Implementing Self Service with Tableau part II

In the first post of this series, here, I looked at the definition of Self Service, myths surrounding it and pros and cons of implementing a self service approach at enterprise level.

In this post I’ll share what we’ve done at my work place. At CACI we provide a Retail Finance Benchmarking service to the majority of banks and building societies in the UK. This means providing Tableau Server external access to banks and building societies to analyse benchmarking data.

The end users will interact with the data but we have to be mindful of the various needs of our users, which meant asking one of the most important questions.


Granted those are two questions, well spotted. But these types of questions are fundamental if you want to deliver a service that’s inclusive and offers options to every single one of your users. What use is there to create the best dashboard in the world if it doesn’t suit your audience?

For us it became clear our end users could largely be split into 3 groups with different roles, thus different needs:


The key here is to understand the hurdle for our users, for instance there may be no benefit to provide a self service to executives as they are time pressed for answers and can potentially be less tech savvy than analysts, who usually have more time to dedicate to a single project. Managers are often bridging the gap between executives and analysts and are the ones more likely to span across categories.

It’s worth pointing out that users sometimes crossover categories and these are not meant to be restrictive, rather a a way for us to understand our audience in a high level sort of way. At your organisation, you may have to split your audience types in more or less segments. I’d caution against looking at it in very detailed levels or it may complicate rather than simplify the way you will approach content creation.

When the time came to design content we already knew who we were designing for, which made our jobs much easier. We also decided to take a similar approach in terms of content and mirrored the 3 layer structure you see above.


Let’s look at each of these in more detail:

Core workbooks are intended to replace a number of reports we had in various forms. This provides a smoother transition into Tableau and it means most of the quick questions are just a couple clicks away.


Tables are meant to be a comfort blanket, some of our users need to just get a couple numbers and export these to blend with their own data in-house, these tables remove the need for these queries to be handled by Excel, and as a Server Admin it means I can understand the needs of all our users by centralising their interactions.


I have written about these types of tables before and provided a detailed explanation on how you can create one. See here.

With regards to core workbooks and tables I always encourage our users to use what I perceive as being two overlooked features of Tableau server. Preferred view which I mentioned above, allows the user to select a couple filters and define that as their preferred view so next time they open the report they don’t have to apply the filters again. The second feature is favourites, which allow the user to mark a view or workbook as a favourite. This makes their user experience so much better by not having to scroll through a large number of workbooks to find the one the user opens every day. See here for more information.

The third option we offer our users is to interact with published data sources using the Web Editing function of Tableau Server.


I mention 9.3 above because that’s what we currently running, but Tableau 10 offers cross data source filters and dashboard creation in Web Edit mode which are two great additions to an already great way of allowing your users to the world of data discovery.

When we created the above structure we hoped to be able to cater for every user and hopefully see a healthy split between the self service offering of the tables and web edit, but also we knew quite a few users would be using the core workbooks to find those quick nuggets of information.

As of late I’ve fallen in love with the Postgres database that comes with Tableau server, on our old technology I couldn’t really find out who what using our product and when. Nowadays I can create a number of views that tell me exactly how successful our self service has been. I can then take this to my boss and have much more informed conversations about the way we move forward. Here’s what we are seeing in terms of usage:


I was rather pleased to see this split, which confirmed that an hybrid approach was the way to go, to unsure the needs of our users are looked after. It also showed that there’s no single answer when it comes to designing content. Ultimately it comes down to understanding your audience.

In the next instalment of this series, I’ll write about, how we are managing it and how we structure permissions within Tableau server. Watch out for a PostgreSQL post coming soon as well.

Until next time, thank you for reading.