Are all support ticket processes created equal?

Sadly, no — looking through social media posts is a fast enough way to see that getting help is a complicated matter. Customers want answers now, and businesses are often trying to handle hundreds if not thousands of complaints of various severity. And that can be in just one day!

So how do you balance collecting not only the right amount of information, but also the right kind of information so that you can resolve the issue quickly without placing undue burden on the customer? How do you simplify the case creation process not just for customers, but also for the human agents who just want to help?

This comes down to case creation UX design. Submitting a support ticket doesn’t have to be a cringe-inducing experience.

We’ll outline eight best practices that we’ve discovered through research and user testing.

Best Practice #1: Split the User Journey into Steps

Chunking, according to the Nielsen Norman Group, is breaking a larger item (whether that’s a concept, content, or, like here, a process) into smaller pieces. This helps users process and recall information. By breaking a process down into steps, customers can focus and provide the info you need to help them.

In testing our Case Assist tool with a group of early adopters, we found that they often used between five and 10 questions in their case creation process. Especially when presented with other hurdles like a required login page or document recommendations that attempt to deflect the creation of a case, that’s a lot to dump on a customer. While there is no “right” number of questions to ask, the fewer, the better.
Simplify by showing the whole process before a customer even starts-this sets expectations of what’s to come, and even primes them to provide certain information at certain times. It can also help motivate customers toward completing tasks! Progress bars are great for this.

Best Practice #2: Start Off Easy

Ask for login only if required

Asking for registration and login might stop a case before it’s started. Always explain why you need a customer to provide their contact information-they’ll be more inclined to do so when they understand how they benefit as people need to trust you before giving personal information.

Our research included asking testers to submit a case in a B2C scenario, where they had an issue with a sports watch. Most were reluctant to log in because they didn’t feel it was necessary in that scenario. “It’s too serious; I wasn’t expecting to have to log in,” one of our testers explained. “I’m used to navigating without having to login.”

If your customers already have an account and are used to being logged in, it might not be an issue as the trust is already built. One of our testers said, “For SaaS, I’m OK with being asked to log in because I’m always logged in anyway. It makes sense to have an account if it’s a tool I’m using on a daily basis.”

Be clear about the main goal of the form

Seventeen percent of our testers weren’t sure if the information they provided in the form was going to reach a human being. Generic labels like “Get help” or “Have a problem?” can refer to other types of help resources like self-service documentation. With all the different support channels available that don’t always involve speaking to other humans, customers might be unsure of where your form will lead.

Avoid ambiguity and say that a human will read their case! It can impact how customers describe their issue. In our test, we saw a user use only bullet points because she wasn’t sure it would actually reach anyone.Testers that did not express doubts about their case reaching a human wrote descriptions in more complete sentences.

Best Practice #3: Help customers describe their problem

Within each step of your case submission process, provide the fields customers need to answer the given question. A good rule of thumb is about five fields per page. Think carefully about what fields are truly important for a customer to gain a resolution to their problem. And if you decide to use optional fields, make sure they bring value and mark them clearly.

If your company usually handles complex issues that might require long descriptions or steps to reproduce, providing text formatting options (bold, lists, paragraph, etc.) can allow for clearer descriptions.

Text field sizes should match the expected input length. If you know exactly how many characters your input requires, the field should only allow that number. For example, if your error codes contain 10 characters, the field should be 10 characters long.

Guide users on what you expect as a description

Sometimes it’s hard for your customers to know where to start when describing a problem. Maybe it’s occurred erratically-enough to be annoying, but not on a consistent basis to know what’s causing it. Or if their focus is solely on the fact that they’re not getting a desired result, they may not realize they’re missing a crucial step. Help users help you by providing tips on what to include in their case description. To achieve this ourselves, we tested a description strength component that showed a customer’s progress while typing.

With a circular progress bar, our testers understood how effective we found their descriptions and were encouraged to add more. Still, some quickly exhausted what details they could think of. “I have to add more?” one commented. As shown above, make sure to provide hints on what type of information you need.

Best Practice #4: Streamline Case Classification

Classifying customer issues is a quick way to ensure problems are routed to the person or group best capable of handling and resolving them fast. Simplify the process with artificial intelligence, such as with the case classification feature of our Case Assist tool.

By adding Case Assist as a second step, the tool reads all of the input from the case description page to suggest the most relevant categories to your customers. This is also a great place to ask for the “severity level” of the case as well.

Show a limited number of suggestions

Giving users too many choices can be overwhelming. Case Assist recommends the most relevant categories, allowing customers to avoid browsing through a long list. It would be best to only show three suggested classifications, which is a good rule of thumb to avoid overburdening customers with choice and ensuring the suggestions stay relevant.

Allow customers to choose an alternate classification from a drop-down that shows all of the choices. If a choice is made in the droplist, the suggestions made by Case Assist disappear, leaving only the droplist accessible. That way customers focus on one element at the time.

How you ask for the severity level matters

Severity level is a difficult concept to quantify, because what’s severe to one person isn’t to another. In our research, testers often selected a higher severity than necessary to avoid “being stuck in the backlog forever.”

Testers questioned why they had to provide that information at all. “Why do they want to know what the severity [of the problem] is? I still want my problem to be resolved!” One commented.

Instead of bluntly asking users to rate their issue from high to low, we recommend giving concrete factors to choose from that might affect case severity. As one of our testers said, “Low-medium-high is kind of the worst way [to rate severity]!”

Best Practice #5: Suggest helpful documents for faster resolution

To aid in self-service (and ultimately case deflection!), this is a good step to offer documentation, videos, or other content that might solve the user’s problem. You can do this via a recommendation system. By pairing this with machine learning, your case creation process can understand the user’s issue and suggest relevant solutions directly within the form. Our Case Assist tool offers such functionality.

How should these documents be displayed in your support request form? We’ve got a few ideas.

Give help documentation its own step

Choosing where to showcase help documentation is an important decision-to find out what works best for your customers, test! During our tests, putting document suggestions on a page of their own increased the clicks on documents by a factor of 2.5. This might be because users have only one task to complete on the page; i.e., look at and evaluate content suggestions without any distractions. However, what works best for your customers can only be decided by your customers, so test different configurations and collect feedback.

Even if you decide to offer related help content, make sure customers can skip this step and complete their case submission process if they want to. It’s important not to block your customers from moving past this step!

Show a limited number of documents

To save time, customers will only click on links that seem like a solution to their problem. And we already know that too many choices makes it harder to choose. Most importantly, users have most probably searched before contacting your support.We would recommend showing between three and five documents, as it is a good balance between providing help and not overwhelming them.

Present documents in an accordion to track more data

Displaying documents in a traditional list works as well, but we suggest using an accordion to display documents for two reasons.

First, it allows for progressive disclosure, which aims at minimizing the chances that users feel overwhelmed by what they see-basically, by not showing everything at once. Instead, let users trigger and control what gets displayed and when. This helps users concentrate on one document at a time.

Second, you can track more data. Was a document ignored because it wasn’t helpful, or because the user didn’t see it? With an accordion, you can track one more data point: whether the document was expanded or not. A click to expand the accordion gives certainty that the document has been seen, but no click to open might imply that it wasn’t what the customer was looking for.

Define how your documents will open

Now that you know how to display the documents in the form, how should they open? We suggest two options: in a new tab or in a quick view option. Here’s a list of pros and cons to help you decide.

New Tab

Quick view

Show how content has helped other users

Customers value the opinions of other customers; as one of our testers put it, they found community commentary “more honest” than corporate information. But you can leverage those opinions to help customers find answers.

Whether with symbols, numbers, or both, you can show how helpful users have found a particular piece of content.

“I love seeing how effective an article has been at resolving a problem,” one of our testers said. “It guides where I click.”

But to display such a rating, you need to first get feedback.

Best Practice #6: Nudge Customers For Feedback On Content

How do customers really feel about your content? There’s only one way to find out, and doing so helps you identify content gaps and better understand your customer’s journey. This is why we added a content evaluation option in our tests, as well as a way to explicitly say they no longer need help after consulting suggested help documentation (i.e., confirmed case deflection!).

Unfortunately, our research showed that few users take the time to provide feedback on content helpfulness. “I don’t have the reflex to mention when it’s [the document is] helpful,” one tester said.

This doesn’t mean you should make feedback mandatory! Instead, here are some suggestions to gently nudge customers for feedback.

Ask for feedback at the right time

Waiters wait until you’ve eaten a couple of bites before asking if it’s good. So wait until after users have read content before asking if it helped! Thus, feedback options should be at the end of documents.

Since there’s a good chance they’ll ignore it, gently remind them to provide feedback at key moments. When users dismiss content is a good time to ask for feedback. If you offer a quick view option, you can track when a user dismisses that view without providing feedback. Once the user has returned to the form, you can ask again if the content was helpful.

Show that their opinion matters

Part of what drives us to provide feedback is knowing that our opinion might help others. Why not tell your customers exactly that? Provide visual cues that the feedback was received by changing the Feedback Widget’s color or wording. Even better, show that their opinion impacted the helpfulness score by adjusting the score based on their feedback.

Offer a deflection confirmation

If customers found what they were looking for with the suggested documents, let them tell you. Even if most users will not use the “Abandon Request” button, it’s still important to design a step that closes the cycle.

Instead of closing the form right away, offer a modal asking for confirmation that the user no longer needs to submit a case. This way, if a user accidentally clicked the button, they have a chance to avoid losing their progress. It also provides confirmation that they found an answer to their question or a solution to their problem-case closed!

