serve our user’s purposes.
It was the shift from a data-centred focus to a user-centred focus. In some of our client projects we would rather proceed the other way around, depending on the circumstances. We aim to always operate in a seamless, flexible fashion. This is due to the mix of skills within Dataveyes and our ability to work closely together to get the most out of data.

Interpenetration of conception and production

The interface naturally went through a series of modifications as the project progressed, as we tried to optimize the way we would represent data. The design phase came much further down the road, almost at the end, in order to make the interface more user-friendly and engaging.

This kind of all-inclusive project got our team to work in a very integrated fashion, beyond the traditional boundaries of coding, data mining and design. Looking back, conception and production were very much intertwined.

WATCH THE VIDEO

Open the application by visiting http://metropolitain.io/.

Want to know more? Get yourselves ready for our next articles by subscribing here on Medium.

You can also follow us on Twitter or discover our last use cases on our website.

A new representation of our transport network shaped by metro data was originally published in Dataveyes Stories on Medium, where people are continuing the conversation by highlighting and responding to this story.

In 2012, we dedicated over 6 months to the conception and execution of a tool visualizing human resources data for a leading French corporation. The tool greatly facilitated the elaboration, implementation and monitoring of talent management strategies within the company.

THE TRUE REASON BEHIND OUR FIRST MEETING

Back in December 2011, we met for the first time two of the main HR managers of the French corporation. Their mission consists in attracting and nurturing talent within the group, on an international scale. This led them to repeatedly question the way they would approach data.

1. What type of information can one rely on to initiate a relevant strategy?

A robust plan of action, consistent with the reality of HR tasks, is required to effectively implement any strategy within a global group

In order to build such robust plan, one must gain a sharp, reliable view of all HR situations and scenarios met across hierarchical structures

When your group has hundreds of thousands of salaries, you can’t obtain such a reliable view counting on your human knowledge only. To master the granularity of information to such degree, you also need to take knowledge from data.

2. How to communicate and assess the implementation of this strategy?

Data knowledge and elaborating strategy is only one step of the way. Implementation issues must also be tackled and assessed so that each branch of the group can expect to benefit from it in a measurable way.

This lead to a second challenge: how to clearly communicate about an operational strategy and measure its success?

Of course, managing data is one part of the solution, and the HR stakeholders had already made significant use of the data to start answering those two key questions.

In this context, we were approached to precisely go faster, and go further.

Great ideas cannot settle for average, limited tools that would hinder their natural progression.

HR managers within the group longed for a tool serving their vision, that would empower them to clearly formulate their strategy and communicate about it; in a word, a tool to help them work better.

As a result, we created a bespoke tool to visualize the company’s data.

Along this mission, we quickly realized that focusing on exploring the data was not enough; we had to design a whole process clarifying how to work with the data.

REVEALING MORE IN-DEPTH NEEDS

Our contacts have access to three distinct sets of internal data that describe the employee populations and their mobility across the group, for each step level of its hierarchical structure.

HR managers within the company had for some time been using Excel to analyze the data, and Powerpoint to present the compiled insight.

At the time of our first meeting, our contacts were particularly concerned with how data insight would be presented and shared:

“What is the point of elaborating a sophisticated strategy if, upon sending graphs presenting the analysis results, our stakeholders do not fully grasp the insight behind the data and fail to take the right decision?”

That questioning marked the beginning of our mission.

We convinced our contacts to work further with us, by sharing examples of how we could visually translate concepts such as branch maturity or the evolution of good recruiting grounds.

Later on, we then understood how their needs far exceeded the framework of information representation…

Following the method with which we approach each of our missions, we started with the immersion phase. We carried out a series of interviews with our contacts. They shared their views and insight on their jobs, vision and how they were used to work with their data and software.

With their help we managed to give birth to more of their needs:

• Need to save time: reduce the high number of days spent manipulating data in Excel and Powerpoint.
• Need to simplify the monitoring and presentation tasks: to facilitate data interpretation.
• Need to unify the information being presented: reach the same depth level of relevance for all indicators, country or hierarchical level.
• Need to identify long term trends: establish comparisons, analyze results in an interface that doesn’t require users to do complex calculations.
• Need to provide a genuine steering tool: beyond the presentation features.
• Need to boost a network of operational managers: through compelling presentations that translate a vision and tell stories that guide decision making.
• Need to inspire good practices around data use: in order to promote a culture of internal reporting and help implement new objective-focused management techniques.

This initial immersion phase led up to a true summary deliverable, listing the identified objectives the tool would be enable the company to reach, and how we intend to proceed.

EXTRACT INFORMATION FROM DATA

We then started working with the data. This is always a sensitive stage in our job. Datasets are in many ways similar to the goldmines of 19th century America: they can prove to be extremely rich but it requires to dig, extract them from existing silos, maneuver a bit and be patient, in order to extract the necessary raw material.

In our case, we also needed to handle two main issues:

Data was very heterogeneous:

• The data comes from three distinct sets from different work tools. Those three tools are not implemented by the same people, nor are they updated at the same time. The hierarchy of the group actually varies from a tool to another. Some of those tools are in English, other in French, which further complicates any attempt at creating a consistent classification by country. Some tools display month by month cumulative figures, while others show absolute data, etc.
• Despite those issues, we decided to build indicators at the crossroads of all those data points, to be as relevant as possible to our stakeholders’ needs.

