Self service analytics is a great way of empowering your business users and releaving the BI team's work load and ad-hoc requests. Modern BI tools give the business user an intuitive and easy to use analytics method, so they can make their own data insights without the need of SQL or other programming knowledge. When implemented correctly, it brings a huge amount of value to your business.
In this blog post I will share some of my best practises for starting with self service analytics.
There are many tools on the market, and selecting the right one is difficult. Gartner lists the leading tools, which is great for an overview, but each company is different. Different companies have different needs. Think about the amount of users you have, the amount of data, the needed quality, and the required easy of use.
Some tools integrate nicely with the current stack of tools. Some tools are better suited for sharing insights outside of the organization. Other tools are really expensive when the amount of users or data goes up. In short, there are many things to consider and selecting a tool will take time and effort. Hiring an independent consultant can be worth it, because they know the differences between tools and can discover your company's needs.
Running a pilot for a potential tool, or looking at other companies that are already using is can also help in selection.
After selecting and installing your tool, you can start transfering your old dashboards to the new environment. Most clients are using this migration period as a way of getting to know the new tool by rebuilding the existing dashboars 1 to 1 and trying to make them as identical as possible. This always makes me a little sad, because this migration should be opportunity for spring cleaning! So many legacy dashboards have grown over the years into becoming huge and unnecessarily complex things. Instead of rebuilding everything you should be asking:
So don't just copy paste your old dashboards, improve them! Tool mgiration is a rare opportunity to do so. It will also help with the adoption of the new tool, because employees will see the added value of the new tool immediately, instead of just seeing it as another thing they need to learn to use.
Your tool is there, your dashboards are there. But for self service, we need a few more steps. Even though these tools sell themselves as being intuitive and easy to use, it is wise to give some training to your end users. At least a 10 minute intro should be done. Not only will this help with the adoption of the tool (change is not always received well), also these tools often have new features (like personalized views and subscription options). If these features aren't shown, they often go unused, which is sad because they can add a lot of value and you've paid for them.
Training wise, you should think about the different users you have. Some users will only interact with the dashboards, others will also be building their own insights. To ensure everyone using the tools properly, you could make a short introduction video for the interactors, and for the builders you could create a training that is manditory before they receive a license.
But what about the data? Is it self service ready?
Your tool is there, your dashboards are there, your employees are trained.. But what about the data? You built the dashboards, so obviously your data is there, but is it self service ready? If you want your business users to make their own insights and dashboards, you might want to take an extra look at your data structures. You cannot expect your business users to join multiple tables to get the necessary data (joining sales data with returns data). Your business users will not be familiar with all of the field names that your BI team knows (3 type of Category fields, but only one is the current correct one). I have also been told by the team: If you use those tables, you have to exclude these records by filtering XYZ, otherwise you are counting doubles...
For self service analytics you need self service-proof data sets. Idealy you have seperate sets for different analytical purposes, that include all fields necessary and don't require additional joining. All naming should be descriptive, intuitive and consistent, so it is immidiately clear which fields are needed. Include frequently used calculations as fields, like growth % vs PY, because these are difficult for your end user to make themselves. Include frequently used filters as fields, like Current Year / week / day, so these can be easily filtered in and out. And, perhaps most importantly but often skipped; make a data dictionary. A page where your users can find descriptions and definitions of the data sets and where they can search for specific fields and which set to use.
Now you have started with self service! Just a few more things... Your users need to know where to go with questions, so make sure to set up support channels (a mailbox, slack channel, wiki page etc.). Communication channels are also needed to update your users, for example when a new data sets is released, or a new version of the tool brings more features. You can think about setting up a news letter or organizing quarterly events. Also providing sample dashboards or templates can help your self service analytics to become more effective. It gives your users some guidelines and a starting point.
But most importantly is to create a self service analytics community, where colleagues can learn from eachother, get inspired and improve their analysis. I will share more about Self Service Analytics Community Management in my next post.
In summary, self service analytics brings a lot of value to your company when implemented correctly:
Thanks for reading!