Best Practice #7: Decide If Your Customers Need a Summary

Should you have a summary or review page before users can submit their case? It’s hard to be sure if a summary/review page is helpful to users.

Best practices (if you’ve been following our guidelines!) say that the case creation process should be quick and require minimal information. With this in mind, a review step seems a little superfluous, doesn’t it? The customer was there throughout the whole support journey, filling out the form. What’s even more important is that, even though they’ll have submitted their case creation form, this isn’t the end! It’s the beginning of the conversation with a support representative. Unlike a purchase that is not editable, cases will evolve during the help journey.

The final decision is yours; maybe your industry requires specific information and intensely detailed descriptions. Because of this, it could be reassuring for your users to be able to review their inputs for errors.

Once on the summary step, avoid asking for more information. We saw some request forms where additional information was asked at this step. Users would not expect to provide more information on a summary. If you ask for additional information, it’s not a summary!

Best Practice #8: Always Confirm the Case Was Submitted!

How do users know when the case creation process is complete? With the confirmation page!

Users should understand at a glance that their case has been submitted. Cues like green check marks and a short explicit title work well. To end the journey on a high note, add delightful imagery or animation. Last impressions are what we remember best!

Offer users an estimation of when they should expect to get a response. Our testers appreciated being able to choose how they would be contacted by support staff-if you can allow that flexibility, it could be a plus!

The Beginning of Your Customer Support Journey

Like we mentioned above, the case creation process is really the first step of a much longer conversation between customer and agent. Making sure the support ticket steps are fluid and capture all of the necessary information just assures that the conversation will be a productive one!

Customer support is an iterative and growing process that does and should change over time; as customer needs change, as products evolve, and as businesses grow. What might work for you today may need to change tomorrow.

These best practices give you somewhere to start, but remember-research, test, rinse, repeat!

Dig Deeper

You’ve looked through how to improve the customer side of the support journey — what about your support agents? Find out why they need an insight panel.

Interested in learning more about self-service? Check out our 6 Techniques for Self-Service Success ebook.

8 Case Creation UX Best Practices was originally published in Coveo Behind the Click on Medium, where people are continuing the conversation by highlighting and responding to this story.

If the last few years have taught me anything, it’s get away from the comfort of my desk. Go outdoors as much as possible, travel, listen to your heart (and your impulses) . . . and never forget your phone.

In my home/work sanctuary, I have learned how much my phone offers a superior digital convenience to me. The use of it — or inability to use it — becomes a deciding factor for me.

I order Friday night dinner delivery while traveling in a taxi. Or buy cat food when playing tennis with a friend. Or check on the support ticket for my paper magazine subscription while waiting in line for my morning coffee. (I haven’t received a paperback issue and won’t let it slip!)

In short, I’ve developed a taste for digital convenience, so now my phone is my best friend. I like easy access to mobile digital experiences. It feels like a super power, an extension of me that lets me achieve my goals fast. And I’m not alone!

Mobile digital experiences are fast becoming a staple in everyone’s lives. Yet not all businesses seem to be clued into that. From subpar mobile site performance to poor mobile site search, a basic digital experience doesn’t cut it anymore.

Companies are grappling with relevancy, especially as we close in on tough economic times. The digital convenience factor has to make its way to the top of the CIO’s mind.

What Is Digital Convenience?

The NPD Group noted that customers are “always looking for the next level of convenience, like not getting out of the car, delivery and drive-thru orders have taken hold, making carry-out the less convenient service mode.”

As my examples above show, it’s not just food: everything in life is looked at through this lens of convenience. And with people on the go, or multitasking, the conduit to all these services and products is their mobile phone.

In my remote job, I specialize in building a product that helps customers realize and scale digital transformation. As a result I have peeked behind the curtain of many companies who say they are customer-centric. Yet their mobile offerings make me worried about the companies’ longevity.

Do they know how much convenience is a competitive advantage for them? And do they realize how much they’re lacking?

When you deal with the public, you can’t just say “your baby is ugly.” So I came up with a quiz to help raise awareness that maybe things aren’t quite up to snuff.

Digital Convenience Factor Quiz

• How many clicks does it take your customers to find an answer they are looking for?
• How do you detect the intent of the user journey on an anonymous digital property?
• Can your customers find things if they don’t know the exact phrase?
• Does your search bar offer type-ahead or drop down-menus to help customers specify their query?
• Do you support natural language search?
• Can you do all of the above from a mobile phone?
• How thoughtful is your UI/UX on a mobile screen? (Alternatively, is the ratio of web/mobile traffic proportional to the UX/UI work you invest in for web and mobile?)

The answer to all of these questions is simple.

If you’ve never thought about any of the points and have no ability to track digital experience convenience, you run the risk of losing a vital audience, creating journey friction, and aggravating customer frustration. Act on changes fast!

Now, here’s the next big question: where do you start to craft digital convenience experience for your mobile search?

How to Craft a Digital Convenience Experience?

Here is how I work with enterprises and their digital properties to create a solution. It is already difficult to achieve a perfect balance between complex customer demands and infrastructure scalability.

Starting with your site search engine is the perfect solution to bringing content teams together into one unified search platform. Then deliver this content via web or mobile to your audiences, all with unique needs, intents, locations (hint: Location-Aware Search). A mobile site search engine, in this case, points to specific content and feeds back analytics to better understand your audience, your mobile app or mobile site behavior, and your content consumption.

To achieve this performance, Coveo customers use Coveo Headless and Atomic libraries to deliver search results and recommendations at lightning speed. Developed and maintained by Coveo for the developer community, they are 10 times faster than the classic JavaScript search UI library.

Does Coveo Headless ring a bell? No surprise that a part of having a strong mobile strategy is having a headless infrastructure is a part of having a strong mobile strategy.

Another trend is moving from a web browser window and building an entirely different experience on a native mobile app. Whether using React Native, building for Android or iOS, mobile apps unlock unparalleled performance and grant full control of the experience in the app.

Cross-device search engine is the unifying thread for digital experiences. Coveo-driven experience can be configured for any journey and any device using the same Administration Console, even if beyond the scenes web and mobile utilize completely different technologies and are done by completely different teams.

Whether your customer buys cat food while playing a tennis match on a Sunday afternoon, or resolves her paid subscription issue while ordering a latte, mobile site search engines deliver on the digital convenience promise for the generations that seize the moment.

DIG DEEPER

Mobile is one of many channels you should maintain. So, when it comes to offering an omnichannel experience, choose a platform that supports your endeavor.

Learn how Hackensack Meridian Health leveraged Coveo to achieve just that.

Mobile: The Gateway to ‘Digital Convenience’ was originally published in Coveo Behind the Click on Medium, where people are continuing the conversation by highlighting and responding to this story.

Search is a very complex problem hiding behind a very simple interface.

From the earliest days of the internet, people have been trying to find web pages (documents, though they don’t always appear as such) that meet their particular intent. To do so, they go to a search box and type in a keyword or phrase such as “cat videos”, and out pops a list of pages showing all kinds of cute cat videos.

This process is so intuitive that the search paradigm, as described here, is never really questioned: you type or speak a search term to a query engine. Then machine learning algorithms provide the (hopefully) correct search result back in the same format.

However, there are problems with this paradigm. Suppose for instance, that what you’re looking for doesn’t fit into a neat category (i.e., is unstructured instead of structured data), or involves very specific, contextual information. Google calls these complex questions.

Here’s an example: Ask Google search about what toys you should release for Christmas this year to make the greatest profits. You’re likely to retrieve ambiguous information at best. Ask which departments had the largest revenues per customer in your company is likely to get even less information than that, possibly because Google doesn’t have your sales data (and preferably shouldn’t). But more likely because Google simply didn’t understand your question.

This is where natural language search comes in.

What Is Natural Language Search?

Natural language search, which uses a machine learning technique called natural language processing, allows users to conduct a search using human language.

Here’s an example of a traditional search query:

Catskill Mountains height

And then reformatted as a natural language query:

How high are the Catskill Mountains?

You can easily see that one is more conversational than the latter, which is why natural language search is also often referred to as conversational search.

Natural Language Search vs Keyword Search

The Evolution of Search

Traditionally, search engines conducted a search by matching the keyword or keywords queried against its index, also known as keyword-based search.

To narrow down these results put the onus of a search structure onto the searcher, requiring users to input long or complex queries. Sometimes they’d have to use boolean operators to communicate better with the query engine.

Because their queries had to be formulated in a certain way, this also often removed the intent that underlies everyday language. Marketers would then have to undertake a sort of manual sentiment analysis, where they would painstakingly mine search query logs to try and better understand what their searchers were looking for.

Search has come a long way since. Whether typed into a search box or spoken aloud to a voice search conduit like Siri, Cortana, or Alexa, users can pose a question to a search engine much in the same way they’d ask a friend.

A Brief History of Natural Language Search

The first application of NLP technology for search was the START Natural language Question Answering Machine, created in 1993 by the MIT Artificial Intelligence Lab. It’s not a web search engine, but it does let users query an online encyclopedia of information using everyday language.

Rebranded to simply Ask.com in 2005, Ask Jeeves was launched in 1996. It was the first web search engine that allowed users to search the internet using natural language search.

