<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by kathryn waldron on Medium]]></title>
        <description><![CDATA[Stories by kathryn waldron on Medium]]></description>
        <link>https://medium.com/@kathryn.j.waldron?source=rss-ce7d000fd342------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*2KOEhD_begNRoD-AL0-1sA.jpeg</url>
            <title>Stories by kathryn waldron on Medium</title>
            <link>https://medium.com/@kathryn.j.waldron?source=rss-ce7d000fd342------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 23:43:48 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@kathryn.j.waldron/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Defining data products]]></title>
            <link>https://medium.com/@kathryn.j.waldron/defining-data-products-4a0178c13b69?source=rss-ce7d000fd342------2</link>
            <guid isPermaLink="false">https://medium.com/p/4a0178c13b69</guid>
            <category><![CDATA[data-as-a-product]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[data-product-manager]]></category>
            <category><![CDATA[data-product-creation]]></category>
            <category><![CDATA[product-manager]]></category>
            <dc:creator><![CDATA[kathryn waldron]]></dc:creator>
            <pubDate>Fri, 09 Aug 2024 09:23:34 GMT</pubDate>
            <atom:updated>2024-08-13T05:15:06.788Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Y_NTb8ovnvfata4sxQsiNA.png" /></figure><p>After years of software product consultancy, being thrown in at the deep end as a data product manager in 2020 was exciting and challenging. Since then, I have had the privilege of working on a diverse array of data products, ranging from risk engines to using remote sensing and computer vision for assessing habitats.</p><p>Initially the world of “data projects” bewildered me — trying to shape the project ask into a product and convince team members and clients to see the value of viewing them as such. Fortunately, attitudes have evolved.</p><p><strong>What is a data product?</strong></p><p>Breaking it down</p><ol><li>The product is driven by data. All products have a data element, but in data products the data is the offering.</li><li>It is being managed as a product, not as a project. It has a product manager, a focus on the outcome and continually improving the product. It is not delivered and thrown over the wall.</li></ol><p>As a data product manager, there is a lot of commonalities in how you approach data products. You will always need to be thinking about your users, their use cases, their skill and capabilities. As you are building a product with data at its heart, data quality, timeliness, and accuracy need to be central to your thinking. Trustworthiness will be central to the success of your product. Understand business requirements like security, access and permission management and audit capabilities.</p><p>Within this, I have found it useful to categorise data products. It provides a framework that guides my approach when joining new engagements.</p><p><strong>1. Data</strong></p><p>This data product type provides users with access to raw or derived data. As a data product manager, you need to recognise and exploit the value of the data itself. Users — internal or external to your organisation — may want access to the data for a variety of use cases , such as analysis or to build a further product on top, and you need to shape your offering accordingly.</p><p>Outcome measurements usually focus on the number of unique users of your data and the volume of transactions.</p><p>I include derived data in this. Derived data can add value, for example through calculations for common organisational definitions that enhance usability and consistency.</p><p>Analyse user needs around themes such as exploreability, levels of friction for change, performance, and availability. If humans are accessing your data considering user capabilities and skills when designing data access methods.</p><p>Examples include:</p><ul><li>Providing access to vehicle roadworthiness data can help users such as fleet operators to manage their vehicles</li><li>Satellite imagery (data is not always numbers) can be downloaded from providers so users can assess land usage</li></ul><p>The remaining three data product types build upon data as the foundational product (if it is managed that way). These types vary in their reliance on the data product for decision-making. As a vigilant data product manager, understanding these nuances is crucial to navigating risks.</p><p><strong>2. Insight</strong></p><p>Insight products provide users with analysed data that drives informed decisions. These insights are often presented as dashboards, enabling users to visualise and interpret complex information.</p><p>Start with understanding the actions users are looking to drive from the product so your product is tailored to their needs. Consider the levels of accuracy and timeliness needed to support these actions. Analysis needs to be clearly presented so users can easily and correctly interpret the information.</p><p>Examples include:</p><ul><li>A command centre dashboard in a contact centre enabling real time management of agents against demand</li><li>A performance dashboard for a digital service manager, showing the health of their service and highlighting areas for improvement</li></ul><p><strong>3. Predictive</strong></p><p>Here the data product’s output is a prediction, or a recommendation, with a human involved in final decision-making.</p><p>You need to have a clear shared understanding of what you are predicting and how that will be used, including expectations around accuracy. Explore the impacts of being wrong (both in terms of false negatives and false positives) and how outcomes get fed back in to improve your product. Ethics and guardrails can become essential as you consider the implications of your predictions. Monitoring needs to cover changes in your data or model output distributions so you can react quickly to unexpected changes.</p><p>You are often now in the world of ML, which is daunting and often not well understood. As a product manager you need to ensure your users and stakeholders grasp the product’s nuances for effective decision-making.</p><p>Examples include:</p><ul><li>A chatbot answering queries from your customers</li><li>A safety product that uses harsh braking data from connected vehicles to predict risky locations on our roads so prioritising investigations by a road safety engineer</li></ul><p><strong>4.Decision making</strong></p><p>Data products now take centre stage in making decisions. However, humans remain integral to the process. Humans may be involved in areas such as providing domain expertise, annotating training data, quality assurance and feedback into the model — as they could be in a predictive data product.</p><p>Essentially, everything that has been said about predictive data products applies here but is now heightened — assuming a more significant impact of decisions being made. The consequences of being wrong, how you will monitor for this and how you can recover from it need to be fully researched, understood by stakeholders and users and agreed to.</p><p>Examples include:</p><ul><li>A credit risk scoring model used to automatically accept or decline credit card applications</li><li>Autonomous vehicles controlling steering, braking and acceleration in real time</li></ul><p>As with any framework, this gives me a starting point when I begin work on a product, but you still very much need to consider the nuances of your product.</p><p>It would be great to hear how other people navigate the world of data products!</p><p><strong>Further reading</strong></p><ul><li>In the early days, a colleague directed me to a blog by Simon O’Regan, <a href="https://towardsdatascience.com/designing-data-products-b6b93edf3d23">Designing Data Products</a>, which provided a great starting point for understanding what data products are.</li><li>“<a href="https://martinfowler.com/articles/data-mesh-principles.html#DataAsAProduct">Data as a product</a>” is one of the four underpinning principles of Data Mesh, an approach designed by Zhamak Dehghani. Her thinking has been pivotal in developing out data products.</li><li>More recently I found the eBook <a href="https://www.dataops.live/dataproductsfordummies">Data Products for Dummies</a> from DataOps which is an excellent overview and has helped me develop my thinking further.</li></ul><p>Image from <a href="http://www.freepik.com/">www.freepik.com</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4a0178c13b69" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>