Analytical Whoopsies (and Other Mishaps)

The ever-delightful Joshua Milligan and I recently did a webinar on the adventures we Tableau analysts take to the perils of mistakes that get us called out by WTFViz and JunkCharts.  It happens to the best of us.

Common Sense is Mythical

Sometimes, we get zoned into our work, forgetting simple things like SUM when it should be average (just don’t look this way), and we think we have something…

Or not…

Sometimes, we study populations based on their composition.  And we look at things based on that population….such as selections, scores or other things.  When you add these across these shifting demographics, you get this VERY clear trend we see here…X even marks the spot!!!!  This will even happen almost every time!!!

Or not…WHOOPSIE!  Next time, I’ll change the aggregation FIRST!!!  Maybe, I’ll try median next time…and then I’ll see the real trend!  (Hint: there’s no X with median.)

Other times, we make a chart and we think we’re on to something…and the thing drops off like The Vallée Blanche in Chamonix, France.  I sure don’t want to ski down that mountain!  And we think we have something!


The line looks great….but, something….something seems amiss….

OOPS!  Despite what we hear from newspapers, teens stop being teens in their teens….

Other times, we take others on their word.  They label the measure something, my favorite being ‘indicator’.  Indicators are magic – they’re glommed up somewhere far away and brought to us mere analysis by the speeds of bits and bites for us to rip apart…

Or throw on a chart…

We’ve gotten schooled on this a few different ways.  Like this.  If we know what we’re measuring, we go a lot farther.

Analysis = (Synthesis + Focus) x Data

Data is like a deluge these days.  As we get more data, we run into the challenge of what to do with it and get taken away with it.  So, we make a bunch of charts and let the user figure it out.  When I was interpreting, we called this making a word salad bar and let the clients pick out what they wanted to eat.  The only problem?  The original message was a four course meal.

I dunno what to say, so DATA VOMIT!


We all have done this at least once.  We get data, we have no clue what to do with it, so we try to answer everything.  Or, we’re too close to it and we think people care about our off the wall tangents with the subject matter.  Sometimes, it works.  Other times, people look the other way.

When we figure out what’s important, or we have great wonderful people give us some much needed perspective, we end up a lot better off.

Pacing. Focus. Happiness.

The more data we have, the more focus we need.  Especially when we try to answer everything.

2 rules, but only 1 answer

But Why Does it Matter?!

We get handed data.  We tear it to shreds like furniture sprayed in catnip.  Then, we hide the mess and sit on our laurels.  Except, we took this whole journey and our logic is hidden.  We put together a bunch of dashboards that, in our head, make sense.

The pressure is ON!!

When we take shortcuts, we lose the logical paths through the dashboard.  It may make sense, or it may not.  WHOOPS!  It’s like trying to follow someone who doesn’t use turn signals that’s diving down back alleys.  Sooner or later, someone will end up lost.

There’s no logic tying these together.

The more we make the data matter, the more we direct and guide naturally, the more likely people are to use it.  The dashboard below doesn’t just summarize the data, it makes it personal and walks through with a little more logic.  It could still be improved.

More logic, still errors – can you spot them?

Video below:


Happy analyzing!

Leave a Reply

Your email address will not be published. Required fields are marked *