Google came along two years later, and the rest is history. With the release of Google BERT (or Bidirectional Encoder Representations from Transformers) in 2019, Google searches were able to understand the full context of a word by looking at the words that came before and after it, instead of one-by-one in order.

About That Complex Question, Or Unpacking a Natural Language Search Example

Natural language search is a shift in thinking about search that goes beyond keyword matching. While keywords (or key concepts) are still important, with natural language understanding the intent is to gain more contextual information by analyzing the questions (queries) being asked, and then using these to help understand the context of those queries.

For instance, consider the question of what toys will be the best sellers this holiday season. This seemingly simple question is quite complicated from a search standpoint because it involves a number of implicit assumptions:

• The query engine can “parse” or break down the question into component sections, and build queries based upon those sections.
• You have, within your data system, entries for toys that you specifically sell, that contain data about how they are selling now, and information about how they sold in the past.
• The query engine can distinguish between your products and your competitors’ products.
• The query engine understands the idea of Christmas as a sales period.
• The query engine can analyze information not only about what’s trendy within your current inventory, but information about what is currently “hot” in the marketplace.
• Upon information retrieval, the system will need to know what you mean by “Show me” (Display the search result in a summary? Tables? Charts? Graphs? Read aloud? Put into a spreadsheet?)

Some of these operations are fairly easy to ascertain and can happen almost instantaneously. Others tend to be ongoing, need a lot of analysis, and can only be queried when put into some kind of indexed structure.

Moreover, the speed of such queries can vary. The first time your best-selling Christmas toys question is asked, it may take some time to assemble this information. Once collated, subsequent retrievals can occur much faster.

The trick is in identifying, given two questions, whether they are similar enough that some or all of the answer can be retrieved quickly from the index without having to do the expensive computations.

Such a search also needs to be sensitive to the person asking the question. Some information may be available to the CEO that might not be available to a visitor to the company website. Thus, the context for such queries includes determining who should be told what, determining what is currently embargoed content, and determining which information cannot be passed on due to privacy regulations.

Natural language search also overlaps with speech to text, the process of converting oral to written speech (and back). This happens via a stochastic (probabilistic) pattern that analyses phonemes (distinctive units of sound) and matches them with specific words or phrases.

Probability comes into play here because specific phoneme combinations usually with distinct phrases that occur in a certain order, and by working with a large enough data sample through machine learning, the number of potential words or phrases often can be reduced from thousands to dozens or less, a process that Coveo takes advantage of.

How Coveo’s Natural Language Search Helps You

When it comes to applying NLP technology to search, Coveo relies on modern tech at the bleeding edge of natural language processing research. Our in-house NLP team is focused on identifying and productizing the best approach for a given use case.

Built on the ubiquitous Word2Vec and BERT neural architectures, we don’t reinvent the wheel. Instead, our extensively trained models are augmented with task-specific heads to apply a specific solution for a given use case.

Some of our in-platform applications include:

• Smart Snippets
• Case Assist
• Automatic Relevance Tuning (ART)
• Query Suggest
• Content Recommendation

Coveo is a powerful AI-powered platform that unifies and delivers relevant, personalized information for individual searchers — and then scales that up to deliver the same quality to audiences of massive size. Over time, all those interactions improve the quality of its service, evolving just as your business does.

It’s like a savvy librarian; not only does she know what books you ask for, but everything that goes far beyond your initial query.

What is Natural Language Search? was originally published in Coveo Behind the Click on Medium, where people are continuing the conversation by highlighting and responding to this story.

How do you choose between unified search and federated search? Well, imagine asking three friends for the best Italian food recipes.

The first time you do, one friend offers magazine recipes, another offers cookbook recipes, and a third gives you online recipes. To find the best, you have to ask each friend individually — then get each friend’s list of top recipes, and then sift through and compare the recipes to find the one you want. Possible, but very time consuming.

Let’s imagine another scenario. You have one Italian food-loving friend who has all kinds of recipes at their fingertips. But when you say, “Give me a recipe for your best Italian dish,” your hyper-organized friend gives you three separate stacks and says, “I put the best recipes from magazines in the first stack. The best recipes from cookbooks are in the second. And the best online recipes are in the third.”

This is certainly easier than having to ask three separate people. But you’re still on your own to figure out which recipe is the best overall.

It’s not hard to imagine an even better scenario. You have one friend who tracks recipes across various sources. When give them the same request for “best Italian food recipe,” your friend says, “Here you go!” and presents you with one list of Italian food recipes, ranked by awesomeness.

This list might include recipes from many sources, but what’s important for you is that the list is sorted by a category of your choice-that is, their awesomeness. You don’t care where each recipe comes from, and your friend knows that, which makes her the best Italian food-friend ever and your future go-to person when it comes to any sort of recipe.

Types of Query Systems

These three scenarios perfectly illustrate the difference between three types of search systems. The first one is largely outdated traditional search (which involved logging into several different systems and searching each one independently), the second one is called federated search (you log in once and search across content sources, but the different content sources still don’t talk to each other), and the last one is unified search (and your best Italian food friend) who knows that relevance is the most important thing when it comes to search and does everything to get you there.

With traditional search being largely-thankfully-phased out these days, it’s important to focus on the difference between the latter two: unified search vs federated search. What is that difference, and what makes one better than the other?

What Is Federated Search?

When federated search first came along, it was a huge step forward from traditional search. Traditional search required users to log into different disparate systems to find the different pieces of information they needed. That meant they had to know which data silo to look in, as well as remember all the passwords and logins for each database.

Federated search changed all that as it gave users a single interface in which to query across all these silos. While most traditional search methods only indexed a single data source at a time, federated search allowed indexing multiple data sources at once, and then presenting the results to users in one unified interface.

How exactly does it work? Federated search retrieves information from a variety of sources via a search application that’s built on top of search engine(s). When a user makes a single query, the federated search engine simultaneously searches multiple, usually disparate databases and ecosystems, returning results from all sources and presenting them in a single user interface.

Federated search became invaluable in complex organizations that have thousands of sources in the cloud and on-premise, where searching this via traditional search was simply unfeasible.

Still, there are both pros and cons to this method.

Pros

• Indices do not have to be standardized: Another advantage of federated search is that indices do not have to be standardized. If one dataset is in one format, and the other is in another format, that’s not a problem.
• Always up-to-date: Since each individual index is maintained separately, as long as each of them is kept up-to-date, the federated search engine does not have to consistently go back and index the source to make sure the content and results are up to date.

Cons

• Impossible to achieve search relevance: The main problem is that with federated search it is impossible to achieve true relevance. Because each index is searched separately, the results are ranked by source-and ranking by relevance across all content sources is impossible. (Remember your organized friend who keeps recipes from different sources neatly stacked in separate piles?)
• Poor search user experience: Users’ expectations from search today are sky-high. If you think a search experience consists of a search box plus a results page, think again. Users today expect much more from their search experience: they expect websites to interpret their misspellings correctly, anticipate their next move, and learn from their previous clicks and searches.

That means that a good search experience today requires different additional capabilities such as autocomplete, query suggestions, filtering and faceting, and so on. This is nearly impossible to achieve in federated search. Unless each separate index supports each of these capabilities, your federated search won’t be able to offer these.

What Is Unified Search?

It remains true and will always be true that content lives in multiple disparate systems and silos. That won’t change. But users don’t need to know whether you’re pulling your recipes from a magazine or a cookbook: they couldn’t care less, they just want to start cooking (or buy the best pair of shoes or the best spice grinder), and they want to get to their goal as fast as possible. The secret is to create a user experience where users don’t have to be distracted by irrelevant information such as which data source the right content is hiding in.

That’s where unified search comes in.

Unified search differs from federated search in that it creates a single unified index of content from across all content repositories — including the option to crawl or push content stored in the cloud or on-prem via connector library or APIs — and ranks the entire corpus in a single user interface, applying relevance across the board and delivering meaningful results.

With unified search, when the index is created, it captures all the information needed in order to apply consistent relevance ranking rules across all of your content for the queries which come in.

Here’s how unified search compares to federated search.

Pros

• Faster response times: Unified search has faster query response times than federated search. This is because results are sourced from one unified index as opposed to searching multiple systems with query-time merging. By contrast, in federated search, the response times are only as fast as its slowest content source.
• Better relevance: With unified search, relevance is applied across the entire corpus of content and can even self-optimize with the application of machine learning. That means that results will always be ranked by relevance, regardless of the source.
• Better UX experience: Unified search guarantees better user experience. Search results are blended and pages can be built from a component library or APIs. Sorting, filtering, faceting and other capabilities of the search UX are built on top of the unified index.

Cons

• Slower set up time: Unified search might take longer to set up than federated search, as the set up is dependent on the number, size, and location of sources. However, for most organizations, the additional time and effort are worth it.

Unified Search vs Federated Search: Which Is Right For You?

Let’s go back to our Italian food recipes for a moment.

Which of these three scenarios makes you more excited about rolling up your sleeves to cook? Likely, when you hear the first one and how much work it will require, you think “I won’t even bother.”

The second one sounds a bit more doable but still quite daunting (and makes you want to find new cooking-obsessed friends).

We bet the third one makes you want to jump from your chair ASAP and run to your kitchen and try making a mouth-watering dish.

In today’s world, people don’t want to spend time on searching and sorting. In fact, a good search experience today is one that doesn’t even require you to search.

Can you get there with federated search? We say that if your content lives in more than one database (as is often the case), it will be nearly impossible.

