Power BI, Excel and other BI tools provide unparalleled power to gain insight, these tools though, do not necessarily prevent grave information design mistakes
that are at best misrepresentations at worst plain lies.
I recently came across a post where a self-proclaimed “Excel/Power BI Dashboard Trainer” was criticizing concepts -presented in a webinar by very experienced information design experts- as being too boring just “BLACK, RED, GREY and GREEN”. He followed it up with a dashboard of his design that “drips with creativity and forethought”:
I politely responded to the post pointing out the shortcomings as above
His response was to delete my comment without any further explanation. What he at least did after my comment is to add the admission that he his “breaking industry rules”:
The aim of this post (finally I found a few minutes…) is to cover some of the key mistakes that I still see way too often at our clients and unfortunately even with “trainers” that should know better.
My tips reflect commonly agreed standards by information design experts (Edward Tufte, Stephen Few Rolf Hichert and IBCS) that are really helpful if your aim is to provide real insights that are understood at a glance as opposed to “cool dashboards” that are fraught with “information noise” and no real added value .
Here is a short overview of what are in my experience key things to consider:
This is the key one: your viewers have no chance of quickly understanding your message/dashboard if the creator “wants to explore her/his artistic side” and is arbitrarily using visualizations, colors and layouts. The same data types should use the same colors, graph types, plot styles, etc. For example, Actual data is always dark and solid, Budget is lighter and dotted, revenues use a consistent style that is different to expenses or percentage KPI’s.
I am not as dogmatic to postulate “you are only allowed to use black, green and red (Sorry Rolf and Juergen) but a coherent approach is absolutely essential. I also think dark backgrounds are OK, which some of the information design experts would frown upon. In my humble opinion it’s easier on the eye and believe it or not saves energy as white pixels consume more:
A situation as in Exhibit A where alert colors like red are used all over the place for completely different things is definitely not a good idea to help your dashboard users. With colors I would aim for subtlety and only use strong, bright colors for some kind of alerts.
Another key area where I see a lot of room for improvement is scaling. For data that should be viewed in comparison it is imperative that you use the same scaling.
This can be an issue with Power BI as separate visuals will automatically scale for the particular data points in that visual. In this case it is important that you set the same axis endpoints on them:
Power BI “out of the box”
Other absolute no-noes are “cut off axis” as in our “Exhibit A” an 11% variance is depicted as if it was more than 50%:
Avoid information Noise
The “ink to information” ratio is another key concept in information design. If the question “does this information pixel add value to clarity or making my point” can’t be answered in the affirmative, it likely isn’t a good idea.
I can see that in rare cases an associated icon can potentially help on dashboards with new users or if they don’t speak the language otherwise the title and labels in your chart should suffice. From colorful gif galore (yes, I have seen animated ones…) I would definitely refrain.
These three concepts are just a fraction of what more comprehensive frameworks like IBCS cover but if these are adhered to, a dashboard/report designer is likely on a good path to achieve clarity, truthfulness and fit for purpose. For creative outpourings, a gallery is likely a better medium….