Last week, we posted a guide to formatting a beautiful, high-level dashboard in just 10 minutes. Alas, the second comment, from Cristian Vasile, was not complimentary. He called it “a complete failure.”


We took a step back and realized that we’d focused on beauty instead of functionality.

The dashboard sure looked pretty, but we realized we’d forgotten a fundamental aspect of visualization: the need to communicate what each mark means. The dashboard was lovely, but we’d gone too far and removed too many labels, leaving the lines and marks without any meaning. Even if you’re doing high-level dashboards, it still needs to be readable at a glance.

Finding the balance of beauty and functionality is one of the toughest things to do. Just because it’s high-level doesn’t mean it should sacrifice functionality. How can someone engage with something that is all beauty and no functionality? Once they ooh and aah at the pretty formatting, what would happen when they realize the charts, without labels, convey no meaning? Who’d invest in a dashboard like that?

Anika and I had a quick chat and we worked out we could maintain the beauty by adding the absolute minimum of labels. Below is our reworked version of the dashboard. What I’ve always loved about Tableau is the speed with which you can redesign a dashboard in response to feedback. It’s fast and allows you to continuously iterate until everyone’s happy.

What do you think? I hope you think it is still a beautiful key-metrics dashboard, now with added functionality.

Cristian, I’d like to thank you for your comment. I would love to know your thoughts! If you or anyone still thinks the dashboard isn’t right, let us know why in the comments. Even better, download the dashboard and show us your own version.

I have also published a post on my own blog detailing all the design choices made in reworking this dashboard. Let me know which you agree and disagree with.

Update: this post was published with an earlier iteration of my reworked dashboard. I have replaced it with the intended final version, which has slightly better (but not perfect) clarity on the marks in the Profit charts.


For sure in business beauty must have form and function. Overall the DB looks good at first; however the color choice on the furniture category leaves something to be desired, as it washes out quickly and does not lend itself to quick ingestion of the information it is showing off. I was also curious why the two graphs to the upper left and middle would show max value in $ and yet the top right is in pounds sterling, all this in conjunction with a map of the U.S.? I think my executives would ask more questions than have questions answered by this db in its current status.

Andy, great work. An improvement over the previous version, although I didn't think it was that bad before. I would agree with Steve above that the color needs to be reevaluated, both the pinks and the overall color concept. One of the key takeaways I learned from your session in Vegas was that color can be used to tell the story immediately (ex: Iraq's bloody toll). If I only had a few seconds to glance at this dashboard my two key insights I would gain would be the total sales and the profit which have some the best real estate on the page, but I'm left wondering if those two KPIs are good or not.

I wanted to examine this under the lens of beauty as meaningful design (see Kelly Martin's take on this here: Please keep in mind, I claim no expertise in this area and I'm exploring myself. For me, the color didn't support what was sales versus profit versus quantity. I'm also anti-pink, but that's just my shade of crazy. I did give myself more than 10 minutes because I wanted to take the angle of if I were giving this to my executives, and less on the 10 minute aspect.

Main changes I made:
- Consolidated sales and profit - it allowed me to show profit a bit more relative to sales.
- Moved labels to the end - yes, this creates additional views to render and we can discuss the optimization/sub-optimization of this, but they felt more logical at the end.
- Muted colors and reduced border effect
- Moved map and added icon to show that this thing filtered everything. I didn't even realize it did until I accidentally clicked when I was deciding to add action filters. I didn't add any, just called attention to it.
- Tried to make cleaner lines in the floating version.

Again, I took far more time that 10 minutes, because my angle was beauty as meaningful design and what I'd give an exec (the likely audience of this dashboard).

Thanks for all your comments. Let me first say, if you wish to go see details of all the formatting changes I made, go check this post on

Steve - $ and £. OMG - how did I not spot that. Yes, that's a fail.

Steve, Jed and Bridget - the colours. Oh man, the colours. Most of the changes were easy, but I played around with the colours for a long time. I couldn't seem to get them right, and I too am not too happy with the mix of blues pinks and greens on this dashboard. One important thing - this post was originally published with those pinks and no mark borders. The image now has borders around the Profit marks. In my first draft, I too had realised the pink was too washed out.