Here is a summary comparing unified search vs federated search, to help you decide which one is right for you:

Today the search UX is incredibly important to users. A good search is one where you barely feel you’re doing the work of searching, but one that gets you to your desired outcome faster than you can say “lasagna.”

Don’t believe us? Give our trial a spin, which will allow you to build a search prototype in minutes!

Dig Deeper

Wondering where your search experience falls on the search spectrum? Find out with our Coveo Relevance Maturity Model. You can also get a search audit done by experts!

Federated Search vs Unified Search, or How to Find the Best Lasagna Recipe was originally published in Coveo Behind the Click on Medium, where people are continuing the conversation by highlighting and responding to this story.

With the pandemic’s perceived closure, everything can go back to normal — right?

From inflation causing skyrocketing prices, to increased geopolitical tensions impacting supply lines (and lives), and, oh, Monkeypox making an appearance on the global stage, there’s still plenty to be worried about. And when people are worried, they consume information in an attempt to allay those fears. Or at least I do; don’t you?

For hybrid workers, bouncing between at-home and in-office desks, a return to the office was supposed to signal a return to how things were before. From watercooler chats to hitting pause on Zoom meetings, it was finally meant to a return to calm, to less, to some margin of relaxation.

And yet…

Today, like every day, 5 quintillion bytes of information will be created; in 2020, 1.7 megabytes of new information is minted every second for every human on earth; yet only 0.5 percent of all data is ever analyzed or used for decision making.

How Does Information Overload Impact Employee Productivity?

Many employees report that their productivity is in a data-induced death spiral — whether remote or at the office. According to a recent report in Tech Republic , a survey of 2,000 remote U.K. workers showed that 18% suffered from “information overload,” (also called “data overload,”) and 8% were overwhelmed by too much data and too many apps they’re meant to check each day.

But these frustrations are not limited to at-home workers. According to an article posted on WorkZone ‘s website, a study of managers in the U.S., U.K., Hong Kong, Singapore, and Australia yielded that 25% of workers reported information overload was leading to poor health and significant stress; 33% of managers also reported poor health due to a large amount of data.

Digital Data Tools Drowning Employees

So, what is the source of this soul-sucking, efficiency-crushing, progress-stalling data? The very companies that employees serve. That’s right — the digital data tools that employers think make employees nimbler, better informed, and more effective are doing the exact opposite. Consider a 2022 workplace relevance report from Coveo; the results revealed myriad alarming effects of information overload, including:

• Workers spend 3.6 hours daily sifting through a large volume of data for information needed to do their jobs — which in a workforce of 5,000 translates to 4,716,000 work hours wasted every year. IT workers spend a whopping average of 4.2 hours per day looking for information
• 60% of employees have to search within four or more data sources every day for the information they are looking for. An additional 18% hops between seven or more sources daily
• 50% of the employees now don’t know where to start their information hunt
• 41% of all information provided to employees is not relevant information to their specific job role
• 81% reported that they were unable to find information in critical moments
• 16% of the respondents are ready to quit due to data-finding frustrations
• Nearly 50% are less engaged in their work and feel less confident in their daily activities

When Coveo looked at the data collected in detail, it found workers were unsure if the incoming information they had was out of date (39%) accurate (25%), or even allowed to be shared (25%).

In short, people are overwhelmed, less engaged, and less confident. And literally ill.

The infrastructure that organizations build to make reams of information available in a few clicks, empower knowledge workers by providing access to information and data, and enable them to work from anywhere — unfortunately, are not always helpful. In fact, these infrastructures are adding to the toxicity of the information age.

Information overload paired with the perpetual barrage of developments surrounding the pandemic created a significant sense of mental fatigue for employees across the globe.

Information Overload = Information Anxiety

How much is this because people are stuck in their own heads ruminating? And how much is the stress of too many options?

And what happens when both phenomena work in tandem?

According to a recent article in Psychology Today, information overload can produce feelings of:

• Anxiety
• Powerlessness
• Overburdening

It creates a situation of ‘mental fatigue’ which has an adverse impact on decision quality. Employees face difficulty making decisions — or worse, it leads to them making hasty, bad ones.

What happens then is known as “cognitive overload”. When we try to exceed our processing capacity with more data than is possible, the author writes, we literally exhaust the human brain. And multitasking, a coveted skill in today’s work market, increases the release of the stress hormone cortisol and the hormone adrenaline, which regulates our “fight or flight” response.

Lindsay Sukornyk, CEO of transformational leadership company Alive + Awake , says that information overload can lead to burnout. “It’s like working out one muscle all the time — it quickly becomes injured,” she says. “People need to be mindful of when they’re reaching that point and take a break, take a walk, connect with other humans by taking a coffee break with a coworker. It’s a good idea to get out into nature for a little while.”

She added that stressful times such as the current Covid epidemic are a challenge. “Balancing cognitive awareness with other emotional influences can be stressful.”

Infobesity: Judgement-Making at Risk

Forbes magazine Councils Member (and giant in cloud business intelligence industry) Sébastien Ricard calls this data-glut pandemic “infobesity”; in a recent Forbes contribution he commented, “…consuming too much information can cloud our ability to make good decisions. The results of these decisions can affect the entire workforce, so it’s easy to see why there’s been a rise in companies working to combat this challenge.”

Looking at the business side of the equation, that uncertainty potentially creates new forms of risks. From governance issues (should this information be shared) to customer service (is the information being given even accurate). That same Coveo study found that customer-service mistakes are costly: people will leave the brand permanently.

Think of your employees as the overburdened Jacob Marley in Charles Dickens’ A Christmas Carol whose chain is described as, “… was long and wound about him like a tail; and it was made (for Scrooge observed it closely) of cash-boxes, keys, padlocks, ledgers, deeds, and heavy purses wrought in steel.” Imagine how Marley felt — that is what all that excess data is like for your employees, weighing them down and compromising their physical and mental health, along with that of your business.

As the Covid pandemic recedes and the months of panic, disruption, and uncertainty dissipate, a rare opportunity arises for reflection. It’s the perfect time to re-examine, re-evaluate, and retool the company infrastructure. Find new ways of addressing their employees’ information needs without drowning them in data.

Dig Deeper

Ready to put knowledge when and where your employees need it? Dive into quick tips and best practices around modernizing your intranet, empowering your team with AI-powered search.

Check out our Guide to Delivering AI-powered Employee Experiences.

The Risks of Information Overload in the Workplace was originally published in Coveo Behind the Click on Medium, where people are continuing the conversation by highlighting and responding to this story.

If you can count on one thing, it’s that customer service will only become more important for banking and financial services organizations. For growth-oriented companies, at least. Research from Salesforce found that growth-oriented companies are 22% more likely to invest in omnichannel service.

Which makes perfect sense. We’re talking about people’s money, here. With financial wellbeing on the line, it’s no wonder that the contact center is no longer just a cost center. Leaders in banking and financial services now recognize the contact center for what it can truly be: a place to communicate with bank customers and build relationships.

And a potential profit center, as a result.

Unfortunately, only 27% think the financial services industry is fully customer-centric. Perhaps that’s why Salesforce report found that building and improving customer trust is the #2 business priority for the next two years (right behind implementing new technology).

As to why, findings from our own Service Relevance Report 2022 offer some insight: in a survey of 4,000 consumers in the U.S. and UK, 96% of respondents said they might never buy from a company again after just one negative customer service experience.

So, which trends will shape the future of the contact center in this space?

Traditionally, a contact center team is responsible for the costs it creates (or helps avoid). Money goes into contact center staff, training, and technology to provide service and support to customers.

In banking and financial services, in particular, there’s tremendous opportunity to generate revenue in the contact center — to leverage relationship building, customer insights, and more to recommend new or adjacent products to customers.

To elevate the contact center experience, banking and financial services companies will need to address three persistent challenges.

These comprise the top three contact center trends that we’re keeping an eye on. You’ll also see what we’re doing to help banks address these issues.

Trend #1: Optimizing Their Digital Channel Strategy

It’s 2022 and contact centers still struggle with high call volumes for what should be low-touch issues. A bank’s customer should be able to find their account number without calling, for example. Pulling up one’s statements on the mobile app should be a simple process.

And yet, customers call in everyday, over and over, about the very same issues.

This logjam in the contact center’s first tier has a cascading effect that degrades the customer experience and increases contact center costs. Customers are frustrated that they can’t figure out the simplest things. They’re opening costly, time-consuming tickets for the same issues when the corresponding self-service content already exists. Many of those customers go through expensive (read: overqualified) support staff to get the information that should be easy to find on their own.

How Coveo Helps Alleviate Disjointed Customer Experience

Our research shows that 40% of Generation Z alone say they will abandon a brand if they can’t resolve an issue on their own. The problem is, this happens a lot: after reviewing 8,000 customer journeys, Gartner found that though 70% of the customers were using self-service channels, only 9% could completely resolve their issue through self-service.

It’s time to help customers help themselves. Moving forward, banks and financial institutions will need to find a way to reduce call volume by offering a better digital self-service experience. The idea is to give customers a frictionless self-service experience through their channel of choice.

What’s frictionless? Not having to dig for answers they didn’t know existed in the first place. It’s almost like the answer comes to them.

That’s how Coveo does it, leveraging AI that automatically considers a customer’s context (such as location, products owned, and website activity) to propose the most relevant content to resolve their issues, proactively and in the moment of need. That could be the contact form, knowledge base search, or even Salesforce community.

