Tableau was designed as a tool for composing visualizations—"vizzes" in Tableauese—of data quantities. These visualizations (and I'll use 'viz' and 'vizzes' hereafter) are central to Tableau's value proposition in that they are enormously effective in helping people see and understand the quantities of their various data elements, and in the relationships between the quantities of multiple dimensions in different contexts. This is a very, very good thing, and Tableau supports the creation of these vizzes extremely well. There's a wealth of quantitative vizzes available for viewing at Tableau's Visual Gallery, Tableau Public Viz of the Day, and at many other places.
But there's a problem lurking in the background, a conceptual blind spot limiting Tableau's potential.
The prevailing paradigm is that vizzes are only vizzes when they're conveying quantitative information, when they're presenting numbers coded into a visual form. In the Tableau world measures are first class data citizens with status and prestige, deserving of special treatment, and dimensions are second class citizens whose proper role is to support the measures in their role of communicating the valuable information and insights they have to offer.
This post argues that this perspective is narrow and shortsighted, that it ignores, if it doesn't actually reject, the concept of data that considers dimensions as first class data elements with value to contribute. In fact, dimensions are valuable in the absence of measures, while measures without the context that dimensions provide have limited value. This can be a big pill for many people to swallow, particularly those whose experience has been shaped and is bounded by the more limited horizons; for them, numbers -are- data.
As examples, here are some presentations of quantitative visualizations of 2012 worldwide GDP (in millions of $US) measures from The World Bank's Open Data Catalog:This visualization shows the value of GDP for a particular data set.
Clearly, the addition of the dimensional data adds real value.
But what about this post's title? Is it possible that a viz without -any- measures, without -any- quantitative data be valuable?
Yes. The following situation came up with one of my clients, a multinational organization with multiple regions, each charged with achieving some set of the organization's strategic goals. Each region was responsible for achieving its own results, each tied to the strategic goals, and specified by the organization's overall policies. The Tableau Public published dashboards below show what happened when we first used Tableau to examine the organization's data, which was sourced from an in-house custom-build budgeting and management system that only the technical person who build it understood.
In this situation the simple cross-relating of SOs and RERs instantly showed an occurrence where the Strategic Object 4 program activity for Region 5 was erroneously being implemented instead of the correct SO 3 program, alerting senior management to the situation and providing them with the opportunity to identify the failure to follow policy.
So what? Is this really a big deal?
It is. There are multiple dimensions along which the framing of what data is shapes the analytical approach to exploring, analyzing, comprehending, and communicating our understanding of it. Tableau, as good as it is, has blind spots that are baked into it as consequences of its initial design paradigm, and as long as that paradigm holds the blind spots will exist, and opportunities to improve Tableau as a highly effective data analytical tool will be lost.