The chronological dimension of the tool was a brain teaser:

• How to display the history of data for a specific Business Unit which changed its names several times during the last few years?
• Similarly, what happens from a data standpoint when a business unit is absorbed by another; which historical data should then be displayed?

Those questions greatly helped lay down the framework of our mission, revolving around the following steps:

1. Defining new indicators based on our understanding of the data and the client’s profesionnal environment

We always approach new datasets from the angle of the what insight they contain and how they can be used in a professional environment.

Spending a long time listening to our contacts allowed us to describe which type of information was needed to support HR decisions across the company.

Consequently, a further analysis of the content enabled us to pinpoint which indicators should be highlighted at the crossroads of datasets, and for which use.

2. Guarantee consistency and temporal continuity through the most appropriate architecture

We have described above the issues relating to data architecture and data quality as of when the mission started.

Acting on this observation, we conceived a new data architecture allowing to unify the three core datasets, in order to ensure chronological continuity.

3. Automate data treatment and storage

We developed a script that receives the source data, processes and extracts a clean dataset which is specifically tailored to provide the insight required by the HR function.

4. Improve data quality thanks to curation standards in line with the client’s mission

Along with our client we also defined quality standards for the data and well as curation rules:

— What are the data ranges for each indicator?

— How should the tool behave when confronted to values that exceed the indicated ranges?

For example, what to do when an employee is listed as being 134 years old in the source database?

By establishing data quality standards, the tool can identify that the value is abnormal.

— Should the tool then delete all information related to this employee, to avoid the risk of skewing the relevance of other potential indicators?

— Or should the tool simply ignore the age of this element, and conserve the other information points related to this employee?

As we agreed with the client, this decision essentially comes down to what the mistake tells us about how the data was collected. In this particular instance, it was the knowledge of the HR job, coupled with an analysis of the data quality, that we managed to establish data management rules.

5. Facilitate the correction of data entry mistakes in source files

Our data treatment script was able to detect different mistakes inside the inital values collected by HR collaborators: bad completion of the organisationnal chart, missing values, sub-totals incoherence, etc.

This way, the tool can offer to the administrator to download a error log to help them correct the sources of data.

In parallel to our work with data, we focused on the graphical and interactive embodiment of the key indicators.

EMBODYING THE MEANING OF VALUES

Our client had initially asked us to formulate a first sketch on paper, early in the project — just after taking in the data specificities and the user needs. The purpose was of course not about having a clear view of what the final tool would look like, but rather showing how we could come up with innovative ideas to visually represent data.

We seized this opportunity to test the match between new visual forms and the meaning of the data, by sketching many small data visualization modules.

We then put our sketched modules to the test, by mapping them up against actual data points. We then kept the modules that appeared to be easiest to read and information-rich.

How to, for example, represent the ratio between two manager populations — juniors and seniors? Our understanding of the client’s work allowed us to see that the key insight often resides in the links between two populations: one is often intended to replace the other, with an expected loss ratio.

We knew here the ‘target’ value of the ratio, which guarantees that the number of senior management positions will be replaced when the junior population will evolve. We thus imagined a graphical representation that materializes this mutation from one population to another via a ‘projection cone’. If the origin and destination populations join up with the two bases of the cone, then the ration is close to its ideal value. It any of the two populations leaves the cone, the ratio goes off the desired values renewing the population of senior managers could reveal problematic.

This type or representation allows us to visualize the true meaning and implications of the data, with regards to the good practices within the client’s work.

The purpose was then to find out to assemble those modules so they generate meaning as a whole inside the interface.

SETTING OUT INFORMATION ARCHITECTURE

We then initiated a phase dedicated to information architecture.

Our objective was here to optimize the interface to deliver the most relevant insight from the data, in order for this information to be pleasant to browse and find. The goal is to make the returning user comfortable.

The first stages of conception felt like a 1000-piece puzzle.

We started working with dozens of different indicators, all of them with their own logic and themes, and offering very different ways of filtering.

Regardless, we worked towards grouping the data and identify a coherent interface structure, from a thousand possibilities.

A few days and post-it papers later, we had derived a solid architecture, where each indicator was in place and in synergy with the others. The architecture was made of several coherent tables, arranged along a natural progression.

We sketched this first architecture using zoning and shared it with our client.

This allowed us to avoid an interaction issue: sometimes a user may click on a button and doesn’t see what it changed on screen because the change operated quite far away from the button they clicked.

We obtained their approval which, in addition to our work on the data, gave us some solid ground to go further.

BRINGING THE APPLICATION TO LIFE

At this project stage, we had brought together an analysis of data and bespoke information architecture.

For the rest, we had to focus on the details to make it work.

We moved from the initial sketches on to wireframes.

We added more matter to the first skeleton, determined how the navigation should unfold for each action: filtering, zooming on the hierarchical levels, picking a country, change time period etc.

All the options given to a user must be displayed in a simple, transparent way within the interface: how does the interface behave/ change at each move of the user’s mouse?

We programmed this interface so that it may address the needs users have, in the most fluid and natural way possible. We attempted to anticipate their issues, logical thinking and their preferred points of reference, to build a navigation path they would be comfortable with.

From updating the datasets to impressing a graph, including interpreting the data and navigating through the company’s hierarchical levels, we have tried to anticipate everything. Being a step ahead of the users would help us give them the information they need, before they even look for it.

REVEALING THE INTERFACE

Once those wireframes were validated, we were able to add a layer of visual design.