Trend #2: Personalizing Every User Experience

It’s difficult to personalize customer experience without the right data. That’s plain enough. But bringing together the right data from multiple sources, at scale, remains a persistent challenge for companies in banking and financial services.

To meet customer expectations for personalized experiences — especially in the digital banking space — banks will need better visibility into customer journeys to provide a better contact center experience.

What does this look like in practice? Visibility into when the customer last called in, for one, or visited the bank’s support site. What was their last interaction with the bank and what did they do? More importantly, do you have a comprehensive blend of this data from all of your channels?

All of this information should be brought together, analyzed, and used to extract the right insights for the contact center.

How Coveo Helps Unify Customer Journey Insights

For many banks, the problem isn’t collecting the data; it’s leveraging the data they already have. There is a solution. For instance, Coveo partnered with the best for analytics (Tableau dashboarding and Snowflake data warehousing, specifically) to offer robust omnichannel analytics for one of our customers.

Today, that customer enjoys a 360-degree view of the customer, which makes it far easier to measure success and optimize the content experience across channels.

What’s more, they can:

• View real time information with little to no manual intervention
• Increase engagement and success across support channels
• Extend timely, relevant content to key touchpoints in the customer journey

Trend #3: Improving the Experience Retain and Engage Distributed Workforces for Quality Customer Service Interactions

If banks want any chance of meeting consumer expectations, they’ll have to take care of their contact center employees better.

Call center employees (customer service reps, or CSRs) require a significant level of training, regardless of experience. Due to physically distributed teams, which is much more prevalent now, training can be difficult and expensive.

Burnout and turnover remain the bane of many a contact center professional.

What’s more, CSRs have to work within multiple tools and interfaces. They often struggle to find the information they need to resolve customer issues, or to make appropriate cross-sell recommendations. It’s not that banks don’t train their CSRs — it’s that they don’t do it effectively and efficiently.

Consider this research from Finastra:

• Globally, companies spend $350B on ineffective training
• Employees lose 80% of skills within a week if they don’t use them post-training
• It takes an estimated 125 hours to create just one hour of compelling training content

We need to find a way to train and support our CSRs a lot better.

How Coveo Supports Employee Training and Retention

Just like customers need personalized content that comes to them, so do CSRs. The trick is to not only integrate enterprise-wide knowledge and training info into the flow of contact center work, but to personalize those content experiences.

With features like insight panels and user actions, Coveo can surface the most relevant and up-to-date information an agent needs for resolving client issues based on a variety of data points. Without switching windows, CSRs will see personalized recommendations based on their role, as well any available customer data (behavior, history, etc.).

Training content can be similarly personalized for CSRs. Here’s what that might look like:

• Layer AI-powered search and information discovery into your workplace apps, intranet, and other important systems
• Centralize and personalize access to knowledge resources from various repositories and for each department and role
• Surface relevant search results and proactive recommendations for new learning
• Deliver actionable knowledge to wherever employees are training

As the aforementioned report from Finastra puts it, “The goal is to make learning and development available anytime, anywhere, via any device.”

Let’s recap some of the more salient takeaways in this blog post.

First and foremost, the contact center is much more than a cost center. Moving forward, growth-oriented companies will realize and act on the huge potential to drive profits and growth through the contact center. They’ll do so by making the contact center part of their digital transformation — by building customer relationships within that channel and, by extension, upselling and cross selling.

To achieve those ends, contact centers will need to address three core trends:

Finally, AI-powered search promises to play an integral role in the contact center. According to Salesforce, almost 90% of financial leaders think autonomous finance will be a “strong competitive advantage.”

I submit that AI-powered search will play a central role in making autonomous finance the best it can be.

Dig Deeper

AI-powered search can extend beyond the contact center; learn more about connecting digital experiences in our ebook, Blueprint for Personalized Digital Experiences in Banking and Wealth Management.

Top 3 Banking Contact Center Trends for 2022 was originally published in Coveo Behind the Click on Medium, where people are continuing the conversation by highlighting and responding to this story.

One way or another, businesses are coming around to the necessity of digital transformation. They’re attempting overhauls of the customer experience that can be comprehensive, compendious, and remarkably compressed.

Yet the aim of these efforts must be consistent and crystal clear.

“The customer experience needs to be omnichannel, and brands need to be where their customers are, when their customers need them, while taking time to understand those customers along the way.” That’s how Annette Franz, Founder and CEO of CX Journey Inc. and a Certified Customer Experience Professional (CCXP), characterized the goal to be attained for digital transformation initiatives.

Franz was speaking on a recent Coveo webinar, Critical Components Of Outstanding Customer Experience. She was joined by Debbie Shutto, Director of Knowledge Management at Finastra. Together, Franz and Shutto prioritized the many moving parts of customer experience, including:

• What exceptional customer experience actually is.
• Where customer support and customer service fit in.
• The requisite components of strong customer experience.
• How knowledge management (KM) and artificial intelligence (AI) fit in.
• What customer experience might look like in the coming year.

First, What Customer Experience is Not

Though we’re all highly invested in the customer service aspect of customer experience, let’s take the blinders off for a moment.

Customer service does not equal customer experience.

Instead, service and support comprise one of many interactions under the customer experience umbrella. And there are many, including:

• Product research
• Purchase process
• Training and onboarding
• Post-purchase customer service and support
• Renewals, cancels, and returns

“In the end, service is what happens when the experience breaks down,” says Franz. “If we get the experience right, then we take the weight off the contact center. If we get the experience wrong, customers will be calling customer service.”

As for a working definition of customer experience, Franz sums it up thusly: “customer experience is the sum of all interactions that a customer has with a brand over the life of their relationship with their brand. More importantly, the feelings, emotions, and perceptions they have about those interactions.”

So, what goes into making that definition of great customer experience a reality?

How can expertise and community become value adds that help retailers compete? Tune into our podcast, The Ecom Edge!

6 Pivotal Components of Exceptional Customer Experience

Even a cursory survey of the customer experience landscape will reveal a fundamental thread that weaves its way throughout: effort. How easily, in a moment of need, can customers (and the employees that serve them) get what they need? And what happens when the effort required is too much?

Across all of the six areas of customer experience that Franz and Shutto highlight, reducing customer effort is a primary objective.

1. The Self-Service Component

The amount of effort required to self-serve — to find answers on one’s own — is directly tied to a customer’s perception of their overall experience. Just the other day I tried to troubleshoot an issue with my cable provider (yes, I still have one of those). Their knowledge base was impossible. The chatbot was an empty-headed robot. In the end, I had to call customer service.

And I’m still this close to switching providers for good.

The stakes are high, as are customers’ expectations around great customer service. Be there or be square, as the saying goes. With respect to self-service, “there” can mean a lot of things:

Debbie Shutto of Finastra recalls how the company’s three siloed knowledge sources made it so difficult for customers to find answers.

“It was terrible,” she recalls.

So she and her team built the Finastra Customer Success Community, a go-to support portal in which one search returns unified results from all of those knowledge sources combined.

2. Knowledge Management

Ah, knowledge management. Old reliable. At least it used to be.

Now knowledge management is back in the spotlight, thanks in part to the needs of hybrid working environments. Increasingly, enterprises are figuring out that, without mature knowledge management programs, other parts of their customer experience will suffer.

Shutto puts it plainly: “If we have the info, our customers and employees ought to be able to find it.”

Indeed, the self-service knowledge base is the #1 channel for Finastra. It helps customers self-serve. It extends content into Finastra’s new customer success community. It even ties into customer service agent and employee experience.

While plenty goes into the knowledge management discipline, Finastra in particular has invested in a culture of creating valuable — and reusable — content. To this end, they’ve implemented Knowledge-Centered Service (KCS) best practices.

“KCS is all about structuring content so it’s easy to follow. We make sure our content adheres to that structure to make sure our clients can solve problems on their own.”

3. The Customer Service Agent/Employee Experience

In our zealous devotion to customer-first experiences, let’s not forget the people making it happen behind the scenes. The better equipped agents and internal employees are to access knowledge, self-serve, and collaborate, the better they’ll serve customers.

“Our customers are able to find what they need quickly and get back to using their Finastra products,” says Shutto. And when cases are submitted, agents get the same experience in Service Cloud.” That means a unified search experience for support engineers. It also means pre-filtered search results and recommendations based on case subject and product.

“For non-support staff,” continues Shutto, “we have a Coveo enhanced search tab so that success managers, account execs, and product managers have the same search experience when they’re working with customers and clients on their own day to day.”

4. AI and Personalization

“AI can really help to facilitate the customer experience. It makes it more personalized, which is what customers want and expect,” says Franz.

At its best, AI and machine learning can personalize content, recommendations, and journeys on a one-to-one basis. It can automatically bring together data sources, map the most useful ways forward, and serve those up to customers.

It’s not easy, of course. Take ecommerce. To properly personalize ecommerce experiences using AI requires a lot of different data:

• Personal
• Contact
• Transactional
• Onsite behavioral
• Offsite behavioral

To name a few. The same goes for exceptional customer service and support. Yet both Franz and Shutto remain convinced about the continued role that AI-powered personalization will play in customer experience.

5. Analytics

Every customer interaction creates data points, whether you pay attention to them or not. Full stop. Within the context of customer experience and, more specifically, service and support, analytics can reveal a lot about where customer experience is now, versus where it needs to go.

Debbie Shutto’s team at Finsatra uses analytics to identify gaps in their documentation. They look at trending topics, opportunities for consolidation, or ways to improve relevance.

