MadCap Software LogoLeft Quotation MarkRight Quotation Mark

October 2019 New Releases Available Now   |   Learn More

Products Solutions Services Help & Support Customer Success Resources Free Trials Buy Now

dunnhumby wins UK TC Award by Modernizing User Assistance with MadCap Flare

In this article, Rachel Johnston of dunnhumby describes how the choice between DITA and MadCap Flare saved them around £100,000 per year, and how the team modernized their user assistance with Flare. This article was originally published in the ISTC Communicator, Winter 2018 issue.

dunnhumby is a data science company that provides tools and services to the retail industry. Our services enable retailers to understand and serve their customers better through strategic pricing, promotions, and ranging decisions.

Part of our strategy to broaden our market and create greater access to customer data science for consumer-facing businesses meant that we needed to re-evaluate our user assistance tools and processes.

We’d been using a very heavyweight DITA solution to solve the business problem of how we pushed extremely tailored content (tailored right down to the individual word choice) to users. But the cost of running and maintaining the CCMS (component content management system) meant we had to make compromises on the look, feel, and content model. Our output was functional but bland and, frankly, hard to love.

As our business model started to change and our analytics had been telling us that usage of the help was low, it gave us the opportunity, and the imperative, to change our user assistance into something more modern, user friendly, and easy to maintain.

Why Flare?

We decided to use MadCap Flare instead of DITA because it fits the shape of the team better. With only five authors, only two of whom were using DITA regularly, we weren’t getting the benefits of an enterprise CCMS – it was like using a double decker bus just to drive to the shops and back. That decision saved us around £100,000 per year that we could invest in staffing, design work, user research, migration, localisation, and video creation.

Redesigning the experience

When we began redesigning the overall experience, we wanted to make the user feel welcome, and clearly signpost the most commonly used parts of the help file. We started by looking at help portals for well-known companies, and looked at the showcase of examples on the MadCap website.

We also used analytics to identify high-traffic areas of the site. For our users, we found that the glossaries that explain our metrics and measures and how they are calculated, are particularly important – so we made sure we kept the glossary prominent in the navigation.

We also asked our UX (user experience) team for help with the interaction design. Based on its understanding of our information architectures, the team built us wireframes that we then passed to a CSS developer.

dunnHumby Online Help Screenshot

Taking it back to the task

It sounds counterintuitive, but we made a decision to reduce the amount of precisely tailored content. We were expending time, money, and energy on maintaining hundreds of variables for product terms (translated into a dozen languages) that didn’t particularly help the user achieve their goals. For example, we’d use variables to describe the same concept (say - a shopper) in whatever way the client asked us to (shopper, customer, householder).

We realised that having a coherent set of terms that we can teach the user, rather than accommodate individual users’ word choices, is an important part of our move to self-service software.

Video strategy

Short-form videos screenshot

In our user assistance, we often have to explain very complex mathematics and data science in a way that’s accessible to non-specialists. Short-form videos have been invaluable in explaining, for example, how our products use machine learning algorithms. We’ve also been using animated gifs to signpost new features in a way that’s immediate and attractive.

Hosting the help

We use GitLab, both as source control and also as a mechanism to push the content to the development team who deploy the help. We used in-house developers to create that process. Our previous solution required a complex setup of ADFS1 groups to give users access to the help but these groups were not always reliable, leaving users without any help. Now, while still being secure, we can authenticate the user more simply.

Adding the experience layer

As well as providing a comprehensive help file, we also wanted to push more help to the user at the point of need. Most mobile applications now have some form of user onboarding – overlays to the UI that help the user to orient themselves and gain confidence in using the application.

We evaluated a number of user onboarding tools before settling on Appcues as a mature, robust, and usable onboarding product.

With just a couple of lines of JavaScript embedded in the product, Appcues enables authors to create and publish attractive guided tours of the product. It uses the application’s CSS to identify where and when a piece of content should appear on the screen, and you can also use CSS to style the guided tour to match your application’s look and feel. With a little more development work, you can push specific content at specific times – for example, if a user hasn’t yet explored a new feature, you can encourage them to try it out.

Appcues also has built-in analytics that enable you to test how engaged users are with each flow. Completion statistics help us to evaluate if a tour is doing its job, or whether it needs to be shortened, reworded, or removed entirely. Appcues also enables us to build surveys into the product, enabling our development teams to begin getting fast, actionable feedback on new features.

Response to the new UA

Since winning the TCUK competition, we’ve rolled out our new help in larger projects and the response has been overwhelmingly positive.

It’s also raised the profile of technical communication within our company. Product teams are genuinely excited about this new way of offering help to the user.

We’re already seeing a substantial increase in use of the new portal compared with the previous platform.

We’ve also implemented Google Analytics and we’re looking forward to making more data-driven decisions about how we work.