Is a table a valid way to visualise data?

Andy Cotgreave
December 19, 2012

There is a great thread on our community right now. It started off with an innocent question about sub-totals. Once our Tableau Zen Masters got involved, it morphed into a great debate: are tables useful in data visualisation? This is a really important question, and one we cover in our Visual Analytics course. It's so important, I've made this post. Let me know what you think on twitter (@acotgreave)

1. Why do people use tables when charts are "better"?
2. Why bother making tables in Tableau?

Why do people use tables when charts are "better"?
Because they provide precision where visuals do not. Consider the chart below.


Now answer two questions:
1. What has happened to Personal Income Tax over the time period?
2. How much Personal Income Tax was raised in 2000?

Question 1 is easy with a line chart, Question 2 is not.

Remember this: EVERY visualisation is a compromise. The second question can't be answered without some sort of precision. A table gives you precision, if that's what is most important.

Why bother in Tableau?
Let's repeat the mantra I just said: EVERY visualisation is a compromise.
And now add Ben Schneiderman's visualisation mantra: Overview....Zoom and Filter.... Details on Demand.
What's that got to do with Tableau? Dashboards and multiple perspectives on data. With Tableau you can get the best of both worlds. You can draw the time series, and the table. Either draw all the info in each or link them with Actions.

In our Visual Analytics course, we use the following as an example of a well-designed Tableau Table:

And then make the point that if you combine charts and tables you get the best of both worlds, like this:

My conclusion from this: Tables are valid as part of the analytical cycle. They are often necessary and form a key part of many well designed dashboards. Therefore - yes, Tableau should continue making our tables amazing and beautiful.


I often have this discussion with clients and for me it often comes down to the goal of the display. If I'm building an earnings call dashboard, for example, that requires the user to be able to easily lookup and answer specific questions from analyst, then a table works well. If you rather need to look at more than just individual data points a more visual display work much better (except pies).

A common technique I use in balancing the need for both specificity and trending information is to use both a table and visual display. Sometimes if the details are too immense to display on initial load I may have the user first select some element of the visual display to unhide the details (eg. table).

Great discussion and post Andy!