“We’re able to look at search queries that return no results and fill those gaps, and publish a knowledge base article that same day,” says Shutto. If her teams see a high-volume search term with no search results, for example, they flag it and create content for it straightaway.

We’ve seen customer experience pros like Shutto use analytics to fix major kinks in the customer journey, too — including improvements to the product itself.

“We use a customer self-service metric within our Coveo insights, which allows us to measure how well we are doing from a customer self-serve perspective,” says Shutto. These metrics directly inform how and where Shutto’s team directs its efforts.

6. Feedback

Though it might not be as flashy as AI, customer feedback remains a lifeline of customer experience. Both Franze and Shutto agree that customer-experience-focused organizations must take the time to not only listen to customer feedback, but use that information to design better experiences.

“When feedback comes in, we’re acting on it immediately,” says Shutto. “If it’s a thank you, we send an email to the client to thank them. If it’s a problem, we’ll follow up with a subject matter expert and get the customer an answer or connect them to a support engineer. Then we circle back around and make sure we’ve updated that article and republished it.”

Even exceptional customer experience is a house of cards if most customers feel like they’re speaking into the void. The solution? Build mechanisms to not only collect customer feedback, but to close the feedback loop and respond.

Finally, Where is Customer Experience Headed?

One thing we know is that things are changing fast.

“We have to listen to customers to stay ahead of evolving trends and expectations,” says Franz. “AI and automation are not going away. Personalization and hyper personalization are here to stay. More than anything, though, it’s people. People are in the driver’s seat. If brands don’t have the foundation right, they’ll struggle with these trends.”

Both Franz and Shutto note a few areas to keep an eye on with regard to the future of customer experience:

• Predictive support: help customers get out ahead of issues long before they become issues
• In-product contextual support: bring personalized help content into the products themselves
• Agility: the ability to quickly spin up needed features or experiences that customers are demanding

All of these are compelling angles to consider. They all depend on getting the experience right, too. And that experience needs to be built around the customer.

“We need to put the customer at the center of a digital experience,” says Franz. “If they’re not, who are we designing this experience for?”

Check out the whole conversation on-demand.

Dig Deeper

Want actionable insights for designing your self-service customer experience? Check out our 6 Tips for Tailoring Your Customer Support Experience for Self-Service.

Looking for the latest customer service trends? Check out our most recent research, the 2022 Service Relevance Report.

The Makings of Exceptional Customer Experience in 2022 was originally published in Coveo Behind the Click on Medium, where people are continuing the conversation by highlighting and responding to this story.

How do you provide great ecommerce personalization?

When it comes to offering a personalized experience, most would say you really need to know a person (or demographic, to a point — they’re not … entirely … monolithic) on a certain level. This is done much more easily in person, where body language, gestures, and direct interactions with products are observable.

But now that the retail landscape has undertaken a substantial transformation, with ecommerce sales soaring to $4.9 trillion worldwide, the ability to offer a personalized experience online is more important than ever. After all, prior to ecommerce, consumers had to get back in their car to go to a different store. These days, your competition is just a browser tab away.

This brings us to the idea of ecommerce personalization. What is it, and how do you provide it? It’s not as simple as putting a name in an email (or an email subject line). Let’s dig into the what, why, and, perhaps most importantly, the how.

What Does Ecommerce Personalization Mean?

Ecommerce personalization is an essential practice for businesses, and you’ll be hard-pressed to find an online store without it.

Essentially, personalization determines relevant offers and content to present to users based on their data — demographics, search or browsing history, past purchases, website engagement, and more. That’s why every website you land on now asks if it can use “cookies” — it just means you are allowing the site to use your data.

A personalized ecommerce shopping experience might look like seeing a “suggested for you” section while you’re browsing an online store, or it could be an email with the subject line “gadgets we think you’ll like.” Ecommerce personalization reaches customers through multiple channels, and there are countless ways it can be implemented.

Today, it’s almost impossible to survive as an online store without offering ecommerce personalization because it’s what consumers not only want, but have come to expect. In fact, 91% of consumers report that they’re more likely to shop with brands that offer personalized, relevant product recommendations.

It’s clearly an important feature but why?

Why Is Ecommerce Personalization Important?

In this digital age, ecommerce websites can easily lack one essential part of the customer experience: personalization. Businesses previously built connections with their customers through face-to-face interactions via their sales team, customer service reps, and cashiers.

The pandemic created new online shopping habits, and with less people visiting stores in person, it’s more important than ever to give shoppers a satisfying online experience.

In-person customer experience relies on actual, human personalization to build strong customer connections. Employees can recommend items, get to know loyal customers and inform them of upcoming offers. In ecommerce, personalization relies on artificial intelligence (AI) to create that same unique, curated customer experience in an online environment.

Customer data can help determine the most relevant offers and personalized content for specific visitors in a way that sales reps can’t keep up with in real-time. While there is a trade-off between real in-person interactions and digital ones, AI-powered personalization has become so effective that it can massively impact a business’s customer loyalty, sales, and competitive positioning.

One important item of note, however: the ever-shifting data privacy landscape. While shoppers do want their experience to be personalized, they don’t want that personalization to come at the cost of their data being misused.

This has evolved into the death of the third-party cookie, but that doesn’t mean you can’t still offer great personalization. Our own research found that 51% of survey respondents would share personal data with a brand they trust. We dig into how to build that trust in our blog, Future-Proof Your Personalization with First- and Zero-Party Data.

Let’s explore the benefits of building meaningful connections through ecommerce personalization.

3 Benefits of Ecommerce Personalization

1. Increase Brand Loyalty

More than half of consumers say they feel more loyal to a brand that “gets them.” When customers feel understood by a brand and receive custom communication, a good product recommendation, or a great offer, they’re more likely to return for future purchases. In fact, brand loyalty, specifically with millennials, increases by 28% if the customer experiences personalized communication.

Good ecommerce personalization boils down to good customer service. It’s meeting your customer’s needs before they know they have them. Feeling taken care of goes a long way in creating loyal customers.

Great personalization can come in the form of AI-powered search. By combining user intent via behavioral data with the “wisdom of the crowd,” you can personalize even for what we term “cold start-users.” (And did you know that 70% of new traffic on ecommerce sites is from cold start users? But not all of those users are really “new”!)

Elements like query suggestions and dynamic facets, powered by machine learning, create a personalized user experience, at scale. Fifty-six percent of shoppers are more likely to return to a website offering recommended products.

2. Boost Sales

This is probably the most obvious benefit ecommerce personalization has to offer.

Personalization helps grab the customer’s attention and reduces the amount of browsing time and energy a customer expends while browsing ecommerce sites. Presenting a customer with a product recommendation or offer they’re likely to prefer helps increase the chance of actually increasing conversions.

Forbes collected and published 50 different stats on the power of personalization. Here are just a few about its impacts on sales:

• Companies using advanced personalization report a $20 return for every $1 spent. — Clickz
• Marketers report that personalization efforts can boost revenues by up to 15%. — Adweek
• 95% of companies that saw 3x ROI from their personalization efforts increased profitability in the year after their personalization efforts. — Monetate

3. Keep Up With The Competition

Now that ecommerce personalization is so popular, ecommerce sites that don’t have them can easily fall behind their competitors. Our own research shows that 93% of consumers expect their online shopping experience to be at least equal to if not better than the in-store experience. Stumbling upon sites with poor or no personalization can be extremely frustrating for users.

80% of frequent shoppers claim to only shop with brands that offer personalization, while 66% of customers report that coming across content lacking personalization would actually discourage them from completing their purchase.

Without relevant personalization, it’s almost certain that customers will visit your competitor. (And although our research found that customers use Amazon as a primary shopping source, 46% would choose to shop elsewhere because a product isn’t available, they prefer other sites, or they didn’t trust reviews — which means there’s an opportunity for your ecommerce site, if geared with the right personalization software.)

Ecommerce personalization is necessary for a thriving online business today, but what are the best ways to use them to harness and make the most of customer data? There are opportunities for this across almost every single part of the customer journey.

10 Ecommerce Personalization Examples To Enhance The Customer Experience

1. Offer a Special Offer to First-time Visitors

First impressions are everything! At their very first stop on the customer journey, you can entice customers with 10% off, buy-one-get-one, or with a special seasonal deal catered to them. This is a great way to introduce customers to your brand and make it easier for them to say “yes” to an initial purchase.

Example: Glossier offers a 10% off deal and free shipping on $30 or more for the customer’s first order

How do the brands Glossier and Pietra create a bond between brand and customer? Tune in to our podcast, The Ecom Edge!

2. Offer Recommended Searches to Narrow Options

For big retailers with different departments and offerings, visitors can get overwhelmed by not knowing where to start — especially if it’s their first time on the site.

Search recommendations can help narrow the customer’s choices by suggesting searches based on the user’s data combined with other factors like past first-time customer data.

Example: Lowe’s offers recommended searches to give the visitor a starting point

3. Suggest Popular Products

Placing popular items front and center not only helps direct customers when they arrive on your site, but it also acts as a form of social proofing to remove uncertainty around a purchasing decision based on other people’s activity.

If a customer sees that a certain type of laptop is more popular than another, it will likely influence their decision and simplify their shopping experience. This is especially helpful for first-time customers who are likely comparing products from multiple sites. “Popular” items presented can be tailored according to behavioral data from a cold-start user or an authenticated customer’s habits or recent searches. This keeps these elements relevant to the category they are shopping in.

