We want our customers to be making best in class dashboards: dashboards that look great, effectively communicate their data by showing relationships and hierarchy and are easily understood by anyone who views them.
I reviewed around 50 customer dashboards to establish the main problems:
These were prioritised into importance of how many times they appeared:
Before taking the project any further, my Product Manager and I took several of the ‘worst’ dashboards customers had created and set out to make them better using the product as it was. We wanted to be absolutely certain that the problem was in our current solution, and not the education that we’re providing our customers. We wondered if actually we could not build anything new and instead perhaps educate or guide users to better solutions.
The result was that yes, we could make them a little better, but it was still really hard to see relationships and there was still a lot of repetition. Given that we, the ‘dashboard design professionals’ couldn’t make something that we considered great, we felt more confident that we needed to make bigger changes to the dashboards.
The ability to group metrics will help to show relationships in data, help with cognitive overload and is a start towards showing hierarchy.
This project was my first after joining Geckoboard, and whilst I had lots of ideas, I knew that senior stakeholders and longer standing colleagues had much more domain and technical knowledge than me, so to leverage this, I invited them to sketch out their ideas with me.
It was great to see how creative people were with their grouping:
I took the sketches and mocked them up into a rough, but idealistic end result using data from a real customer dashboard.
It was clear at this point that we saw real value in being able to group metrics. This has multiple benefits: clear relationships shown, less repetition of text in titles, and less cognitive load.
Once we were happy with the notion of grouping; how it would look and also the technical feasibility, we began protyping the UX to group metrics together.
In this phase, I was working with one frontend developer. We were making fast prototypes, testing and then making quick changes. As we were such a small team (one PM, one dev and me) we were testing a LOT but taking it in turns to lead the sessions. This was great as we all got to be involved, see the pain points and come to the same conclusions together.
At the end of the project, I gathered screenshots of the new dashboards and reviewed them using the same criteria as my initial review. This exercise also doubled up as a review of these dashboards to highlight the main bugs and issues. In true meta fashion, I made a dashboard of the results:
Next up will be iteration! Watch this space.