What would Joe Mako do?

One of my clients is getting fairly far along with their Tableau experience and competence. Which is really rewarding, and bringing along a raft of challenges. As their sophistication grows their interest in more substantial and innovative analyses keeps pace.

A week or so ago I was asked to assist in creating out a particularly interesting analysis. The details don't matter, but it was subtle and complicated in some really interesting ways. I wasn't even sure if it -could- be done in Tableau, much less how to do it. So I started puzzling it out – the usual process of contemplating the data, the desired outcome, and Tableau's capabilities. It was hard going, the pieces weren't falling into place, remained half-glimpsed shapes veiled in the dim corners of possibility.

And then, my breakthrough—I thought to myself:

"What would Joe Mako do?"

So I gave it a shot, tried thinking about the things Joe has contributed to the Tableau community and the thinking behind them. Dragging mostly-forgotten or unrecognized insights up from the deep. Remembering what Joe has said about the nature of VizQL and Tableau's data operational model as best I could. (something in there about the six layers/levels/steps involved in getting data from the source to the viz)

Slowly, slowly, shapes began to emerge from the dust, noise and confusion and cluster themselves into formations, much like structures revealed by X-ray crystallography.

After worrying it for a spell like a Terrier with a rat I felt increasingly comfortable that not only could Tableau handle the problem, but that I was able to coax it into shape. At the end there was a bit of a leap of faith, but I was confident that I'd found the right approach and the path was clear, so I took the last step. And everything fell –snap into place and voila!

Thanks, Joe.

Failure to Identify -or- Who is that mystery measure?

Table with One Measure – Nice Numbers, But For What?

This visualization is incredibly easy to create.
Unfortunately, it's impossible to interpret meaningfully because Tableau doesn't provide any information about what the numbers shown mean.
There are ways to have Tableau present a label for the numbers, but they're hacks taking advantage of side effects in Tableau's mechanisms to produce what should be a first-order functionality.
Which is a long-winded way of saying that Tableau should label everything it shows by default, making interpretation clear and easy.
We shouldn't have to 'fix' things like this, that shouldn't be broken in the first place.

Table with Two Measures – Each Clearly Labeled

This visualization can be created from the One-Measure table with a simple double-click of the second measure.
Strangely, when adding the second measure Tableau finds it appropriate to label them both. Why is this? Why label two but not one? Is one measure, presented solo, somehow not worthy of being indentified?
It's a bona fide mystery.

The Tableau UI for Two-Measure Table

The Tableau UI for One-Measure Table

When adding the second measure, Tableau also adds "Measure Names" to the columns shelf, which in turn provides the Measures' labeling. This seems all nice and reasonable: "Measure Names" => labels.
So it also seems reasonable to assume that adding "Measure Names" to the One-Measure table's Columns shelf will label the single column with the appropriate Measure Name. (Even though this is an imposition upon the User, making him/her do something to make up for Tableau's shortcoming).

But it doesn't work as intended.
Tableau really doesn't know what to do with "Measure Names" on the Columns shelf when only one Measure is in the viz. here's what happens:

Tableau obligingly, and erroneously, informs us that there is "No Measure Value" in the viz.
This is wrong, and makes it harder to convince people that Tableau is easy and trustworthy.