Example: Best Buy promotes their current most popular items as product recommendations

4. Show Items Your Customer Will Like

As you start collecting more data from frequent shoppers, machine learning helps suggest relevant items that match their tastes based on past searches, clicks, and other website interactions. This type of personalization can make a customer’s user experience more enjoyable and easy, while making them feel seen by your brand.

Example: DSW provides product recommendations that the user is likely looking for

Blog: Need ideas on the best spots to put product recommendations? We’ve got you covered

5. Help Them Complete the Collection or Look

This type of personalization acts as a friendly salesperson by suggesting what goes well with the product a customer is considering. It helps the customer see a fuller picture of the product, how it might play out in their life, and how other products could complete their purchase.

Example: Hobbs, a Coveo customer, provides styling suggestions on a trench coat product page

Blog: Learn how machine learning can power suggestive selling

6. Suggest Complementary Products

Once the customer has landed on a product page or added something to their cart, complementary products are a great personalization strategy to upsell. It’s similar to suggesting items that go well with the product the customer is considering or purchasing, but rather than focusing on completing a look or a collection, complementary products can help make the experience of the current product better.

For example, someone who is purchasing shoes may need socks, or someone who purchased a table may need floor protectors to ensure they don’t scratch their hard-wood floors. This type of personalization focuses on the customer’s next step and moving them toward it based on their current activity and past purchasing data.

Example: Amazon suggests a screen protector to go with a phone case

7. Personalize Exit Offers

If a visitor clicks the exit or back button without purchasing anything, exit offers can act as a last-ditch effort to make a sale. These can sometimes come across as desperate, so framing them positively and keeping them customer-centric can help make them more effective.

Example: Express Gifts used Coveo-Qubit to detect exit behavior and persuade shoppers to complete a purchase on their site.

8. Entice Lost Buyers

Most of us have seen a cart recovery email, but the key to making them work is making it easy for customers to pick back up where they left off shopping. A “return to cart” or “continue checkout” button minimizes the customer’s efforts to get right back to it and is more actionable than simply saying “keep shopping.”

Example: Crocus, a Coveo customer, makes it easy to go directly back to the checkout cart where the customer left items waiting.

Blog: Get 9 more cart abandonment recovery tips

9. Email Customers With Relevant Suggestions and Deals

Sending suggestions or deals directly to a customer’s inbox can pique their interest and remind them that you’re interested in what they like. Whether it’s new items they might like or a deal especially for “insiders,” sending them something just for them can make them feel special.

Example: Banana Republic sends customers items curated especially for them

10. Bring Customers Back With a Special Deal

This personalization tactic is based on a lack of interaction with a site and can help bring back customers that have spent time away from a certain brand. This is often a profitable effort for companies because acquiring new customers can be more expensive than simply retaining old ones. The come-back personalized email can help revive an old customer and invoke their loyalty.

Example: Sephora sends Insiders who haven’t purchased in a whole a special 10% discount

These 10 examples highlight the importance of ecommerce personalization in your online shopping experience. Employing these examples in your own ecommerce efforts can help move the needle toward a profitable business — and away from a website that no one visits.

If you’re struggling to achieve successful personalization on your site, check out our Ultimate Guide to Personalization in Ecommerce for insights and tips on how to change that.

Ebook: The Ultimate Guide to Personalization in Ecommerce

DIG DEEPER

Want to learn even more about personalization? Check out our blog post One to One Digital Personalization: Fact or Fiction?.

And while you’re at it, read our blog Powerful Personalization in Ecommerce — No Big Data Required to learn how to personalize your website while with limited resources.

What is the №1 thing retailers need to do to stay relevant in today’s market? Tune in to Episode 4: Is Physical Retail Dead? of our podcast, The Ecom Edge!

Take your ecommerce website to the next level with Coveo’s Content Personalization Engine.

Ecommerce Personalization: More Meaningful Connection in a Digital World (+10 Examples) was originally published in Coveo Behind the Click on Medium, where people are continuing the conversation by highlighting and responding to this story.

If you have been researching Natural Language Search or Natural Language Processing (NLP), you may have heard of word2vec. If not, we have you covered.

Word2Vec is a machine learning technique that has been around since 2013, courtesy of Tomas Mikolov and his data science team at Google. It’s used to train a computer to learn about your language (vocabulary, expressions, etc.) using a corpus (content library). If you aren’t a data scientist, this article digs into what you need to know to be conversant with the data scientists in your life.

How Words Are Encoded in Your Brain

Word embeddings are the vectors that are created to represent words — and their context — or the relationships between words. Let’s take a step back first to consider how we learn as individuals.

Ever have someone come up to you with a question — and you have no idea what they are talking about? Chances are they have not given you enough context to understand what they really need. Maybe they are using different words than what you are used to, or maybe they are using words that have different meanings in a different time and place.

We can try and decode what a person wants by asking questions until we can finally winnow down what they want.

Understanding natural language is not a trivial matter. And as humans, we don’t realize how incredibly complicated language is, which is why it is hard to teach computers NLP.

Structural Relationships Between Words / Context

Words rarely have meaning in isolation. If I say “leopard print” — it might be referring to the indentation of the animal’s paw on a given surface, or maybe it’s a description of a pattern that replicates a leopard’s skin. Then again, it might be a poster of the magnificent beast itself.

Without any context, it’s hard to know what it means. That’s where word embeddings come in. They show the particular relationship certain words have in relation to others in the same phrase or sentence.

When paired with clothing or accessories — we understand that “leopard print” is describing colors and markings. When paired with sizes like 24 x 36, we understand that it is a poster, and when we use it with words like sand or mud, we know that we are tracking the animal.

When paired with clothing or accessories — we understand that “leopard print” is describing colors and markings. When paired with sizes like 24 x 36, we understand that it is a poster, and when we use it with words like sand or mud, we know that we are tracking the animal.

Semantic Relationships Between Words

Your brain also understands semantic relationships between words.

For instance, it understands that word pairs like “king” and “queen”, “blue” and “yellow”, “running” and “jogging” each have a special relationship. A word pair of “king” and “running,” and “queen” with “yellow” — don’t! Our brain also understands that the semantic meaning between “shoes” and “socks” is different than, say, the relationship between “shoes” and “sandals,” or “shoes” and “running.”

Your brain knows that the word “queen” has a certain relationship with the word “England” that it doesn’t have with the word “California,” (unless we are talking about Gen Zs in California talking to each other). The word “toast” has a relationship with the word “French” that it doesn’t have with the word “Spanish.”

Now imagine the area of your brain that contains language as a well-organized storage space. Words like “French” and “toast” would be located closer together in that space than the words “toast” and “Spanish.”

Target Words, Word Vectors, Vector Representation

Finally, every word evokes a set of associations that are partially shared by all speakers of the language and partially a result of your personal experiences/geography/etc.

For example, your associations with the word “milk” might be white, 1%, breakfast, cereal, cow, almond, and if you have experience with dairy allergies, danger, ambulance, anaphylaxis.

This ability to draw these associations is the result of complex neurological computations honed by our brains over thousands of years of evolution.

Word Embeddings

Now that you have an idea of the complexity of the human language, you know that it’s not enough to just feed computers the dictionary definition of words and hope for the best.

Yet, NLP had to start somewhere, and this is where it started. Since computers understand only numbers, in order to teach them natural language, we have to represent words numerically. For a long time, it was done by representing each target word as a string of zeros with 1 in a different position for each word. This method is called “one-hot vector encoding.”

So if you have a vocabulary of four words, they would be represented like this:

This computational linguistics method creates a unique representation for each target word and therefore helps the system easily distinguish “dogs” from “cats” and “play” from “Rome.” But there are two problems with it.

First — you guessed it, this method has no way of encoding the relationship between words that as humans we take for granted. It has no way of knowing that the word pair DOG and CAT are similar — in a way that DOG and ROME are not. That CAT and PLAY have a special relationship — that CAT and ROME do not.

Second, this vector representation is problematic if you have a very large corpus. What if your vocabulary size is 10,000 words instead of four? This would require very long vectors, with each vector consisting of a long string of zeros. Many machine learning models don’t work well with this type of data — as it would make training the data really hard.

Today, both these problems are solved with the help of a modern NLP technique called word embeddings. Word embeddings are the process of converting words into dense vectors — or a series of smaller vectors. This process assigns values to each word along several different dimensions, creating a dense vector that isn’t just a string of 0s but actual coordinates in abstract space.

[For geeks out there, sparse vectors are when you have a lot of values in the vector that are zero, such as in the one-hot-vector method above. A dense vector is when most of the values in the vector are non-zero.]

As an example, lets see how this method would encode the meaning of the following five words:

Each word can be assigned a value between 0 and 1 along several different dimensions, for example, “animal”, “fluffiness”, “dangerous”, and “spooky”

Each semantic feature is a single dimension in an abstract multidimensional semantic space and is represented as its own axis. It is possible to create word vectors using anywhere from 50 to 500 dimensions (or any other number of dimensions really…).

Each word is then given specific coordinates within that space based on its specific values on the features in question. The good news is, this is not a manual job. The computer assigns these coordinates based on how often it “sees” the co-occurrences of words.

For example, the words “cat” and “aardvark” are very close on the “animal” axis, but are far from each other on the scale of fluffiness, and the words “cat” and “duvet” are similar on the scale of fluffiness but not on any other scale.