In the meantime, we started building the technical architecture. It was intended to support the entire process from data collection to processing, including complex situations that involved chart restructuring or error spotting.

This backend then delivered refined data to the frontend via APIs. This is where the Dataveyes magic operated: our wizard developers structured the application interface and brought the data movement to images.

In less poetic terms, that means writing lines of Javascript code. A lot of them.

Those tasks kept us busy for about 4 months and accounted for over half the time spent on this mission (6 months in total).

In the end, we had a finished, tangible product to show our client.

Very quickly, the stakeholders on our client’s side proved to be very comfortable with it: they understood the information and performed interactions seamlessly. They were eager to dive deeper and start telling stories from the data.

A FEW MONTHS LATER

1. The first data updates enabled us to carry out final adjustments
2. Our stakeholders now feel comfortable enough to run some of their internal strategy meetings using the new tool rather than the old Excel/ Powerpoint combination
3. The first version of the tool received overwhelmingly good feedback, beyond the circle of the stakeholders we were directly in contact with.

Looking back, our contacts initially requested us to create a tool that would draw their colleagues’ attention to the importance of data. Six months down the line, their colleagues are now proactively asking to access the tool.

As a result, the second phase of the project will now revolve around deploying the tool across a larger number of users, in order to have it benefit more managers around the group.

CONCLUSION

The mission lasted six months, during which we carried out 15 meetings and catch ups with the client, and just about 100 workshops and scrums within our internal team to progress.

In addition to the first version of the monitoring tool, we also provided the client with 12 deliverables, describing our methodology step by step, as well as recommendations on how to maintain the tool.

Want to know more? Get yourselves ready for our next articles by subscribing here on Medium.

You can also follow us on Twitter or discover our last use cases on our website.

A tool to reconcile HR with the use of data was originally published in Dataveyes Stories on Medium, where people are continuing the conversation by highlighting and responding to this story.

(A French version of this article is available here)

Data visualization can serve GDPR compliance: to improve interfaces for the general public, to map the digital assets of businesses, and even to support the audit of algorithms. That’s why data visualization must look to improve interactions between humans and data. Here’s our feedback on uses and best practices.

The General Data Protection Regulation (GDPR), which came into force in May 2018, amended the rules for the management of personal data in European Union countries. It aims to bring more transparency to citizens, strengthening their rights over the use of their personal data, and making the various players processing their data more accountable.

The GDPR not only introduces technical challenges, related to anonymization or data storage, but also design challenges. The design of an interface can influence users’ ability to consent to data collection: it can either help them make an informed choice, or, on the contrary, divert them.

Data visualization is an integral part of interface design. Through visuals and interactions, the translated data becomes easier to grasp and understand. As such, it has a role to play in building relationships that are more respectful of citizens’ rights and liberties.

The context is favorable. Six months after the GDPR was implemented, the CNIL (National Commission on Informatics and Liberty) carried out a survey in partnership with Ifop (French Institute of Public Opinion). The result: out of 1,003 respondents, 46% had already noticed the misuse of their personal data, and 66% said they were more sensitive than before to the protection of their personal data.

1. Leveraging data visualization to ensure GDPR compliance

1.1 Improving the interface of services collecting personal data

The GDPR entrusts actors involved in data collection: it’s up to them to design collection methods that don’t fool users, and clearly inform them of potential implications. The regulation states that an application or a website that collects personal data must ensure the users are in a position to give, or not, their informed consent. For this to happen, users need to understand how their data is used and where it comes from.

Using data visualization in the design process of an application or a website can help achieve this goal. Visual storytelling and interactions are key to helping users understand the information that lies in their data. They become able to grasp more complex concepts. The data reveals itself and becomes easier to apprehend.

We have worked on a prototype application connected to the data of a smart meter, to help the general public better understand how to monitor their electricity consumption. We started with comments and feedback from the inhabitants of newly built homes, equipped with electric sensors and smart regulators. They had feelings of intricacy, fear of disruption, doubts about their capabilities, difficulties to read the data, etc. Our prototype solves these problems by leveraging human-data design and interactions: visualizations facilitating the monitoring of electricity consumption over time, a notification-based approach, a gamification system with reward badges, and even a comparison with the consumption of the neighborhood — based on the principles of green nudges. Since smart meter companies often deprive users of useful information, by not giving them access to raw data or informative graphs, we wanted to show how to empower citizens through design.

Here’s another of our projects: an educational simulator to explain the concept of energy flexibility, using data from sensors installed in test homes. Users can visualize data around a clock and interact with the home’s regulatory systems, following different scenarios. This way, we facilitate the user’s understanding of the complex electrical control system, and make it more transparent.

As we can see, GDPR’s deployment is as much an ordinance as it is an opportunity: the one to build better mediating schemes between users and data.

1.2 Mapping the collected data

Data visualization can help raise awareness on the magnitude of the data being collected.

First, for the general public: the striking clarity that can emerge from a data visualization is effective in making users more aware of the data they share. OCR’s project “Behind the banner” for Adobe, Mozilla’s “Lighbeam” or the CNIL’s “CookiesViz” are such examples: they explain the process of targeted advertising through data storytelling, and encourage citizens to take an interest in it.

Secondly, for companies that collect personal data: they are now obligated to have a clear vision of their databases and processes in order to verify their compliance with GDPR, and conduct impact analyses. Visualization tools can help them, such as Datagalaxy or Dawizz. These tools can label and map their data in meaningful ways, with visuals such as treemaps or social graphs. They contribute to better data understanding, a better ability to inventory it, and better governance. To achieve objectives that are hardly solved by relying on lengthy or tedious technical documentation, data visualization introduces a visual language understood by all members of the company.

