8 most common mistakes when designing a dashboard
Unfortunately, there are so many so-called "beautiful" designs of dashboards circulating on the Internet, which ignore the basic rules, that one is quite surprised.
And later then, you are even more surprised, when your client brings you this "beautiful" horror, saying that this is how his dashboard might look like. The analytical team sometimes defends these transgressions, because "business wants it that way". What types of transgressions are we talking about? Let's take a look at the most common ones.
1. The "ultimate dashboard" for everyone and no one
Even more humorously, this is often called as “the cockpit". You just make a dashboard with all the possible KPIs that come to your mind, because "it's interesting", "we need to see it", etc. I saw dashboards where there were more than 30 (!!!) KPI widgets at once, ideally even compared to the previous period.
So why not like that? Such a dashboard violates the basic rule of data storytelling: It is always necessary to consider WHAT and to WHOM you want to tell, whereas taking a solution in style EVERYTHING to EVERYONE does not respond to the basic limitations of human perception. Too much information at once means that we just don't notice what's really important.
These solutions are usually a consequence of following reasons: either the company’s management doesn't really know what to track and tries to track "everything." This is typical in the beginning, when a company embarks on a data-driven path. Or, the second reason often lies in the pragmatic way of saving the visualization tool licenses. Such an ultimate dashboard is usually distributed by email to anyone who requests it. And when you're doing something to suit everyone, you have to make a lot of compromises.
So, what about this? Simply, try to split it according to a simple rule: one use case for one role on one dashboard. E.g. continuous fulfillment of monthly quotas by teams for the sales director is a pretty good example.
Often dashboards, presented as the "beautiful ones" are more a display of the possibilities of a given visualization tool. On one page we find all the types of charts and widgets that the tool offers. Why this is not the right way? Still the same problem. The human brain has limited ability to absorb information, and these galleries of various charts are in fact shattering the attention. So, what to do? Less is often more. Choose the individual visualizations well, with regard to WHAT we want to communicate.
Often because "business wants it that way", these bumfs are created. One loooong page with just about everything. Because someone doesn't want to "click" etc. In addition to the fact, that it is proven, that users simply do not scroll (and such site also suffers from all of the things mentioned above) it has another problem, which is typically performance. If we place 118 visualizations on one page (yes, we have such an example), then we cannot be surprised that loading takes a long time and the reactions of such a dashboard are very slow. Simply, given analytical engine has a lot of work - often quite unnecessarily - and so it takes time.
4. Hell of filters
There is also a filtered hell connected to the bumf - a dashboard, on which dozens of different filters are placed. It is often not clear which filter corresponds to which chart, etc. Such a yuck solution reflecting the effect of automatic filters in Excel. Additionally, loading values into the filter - especially if it has to display only relevant values - simply takes some time. Then complaining about the performance is a bit like locking the barn door after the horse is gone.
5. Wandering through time
This pattern looks innocent at first glance: you have a dashboard called "Monthly plan fulfillment", a filter for switching months, and at the end of the dashboard you place a report that shows long-term sales since the beginning of the year. Confusing. In addition, when you start filtering (this month, last month), at its best, the full-year revenue report will not change by month.
6. War of colors
It's a bit related to the "exhibition". The key is, that the same attribute values (e.g. individual sales teams) are always assigned the same color everywhere. Because otherwise the colors are rather confusing. If the teams in each part of the dashboard are assigned different colors each time, this again reduces the readability of the whole information.
7. It all starts from "Z"
It sounds strange. But the letter "Z" plays a key role in dashboard design. It shows, how to place individual visualizations on the dashboard. Meaning from left to right and from top to bottom. The most important information must be at the top left, then more detailed information (drilldowns etc.) is placed downwards. Again, it has to do something with how the human brain consumes information. Violation of this formula when designing the dashboard again leads to its illegibility.
8. Export table
Perhaps the most common transgression ever. Dashboard, on which, among other things, there is a table that shows all your eight thousand (!!!) clients and 20 different metrics. The goal is to click on Export to XLS and then happily filter everything in Excel. When this happens, that you are forced to create such a solution, believe that there is something very wrong somewhere. In addition to the fact that displaying this kind of table will take a very long time, you mainly change completely the use case of the whole visualization tool. It then does not serve as an analytical / presentation tool, but only as a tool for data acquisition. The actual analytical work is then done in Excel. Do you really want that?
Do you have any questions or comments about this article? Feel free to write to me, I'll be happy to discuss it with you :-)