Word embedding algorithms excel at encoding a variety of semantic relationships between words. Synonyms (words that have a similar meaning) will be located very close to each other.

[The counterpart is that often antonyms are also very close in that same space. That’s how Word2vec works: words that appear in the same context — and antonyms usually do) are mapped in the same area of space.]

Other semantic relationships between words, for example, hyponymy (a subtype relationship, e.g. “spoon” is a hyponym of “cutlery”) will also be encoded.

This method also helps establish the relationship between specific target words, for example

• A leopard print was seen in the mud.
• That’s a beautiful leopard print on your coat.
• I bought a leopard print with a black frame.

The system will encode that the target word “leopard print” appears within sentences that have the words “mud,” “coat” and “frames.” (This is a called a word window and older models like Word2vec have a small window — with the targets being within 3–5 words.) By looking at what is before and after a target word, the computer is able to learn additional information about each word and locate it as precisely as possible in abstract vector space.

So to summarize, under this method, words are analyzed to see how often they appear near each other (co-occurrence). Thus, word embedding algorithms are capable of capturing the context of a word in a document, semantic and syntactic similarity, relation with other words, and so on.

As you can see below, by assigning similar words and similar vector representations, this method was able to encode the meaning of words and cluster them according to categories, such as cities, food, countries, and so on.

Over the next couple of months, we will delve into more machine learning models that come into play with deep learning and NLP.

Originally published at https://www.coveo.com on May 24, 2022.

Word2Vec, or How Computers Learned to Talk Like We Do! was originally published in Coveo Behind the Click on Medium, where people are continuing the conversation by highlighting and responding to this story.

Your customers are used to immediate service. They order their favorite pizza with just a tap in an app, and binge their favorite shows their living rooms on their streaming service of choice. (Sounds like a great evening to us, don’t get us wrong!)

Those same customers expect the service they receive from your agents to be just as easy and seamless. Ninety-six percent say a negative customer experience would affect their decision to purchase from that company in the future.

IVR systems, self-service channels, and chatbots are part of the equation to delivering service that gets things right — fast. But those tools cut both ways… because by the time a customer lands on the line with one of your agents, they’ve already tried every other option to no avail. They’re frustrated, and it’s up to your agents to sort things out — again, fast.

And those agents are likely feeling frustrated, too. Their whole day is filled with more challenging cases since the easy stuff is already covered. Implementing stronger self-service options can take off some of the load, but you still need to grapple with the fact that the calls that do come in will always be the most difficult ones (with the crankiest customers) to handle.

With that in mind, the question becomes how do you best support them: with the traditional tiered support model, or by introducing case swarming?

Tiered Support: Playing Ping Pong

In a tiered support model, cases are moved from generalist agents to specialist agents in increasing steps.

It works like this: customers calling for help are ushered to a tier 1 agent who handles most of the known, most common issues. If tier 1 can’t resolve the problem, the case gets escalated to the next level of specialist who has more expertise to offer.

Step by step, customers rise through the ranks of knowledge and skill until they get the answer they need. In most organizations, the levels cap at three or four, though five-tier structures aren’t unheard of.

Pre-self-service and pre-on-demand everything, it worked. Generally speaking, most questions were basic ones that the initial agent could answer, and the upper-tier agents were fed only the new or complex cases. The majority of customers left satisfied after one call, with only a handful needing a bump up the ladder.

And Then Everything Changed

With self-service and other technology like IVR and chatbots deflecting many of those basic calls, tiered support falls apart.

Because the easy stuff’s been answered already, more customers need to move beyond tier 1 and get bounced up the chain. With all the shunting between tiers, customers end up frustrated with the process. Escalation is time-consuming, and it requires customers to repeat their story over and over again, with each new agent they reach.

Add to that, some of those customers would rather not be speaking with a customer support agent to begin with. While talking directly with a human being is a preferred method for some customers (52% of baby boomers say they’ll drop a brand if they can’t speak to a person to solve an issue), 40% of Gen Z say they will abandon a brand if they can’t resolve an issue on their own.

Your support staff feels the pain, too. Tier 1 agents do little more than ask the same repetitive questions and forward calls up the chain — without ever learning the outcome or how to resolve them themselves. Higher tiers end up feeling overwhelmed with higher volumes than expected. Agents in lower tiers don’t get much opportunity to develop their skills or learn new ones via training.

That’s a major blow when it comes to employee satisfaction… because when employee satisfaction falls, so too does customer satisfaction.

Case Swarming: Playing Catch

Case swarming, or an Intelligent Swarming ℠ model, disrupts this three-plus decade legacy of customer service practice — with good reason.

Pared down to its core, an Intelligent Swarming model (or case swarming) is a tier-less service structure where customer support tickets are pooled and agents pick the ones they’re best suited to handle. Agents who opt in then “own “ the case until resolution, using two options:

• Using their own skills or expertise to find an answer, or
• Collaborating with experts across your organization when they need assistance.

Go even further, and you’ll find service organizations using intelligent, AI-powered matching to pair individual cases with a swarm of agents who are most likely to have the interest, skills, or knowledge necessary to solve them.

It’s a perfect approach for those increasingly complex cases because it considers the talents of your service team. It also enables collaboration with other agents and subject matter experts, across areas like finance, product, operations, sales, legal and more.

No longer are agents compartmentalized, where those at the “top “ of the hierarchy feel they’ve moved “beyond “ customer interactions. No longer is knowledge hoarded away in departmental silos, unavailable to other agents who want to grow in their roles or for future self-service content. Agents have more… well… agency over their roles. Case swarming lets them choose what challenges they’d like to tackle, what they’d like to learn more about, and how they’d like to grow within the organization.

Customers win, too, because they don’t bounce around between agents. With agents self-identifying which cases they take, customers interact with just one agent — and are more likely to get an answer the first time, without all the waiting, queueing, and bouncing.

It’s no wonder companies like Salesforce are adopting (and sticking) with case swarming over tiered support. The Technology and Services Industry Association (TSIA) reports that 30% of its members currently use case swarming, a number that’s grown more than 8% in the past year.

And It’s Gaining Ground

Fifteen years ago, customer service simply wasn’t backed by the kinds of tools we have available today. But with service innovation around advanced collaboration and AI, service is shifting quickly. Particularly when it comes to cracking that difficult nut: delivering quality service at scale, the way customers want it, without wasting their time.

Let’s illustrate that in the context of Knowledge-Centered Service, or KCS.

In a tiered support model, all that juicy knowledge about complex cases and never-been-seen-before issues gets locked away in silos. And in a silo, it may never be captured and used to improve your self-service content or internal knowledge database. If you don’t have a way to surface those solutions, they’ll never serve your agents in the future when the same issue pops up again.

Beefing up your self-service channels to deal with the increase in upper-tier cases isn’t in the cards, either. If you aren’t connecting all of your relevant data — including case information and resolutions — you’ll struggle to identify where your team members need a boost, much less provide content that’ll help customers solve their own issues.

On the other hand, within an Intelligent Swarming model, Slack and search make for a powerful duo that allows for the creation and maintenance of organizational knowledge. Here’s how they work together to make customer service more effective:

• Slack connects agents with expertise from around your organization. When agents need assistance with customer issues, they have immediate access to the people with the right knowledge. A quick back-and-forth later, and the agent has the answer they need.
• AI-powered search then indexes those conversations, so the next agent who looks up that issue on your knowledge database can access that conversation and get their answer right away. Not only that, but connecting information about resolved issues from Slack with your knowledge database can help you build stronger self-service content.

No black holes of knowledge. No frustrated customer support agents. And no (well, fewer) cranky customers.

Is There a Happy Medium with Case Swarming vs Tiered Support?

Short answer: “Yes, but…”

A hybrid support organization works like this: a traditional tier 1 agent takes the initial call, and if it’s a basic question, it gets answered right away.

If not, that agent gathers information and then routes the case through to a swarm and then acts as a middle person between the customer and the swarmers. There’s still friction, but it’s less than you’d get in a fully tiered system.

But that doesn’t solve the issues of satisfaction or speed. Customers are still left waiting for answers, and tier 1 agents are still limited to repetitive work with few professional development opportunities.

Basic case swarming alone isn’t the answer, either. Tossing every customer case willy-nilly into a generalized pool and expecting agents to pick and choose from the chaos? A rather noisy approach.

To be truly effective, your swarming model needs to be intelligent. That means it’s backed by technology like:

• Swarmer identification and intelligent case assignment that understands who your agents are, what the customer issue is, and funnels cases through to only the support agent whose profiles are best fit to answer
• Knowledge management that wraps around all of your most valuable troves of knowledge, including real-time conversations on platforms like Slack
• Unified search that surfaces only the most relevant information to your customers when they’re looking to self-serve — and to your agents when they’re checking their knowledge database

The Intelligent Swarming methodology, while certainly innovative and effective, isn’t a cure-all. It also isn’t perfect for every company and every situation. Making the move from a tiered support model to an intelligent swarm requires quite a shift in terms of company culture, workflow, processes, technology and more.

But when your customers can access the support they need with the same effort that it takes to choose a movie to watch, you’ll see the benefits first-hand.

*Intelligent Swarming℠ is a service mark of the Consortium for Service Innovation™.

Case Swarming vs Tiered Support: Which Does Your Company Need? was originally published in Coveo Behind the Click on Medium, where people are continuing the conversation by highlighting and responding to this story.