1.3 Making data processing algorithms auditable

Finally, data visualization can be a valuable tool for auditing data processing algorithms and ensure their GDPR compliance. Especially when these algorithms involve machine learning: their inner workings is often opaque, and even referenced as black boxes by data scientists themselves — especially for neural networks. They can introduce biases and sometimes lead to discrimination, if left unaudited. For example, yield management algorithms that are used in dynamic pricing systems use many parameters, including personal data such as your postal code, in order to optimize sales of products or services.

In this case, data visualization can be used to make these algorithms and their results more tangible, explain how they transform data, and help identify biases.

We’ve made such an example of a visual tool, called “Affinity Index”, created for the sponsored content recommendation specialist Outbrain. The aim of this project was to clarify the behavior of algorithms that were perceived as opaque: the ones that allocate sponsored articles in an inventory of available spaces, taking into account the reader’s affinities. We have visualized the results collected over a year, and, using colored bubbles, we made the recommendation dynamics visible.

2. Human-data interaction design and GDPR best practices

2.1 Adopting an iterative, user-centered design approach

Beyond these different uses of data visualization, practices from human-data interaction design can contribute to a greater alignment with GDPR’s ambitions.

Adopting a user-centered design approach, such as design thinking, helps create GDPR-compliant projects by considering end users from the start. And that’s a practice that goes hand in hand with the concept of privacy by design, present in the GDPR (data protection by design and by default). Thus, we recommend an iterative approach, guided by small-scale tests, prototypes, and user feedback. It makes it possible to verify, for example, that presenting certain data sets together doesn’t raise confidentiality problems, or that an offered service is comprehensible and valuable. Putting the product in the users’ hands right from the start is a motivation for the designers to beget confidence and support.

Working with data from the beginning of projects and prototyping visualizations early are also important, enabling a clear understanding of the data’s behavior. By testing several visualization options, it’s possible to identify the ones that cause the least bias and highlight the most valuable information for users.

Finally, we recommend a modular and iterative approach to projects, so that applications and services remain easier to maintain, as the user’s consent is likely to evolve over time.

Using these methods, we have designed a tool for the HR sector. Using data from job posts and internal role changes, we created an application that summarizes individual career histories and structures it into paths followed by employees, from one job to another within their company. Where the HR reporting tools are often simple dashboards with general statistical indicators, we have made a career guidance tool for employees. The application shows the career paths in the company, without any institutional filter or company bias, thus providing more transparency.

2.2. Making data collection explicable, progressive, and optional

It all comes down to the moment the users are asked to give consent and allow the collection of their personal data. In practice, they trade personal data for a service. From the moment the product is designed, they must be able to evaluate whether this trade is a win-win… or not.

When users are asked to provide or allow access to information, such as geolocation, a good practice is to explain what this information will be used for, and especially what won’t be available if they don’t consent. For example, without geolocation, a calendar will not send a reminder when it’s time to leave for an appointment. Without a viewing history, there will not be relevant suggestions for upcoming videos. With such explanations, the user is better equipped to assess the impact of a refusal to data collection.

Allowing users to progressively share their data is another good practice, asking for it only when it’s essential to their experience and not before. All too often, the sharing of personal data is a prerequisite for the use of services: it’s an all-or-nothing approach. The gradual sharing practice is consistent with the GDPR’s requirements to only collect useful data, ideally matching the exact purpose of its subsequent processing.

This also implies that applications can run in degraded modes, with features that don’t require access to personal data. For example, it’s entirely possible to use a mapping application without continuously sharing location: the “directions” functionality may be disabled, but this doesn’t prevent the search of a place. A privacy-friendly practice is therefore to design applications with a minimum set of features that don’t require any personal data, and advanced features taking advantage of user data.

When we undertook the “Crea Carte” project for Société Générale in 2015, we had these good practices in mind. Crea Carte was a competition on the theme of generative art. Participants were invited to create an aesthetic pattern from data. The users could submit their designs to the jury and try to win a prize, or simply add the design to the collaborative gallery, as a contribution to the collective artistic experience. During the process, the participants’ contact details were only asked if they wished to show their creation to the jury. All other features, constituting most of the experience, could be used anonymously. As for the data used as the base of the aesthetic pattern, we made sure they were unintrusive: they were spatial coordinates describing a gesture.

2.3. Improving the understanding of the data’s origin

Designing an application that respects the rights of users and provides value is not always enough for a good human-data mediation. It’s also useful for users to understand how the data is used, so that they can eventually question their consent. Not everyone knows about the technical intricacies of platforms such as Google or Facebook, which can feel like magic and lead to distrust.

In order to overcome this problem, we recommend making the data sources explicit, in all projects, even when the data is not personal nor confidential. This can be done by providing access to a paragraph or a page describing the data collection and processing methodology. Pictograms with tooltips are another option to display the data’s origin without cluttering the interface.

Providing context for the data presented also helps to understand its origin. For example, indicating the number of people interviewed in a survey, the date when their opinion was collected, and the exact questions they answered. Or specifying when the data is processed by an emotion recognition or term identification algorithm, as may be the case with the analysis of social network content.

Ideally, the source data should be accessible in open data, when it doesn’t raise confidentiality issues, in order to contribute to data transparency and offer users more opportunities to criticize its use.