Bridget - just wow. You smashed it! I love the banner. I love the all caps title. I love the proportions of the different charts on two rows. That's a really really smart reworking. Congrats!

What's my conclusion? A dashboard will never be 100% complete. But thankfully I use a tool which allows me to iterate and get feedback and change things quickly.

Clearly better than the previous version.

Not sure "quantity" is a good metric, especially when you get several categories of item. Two desktop and three computers may make five but this doesn't make sense.


Yes, thank you, Andy, for sharing this dashboard design. I appreciate that obsession over the details to make sure the dashboard makes sense at a glance to a business leader.

That said, I've found that the design hierarchy is best when it aligns to the importance of the business KPIs. On this dash example, the first thing I notice is the big map, then the prominent purple bar charts. But in this case, wouldn't "Total Sales" or possibly "Total Profit" be the No. 1 KPI?

If so, the dashboard would be even better if the No. 1 KPI was the No. 1 dominant visual, then the No. 2 KPI would be the No. 2 visual and so on.

Hello Andy,

Thank you for involvement and nice words, you are a real gentleman. On a scale from 1-10 I give 9 on updated design.


Hi Andy,

Did you simulate margin by shading the containers? One thing I really wish is that Tableau had margin possibilities between containers built in, and also that I could always set the size of containers exactly.


Nice remake of the first design. A definite improvement. I also had a look at Bridget's version and I think she did an excellent job improving it further.


Here's what occurs to me when I look it over:

The pink/purple colors are actively offputting, both draw attention and and repel vision, particularly since they the charts' bars have such different visual weights.
Tt's difficult to pick out the Furniture bar segments (too light, at least in web view)

Data content / Time Span / Granularity
Order Date is filtered to "Last 6 Years"
It's not clear what "Last 6 Years" means.
Is it all of the five years prior to "this year", plus all of this year to the date the dahboard is rendered, or
is it from six years to the day prior to when this dashboard was rendered, or
something else?
In any case, "Last 6 Years" is at odds with the dashboards' other presentations of date spans.
The Total Sales and Total Profit time spans shown are from 2011 (when in the year isn't clear, is it the beginning of the year?) to almost 2015 - clearly not the last 6 years

It's not clear whether Total Sales & Total Profit are per Month or some other period (Month is a reasonable guess, given the # of dots/marks, but a small label would make this clear).

Profit by Category and Sales Trends are horizontally partitioned into Months, but it's not clear for which Year(s).
And it's a legitimate question as to whether lumping Monthly Sales and Profits across Years makes sense-it may or may not; this is particularly relevant when showing the daily/monthly Sales Trends, e.g. does the dark blue mark for M/18 indicate high sales for one M/18, or all M/18s across some number of years

The different segmentations for the date distributions of data imposes a cognitive load on the dashboard interpreter.
Particularly for the top three charts - the leftmost pari appear to have the same structure, but Profit by Category is different. Given that the dominant geometric visual organization groups these three into a horizontal band the discontinuities between them is discordant. If the intent is to tie the vertical bars together to show the distributions of Profits by Category both Monthly and Yearly this dashboard doesn't do it well - the two charts don't tie together visually.

I agree with Chris Tauber. The map is visually dominant, drawing attention to itself, which feels disproportinate to the information value it's conveying. If the Quanity by State shown mapwise is the most important information to convey, it would work better top-left.
Also: is the map showing the aggregated quanitity by State for the same time period as the other charts? Which brings the time period questions into play again.

Total Sales v Total Profit
It would be really helpful for these two charts to be as similar as possible in order to support comparison of their mettrics.
They fall short in a couple of ways.
The Total Sales under-line area is shaded, but the Total Profit under-line area is not. This is confusing, at least at the level of needing to reconcile the different presentations.
The zero lines are not aligned horizontally, making it harder to compare deviations from baselines than it should be. It can be tricky to get this alignment, but it's useful.

Subscribe to our blog