2.4 Relying on user intelligence

We must finally ensure that users are cognisant of the information extracted from their data. We believe we should always consider that users are able to grasp complex information if they’re explained clearly, rather than presume it doesn’t interest them, or is out of their reach.

Thus, we prefer visualization modes that make the data tangible and, for example, assign it physical properties. All too often, the designers’ reflex is to simplify the data, in order not to frighten, by using simplified scores or composite indices that are easier to display. These KPIs (Key Performance Indicators), however, can be black boxes: they don’t have meaningful units, they are often cut off from any analysis context, and they aggregate too many dimensions. This makes their variation hard to interpret — or impossible. Under these conditions, it’s difficult to know which underlying personal data is being used and whether it was worth sharing them.

We recommend expressing the data in meaningful units as often as possible. For example, by displaying a number of people instead of a percentage. Or by showing probabilities assigned to categories rather than imposing a single verdict.

Transparency, purpose, relevance… if all the ingredients are there, it’s possible that users voluntarily entrust their personal data. To serve their individual interests, of course, but also to serve collective projects. The sharing of individual data can go beyond the relationship between consumers and digital service platforms, and become part of a voluntary, citizen initiative. This is already the case with quantified self applications where sharing a car’s journey, sharing individual pollution sensors or noise measures, make it possible to better document cities and to give their citizens new levers for action.

In summary: best practices in human-data interaction design to comply with the GDPR

1. Design projects by iterating, prototyping, with a user-centered approach.
2. Describe the data collection and processing methodologies, and provide access to source data.
3. Make the sharing of personal data progressive, desirable and explicable: it must be mutually beneficial.
4. Visualize, make the data tangible and interpretable.
5. Use meaningful indicators and avoid black-box-type composite indices.

Three projects designed with a thoughtful use of personal data

My neighborhood

While data collection is sometimes necessary at the individual level, it’s not necessarily the case for visualization. For the project “My neighborhood”, carried out for the Nexity real estate group, Dataveyes collected precise locations of a sample of inhabitants’ frequented places, but the application only displayed areas of density or distances. The value for the end user is not in a precise geolocation, but in the accessibility of services in a given area.

Scope Santé

This website gathers information to help the general public choose health institutions. The aim is to deduce which informations are relevant to users from their behavior on the website, without asking them to provide their medical situation on their profile (Facebook connect, a form, cookie retrieval, etc.). For example, the users who performed a hospital search with a “pediatric emergencies” filter will get specific information on pediatrics, without being identified at any time or asked for personal data.

Commute

The input data for these two experimental projects are audio files, recorded when users travel around the city to analyze the noise level they’re exposed to. These projects have the potential to be very intrusive, except that the audio signal is not stored as is but is immediately converted to frequency, volume, tone, and other signal dimensions. This way, it’s impossible to listen to what has been said or to find the places covered during the journeys.

GDPR and Data Visualization: When Design Meets Compliance was originally published in Dataveyes Stories on Medium, where people are continuing the conversation by highlighting and responding to this story.

(A French version of this article is available here)

At the beginning of May in London, we took part in the Strata Data Conference, a well known international event about data. An opportunity for us to stand out in an increasingly complex and congested ecosystem, to deep dive into the current trends, and refine our understanding of the issues many actors face.

Be data-driven, or die

Since its debut in 2011, the Strata Data Conference has been the place to discuss topics at the forefront of the data-driven world. Data-driven, or AI-first as its more fashionable known, is a term encapsulating all the concerns of big companies in recent years. According to the recruitment platform Indeed, the demand for the data scientist profile alone has increased by 344% since 2013, with growth of + 29% each year. Is being data-driven the major difference between innovative new cool kids and older more established companies? Under the fear of letting the innovation train leave them behind, many companies decided to create a data lab (or data foundry, or data factory).

Are data labs failing?

What is a data lab if not humans, who master data science, and data? The word Lab here emphasises the experimental nature of the projects that might be implemented. On paper, the data lab seems to be a good idea: a team within an organisation dedicated for making the most out of data and valuing this essential business asset of a 21st century company. But between structures often far from business operations, and complex topics like quality, accessibility and governance of data, the model seems to struggle to create value.

During Strata, I discovered this study by NewVantage Partners that asked top managers and C-level about the transformation of their companies. The study shows a strong awareness of the importance of mastering data: in 2018, 97.2% of the leaders who responded said that their companies invest in Big Data and AI initiatives, in order to become data-driven companies. However, the 2019 study reflects an impression of failure: the percentage of tops managers identifying their companies as data-driven has decreased over the last 3 years (37.1% in 2017, 32.4% in 2018, 31% this year). This perception dissonates with their investments in data projects: according to the 2019 survey, 92% of respondents are accelerating their investments in Big Data and AI.

I conclude that despite the vital and well known importance of becoming data-driven, just a few believe their company is doing so. How can we explain this impression, despite a growing financial and organisational effort?

Myth #1: data labs are all about data scientists

The first problem companies faced in creating a data lab is the difficulty recruiting candidates with technical skills in a new and extremely volatile environment.

Companies first hired data scientists, and only data scientists. They have been seen as a miracle solution, profiles at ease with all the components of data science: from the finest mathematical knowledge to the most complex database computing, to an ability to conceptualise business and industrial processes. These people are computer scientists, statisticians and business consultants rolled in one, the unicorns: we all dream of them, but they do not exist.

With this fantasy in mind, the data-related positions are often defined in a vague way, even though data jobs are constantly becoming more and more specialised. For example: you can not be a data scientist expert in both computer vision and natural language processing. All the more since the reality of an everyday data worker’s routine is less about machine learning and neural networks but rather the cleaning, structuring, and moving of data.

Not to mention: data scientists alone can’t make good data products. For example, technical infrastructure can quickly become a problem and in addition to data scientists, it becomes necessary to recruit data engineers. Then, other problems take place: the lack of skills dedicated to the actual crafting of software or interfaces. It’s endless! Each time, all the initiatives start from scratch, time and budgets are running out and convincing results do not arrive.

In data labs, the competencies brought in-house tend to not match the needs to identify, implement and disseminate use cases that bring value to the company: the roles are too compartmentalised, too far from the business, with a lack of global vision. The challenge has been viewed the wrong way around, rather than focusing ourselves on what brings value to the business.

Let’s do something with data and hire data scientists.

To innovate with data, a data lab should integrate other professionals rather than only the data scientist, people capable of building strategic software and integrating them into an organisation. That is, consultants, analysts, product managers, data engineers, developers, designers, etc. This diverse team breaks the isolation of the data lab, creating value for an organisation.

Myth #2: value is in the pipes

The second problem has been to take refuge in technological solutions, focusing on the tools and their technical capabilities, rather than trying to demonstrate the value that can be created from data.

Thus, many companies have focused on the data pipeline, the “data plumbing”, in order to be able to respond effectively to the operational challenges that arise from the daily work with data. They have invested heavily in technology platforms: cloud architecture, data science platforms, business intelligence tools, etc. The landscape has become more and more complex in recent years.

Who has not heard of a data lake project? Centralising all the data, structured and unstructured, is indeed a nice promise. But by not asking the questions about the usage, the usefulness of the data and the value that we will draw from it, this promise becomes the first reason for failure. Implementing these holistic technical architectures is time consuming and expensive. There is a high risk of ending up with an infrastructure that does not respond effectively to the most impactful use cases. Even though some of the more mature companies have overcome the challenge of streamlining their data processing pipelines by investing heavily in data engineering. This has required considerable resources… that not everyone can afford.

It is much easier and less risky to develop some case studies at first, to demonstrate their impact, and then work on their scalability in terms of technical architecture. Of course setting up a data lake can be useful, but time is an essential element of a well-managed strategy: in the first place it seems more judicious to create value with the most impactful projects, with measured impact on the business and the organisation.

The main challenge of the data lab is a cultural one

The latest version of the NewVantage Partners study shows that only 7.5% of data managers cite technology as an issue to accelerate the transformation of their company (11 points less than in 2018), compared to 62.5% who quote the human aspect (14 points more than in 2018). In most of the data labs I have observed, it is obvious that the human dimension of a data-driven enterprise is often put aside or treated at the end of the chain. The impact on work, processes, the organisation itself, and how we bring the value created have been forgotten.

I have already written in a previous article about the place of humans in machine learning software and its impact on work. I did explain how the tech “champion” companies are addressing in a very serious manner the human aspect of creating data value, and the related strategies they are currently implementing. As an example, I invite you to take a look at the dedicated teams of Google and Uber on this topic:

— the PAIR team within the Google Brain department, and their Big Picture group, at the forefront of research about data visualisation in machine learning;

— the visualisation department of Uber, that designs and builds interfaces for the whole company, and beyond.

These companies have realised that investing in data science skills and tools is not enough: it’s not just about training people to data science, it’s about bringing the use of data to people.

A data-driven business is using data outside of the data lab

Uber internally develops many custom-built, data-rich interfaces designed to serve business experts who are not data-savvy. In doing so, Uber turns each of its employees into a data scientist. For example, City Operations Managers, a key position at Uber dedicated to the operations in each city, have access to software enabling data exploration in a simple and intuitive way. To achieve this, Uber focuses on people who can translate data into business needs and usage, that is, translate them into interfaces that employees can explore, analyse, understand, and then make better operational decisions. These interfaces allow the data to have a business impact. Uber’s competitive advantage lies in its interfaces, and therefore its consultants, product managers, designers and developers, working hand in hand with its data scientists and data engineers. They make possible a virtuous circle of information management. The information available internally becomes richer, shared, smarter. Thus interfaces make use of human understanding and analysis capabilities coupled with the machine’s ability to calculate, sort and rank.

Data labs must become human-data labs

The data is not the situation, just as “the map is not the territory”: it’s a representation, and therefore describes a situation more or less accurately. This is key to understand. The data is there to inspire the creativity of people and stimulate their analysis. Whether the core of a data-rich software is descriptive, predictive, or even prescriptive, it is the interface that makes data work for the business. Interfaces allow the creation and ongoing learning by giving people the capability to identify, interpret and communicate the information pieces of the puzzle and thus develop useful business knowledge. Once analysed, understood by most of experts, that data will bring interesting perspectives. Interfaces have the ability to speed innovation in a company with data, and ultimately leverage it. Human-data interactions must be designed with business processes and decisions in mind, where their usage will be the most impactful.

This work can no longer be underinvested in data projects. On the contrary, it must be the first priority, the keystone, that ensures a project lasts.

At Dataveyes, we put humans at the core of our approach, because we believe that data gains value when it reaches the hands and minds of users, rather than staying in data centres. Inspired by service design, our approach is user centric, and takes into account the maturity of the company’s internal culture. Our methodology leads us to analyse its company’s business processes, and structure its information landscape too.

Making data work for people is key towards an actual transformation of the company. Data labs seem to have put aside this aspect. Nothing is lost, it is about time to change the approach, leverage the bullet time effect of interfaces on their data, and transform data labs into human-data labs.

In need of a human-centric data revolution was originally published in Dataveyes Stories on Medium, where people are continuing the conversation by highlighting and responding to this story.

(A French version of this article is available here)

Ever since the Industrial Revolution, we have been constantly striving to optimize the way we work. From the first scientific organizations of production, such as Taylorism and Fordism, to the currently popular agile methods, we have been looking for ways to break down tasks and allocate resources to move towards greater efficiency. Nowadays, new data processing techniques, classified under the term “artificial intelligence”, question our relationship to work. So much so that some herald the advent of a 4th industrial revolution.

Without going so far, significant advances in AI bear questioning. Because, although the AI ​​dates back to the 1950s, the new techniques derived from it, such as machine learning, have profoundly changed the way we think and conceive computer systems.

AI is the new black (box)

Usually, software is explicitly programmed to follow procedures developed by an engineer. With machine learning, the system learns by itself. The engineer directly specifies the target objective, and the system then has to learn how to reach this goal, using a set of training data. The system achieves this task by performing intertwined operations on very large datasets. It is hard to figure out the path followed by such a system towards decision-making.

The operation of machine learning algorithms is much less explicit than that of classical procedural algorithms. Deep learning algorithms in particular may appear to us as black boxes. These algorithms are supported by deeply interconnected neural networks with up to several hundred million links. Each of them can learn, and contributes by specializing in a part of the problem. This is why it is so difficult for us to understand and represent these complex models. I have often heard: “We can build these models, but we do not know how they work”.

We cannot live with impenetrable systems

The algorithms in question are not reserved for a few data-rich professions or for mathematical research. Two recent examples prove that they are connected with what makes our society and identity:

— This software used by American police to predict future criminals, exhibiting bias against black people in its risk assessment.

— More recently, this algorithm that can “detect” the sexual orientation of a person from a simple photo of their face.

Of course the technical specifications and the source code sometimes provide enough information to insiders to allow them to work with a machine learning model. But this is not enough for the great majority of us, data specialists or not, to understand these models, and therefore trust their results. Not even data scientists can claim to read these algorithms as an open book.

However, our need to see clearly in these algorithms is all the more justified that they are not as reliable and robust as they seem.

Machine learning has emerged as a promise to solve problems that were up until now unsolvable by men. When a very large number of criteria are involved in the determination of a phenomenon, and can only be evaluated through gigantic samples, machine learning algorithms achieve useful modeling where classical algorithms reached their limits. Nevertheless, their modeling is only autonomous in a deterministic view of the world, that is to say, assuming that the analyzed data describe comprehensively the many facets of the problem to be solved. In real life however, the universe is not as perfectly delineated: it is difficult to clearly draw all the contours of a problem, and it is difficult to list all the possible answers on which an algorithm must be trained. Some answers may still be unknown, or may be neglected because of our own human biases. Machine learning algorithms mechanically reproduce these biases, and can only model part of the reality: the one we were able to expose to it. They give us the impression of better understanding the world, when in reality they merely reproduce the facets of the world that we already know. This is called quantification bias: the unconscious belief that drives us to value much more what we can measure than what we cannot. In an “AI-first” world, this phenomenon is exacerbated.

“What gets measured gets... attention.”

— Not Peter Drucker

As a result, it is difficult, if not impossible, to prove that a machine learning model will work in all cases for which it was designed, as some situations may be poorly represented in the training data. Unfortunately, correcting what is wrong with this type of case is a complicated task: since the underlying structure of the system is extremely complex, finding the sources of bias is tantamount to looking for a needle in a haystack. Not to mention the difficulty of constituting an important and healthy training dataset, representative of the problem that one seeks to solve.

For these reasons, we need machine learning systems to be more intelligible for both data scientists and the general public. Because an algorithm that can be interpreted is an algorithm that can be improved. We are still a long way off. Yet I think this is a crucial point for strong adoption of machine learning. When our societies realize that sophisticated algorithms have invaded professional and intimate spheres, they may reject en masse the presence of these black boxes.

This poses two challenges.

On the one hand, machine learning specialists need to better control the algorithms they design, be able to explain their decisions, and discuss them with a wide audience, in order to obtain safer and less discriminatory models.

On the other hand, other professions working with data scientists, as well as everyone in their daily lives, must be able to appreciate the effects of algorithmic models, and better understand them to gain better insight into them.

To succeed on these two points, I am convinced that we need to improve the way we interact with these complex systems: we must create mediation interfaces for algorithms.

HDI: the science of mediation between humans and data

We developed the habit of using the term “Human-Data Interactions” (HDI) to describe what we do at Dataveyes. HDIs include all devices designed to improve the way we understand and use the information contained in data. I encourage you to read our article on the subject, explaining this term and our approach.

In a conventional HDI approach, not confronted with machine learning, humans are at the heart of the system and do most of the work: they get information by interacting with the data, and use it to make informed decisions.

One may assume that an AI system and an HDI approach are incompatible, or even opposed, since in AI systems humans seem to disappear from the processing chain.

On the contrary, I think they are in fact complementary rather than contradictory, provided roles are redistributed. Humans and machines must each be responsible for meeting needs of different nature and complexity: the responsibility of learning, calculating, classifying, and so on, should lie with AI systems. And humans should bear the responsibility of understanding, analyzing, sensing or experiencing reality. The two should therefore work together in confidence.

Confidence and understanding are linked. Understanding is an essential objective of human-data interactions because it makes trust possible. HDIs address this issue through design: the design of interfaces, interactions, and information flows must provide humans with the understanding necessary to perform their role beside data-rich systems.

Taming AI through design

What interfaces can help very different audiences — data scientists on one hand, and non-experts on the other — to better understand algorithms? Interfaces that focus on data visualization and interactivity.

Such interfaces do not aim to teach us how to read the mathematical formulas of algorithms. They instead display how algorithms transform data, they help us understand and deduce what these algorithms produce and how the system operates, by showing us the structures, groups, hierarchies, distributions, relationships and correlations in a dataset.

Often, these interfaces feature filters, zooms, cursors to move, buttons to click, etc. Because an important aspect of human-data interactions lies within interactivity. When a visualization simulates data that evolve over time, or when it allows us to vary parameters, it reveals to us the influence of input variables on output variables. The interface gives us a feel of the sensitivity of the data, and allows us to build a mental picture of their interdependencies.

Such interfaces can remain simple and playful, thus allowing the general public to gain better insight into the main algorithmic systems that surround it. They provide non-experts with the level of information necessary for a trustful relationship with data.

In their most advanced versions, visualization interfaces can also help data scientists evolve their work methods.

Towards better control of AI by data scientists

To meet the challenges of AI in our societies, data scientists must be able to thoroughly audit the mechanisms of machine learning and provide context for the answers it offers. To do this, they must extend their scope of intervention to incorporate good practices:

— Properly preparing data and ensuring its quality. In most cases, algorithms are not fed with raw data, but with data re-processed to bring out features relevant to the problem.

— Auditing the operation of the whole system by probing its internal workings.

— Analyzing results by testing their relevance.

With appropriate visualization tools, data scientists can see what happens to data when they initiate and set up a machine learning model. Appropriate visualizations promote their understanding on both a global and local scale: at any given moment in time, they display the whole dataset and a granular view of certain parts of the set. These visualizations allow data experts to more easily test stability or important specific cases of a learning algorithm, at each processing step.

These days, such visualization tools are increasingly used by specialists to carry out the design and training of machine learning models. Many initiatives are heading in this direction:

— Fast Forward Labs recently released a research paper on the concept of Interpretability in machine learning models.

— Google has shared open source visualization works on the topic of machine learning, with Facets and TensorFlow.

— Uber builds its own machine-learning-as-a-service platform with a strong visualization focus.

— The open source Deep Learning platform H2O.ai has also made a lot of efforts in this direction.

— Finally Distill.pub wants to be a platform dedicated to the explanation of machine learning.

Thus, a new approach is emerging, where the AI ​​system is supervised by data scientists who are themselves assisted by interfaces, upstream (input) and downstream (output).

Extend our natural ability with AI

If machine learning algorithms are better when they are assisted by humans, the converse is also true: men can go beyond their limits thanks to machine learning systems.

Joi Ito, the MIT Media Lab manager, gave an interesting perspective on the design principles for working with AI. According to him, humans should concentrate on the concept of extended intelligence rather than on robotics and AGI (artificial general intelligence), for it is the essence of humans to use technology as an extension of themselves.

I would go further by drawing a parallel with the concept of geosophy, introduced by geographer John Kirtland Wright in 1947 in his essay Terrae Incognitae. For him, geography is an “academic” science that must be enriched by what he calls geosophy. Geosophy extends the knowledge of geographers to peripheral conceptions: those of farmers, fishermen, business leaders, but also poets, novelists, painters, etc. J.K. Wright does not hesitate to forge a link between imagination, intuition, knowledge and science. He preaches thus for the widest possible knowledge of territory.

This approach is similar to the one we propose when discussing human-data interactions. HDIs include data science as geosophy includes geography; but they also go beyond data science, integrating the point of view of end users: practitioners of all trades in contact with data, but also inhabitants of smart cities and connected houses, drivers of autonomous cars, users of networks, citizens, etc.

Like geosophy, HDIs mix instinct and knowledge. They use machine learning models to improve our perceptions, and develop our intuitions. In so doing, they enable us to extend our decision-making and creative capabilities, to solve problems more efficiently, and to understand in depth the world we live in.

In business, HDIs bridge the gap between data specialists and domain experts. They enrich a common knowledge of data, favoring a data literacy essential to the improvement of everyone’s work. Organizations that smartly combine design and machine learning will therefore gain a definite advantage over others.

So it is not man versus machine, but rather man AND machine, trying to learn from each other. Our future will be algorithmic, and design will assume an increasingly important role in it. Not only because it allows a safer and more productive relationship with machine learning, but above all because it allows a more collaborative work.

How will design and machine learning enable physicians, technicians, farmers, designers, musicians, etc. to improve their professions? At Dataveyes, the company I co-founded, we are interested in the complete spectrum of interactions between humans and data: from the practices of computer engineers to those of experts in all fields, to users of the products and objects of tomorrow. We are firmly convinced that we can open up new possibilities by designing AI systems that integrate the human dimension at their core.

Design is the keystone to reveal the potential of AI was originally published in Dataveyes Stories on Medium, where people are continuing the conversation by highlighting and responding to this story.