Linguists are to text what _____ are to data?
A lot of people are talking about data literacy these days. In fact, I’m one of them.
But what about the flip-side of literacy — the actual “writing”? Shouldn’t we be talking more about the skills needed to effectively communicate the insights gained from pulling together, analyzing and exploring data and numbers: data narration and presentation?
But what are those important skills? And what is important to keep in mind when building a tool to empower the masses to communicate with data?
Keep it simple
Most data tools are power-user tools. In other words, they are not aimed at the every-day knowledge worker, but at someone that can put in the time to learn a lot about data and the tool in question.
The need for data preparation is a good example. Almost every data tool out there requires data to be in a fairly specific format before you can do any work with it. And these formats are far from the same from one tool to another.
The vendor is really asking the customer to translate their data from a format that made perfect sense to them, to a format that made sense to the tool’s developers. And that requires a lot of abstract understanding of data and data models: dimensions, measures, hierarchies and relations. All concepts that most of us are not familiar with in the abstract, although we may surprise ourselves how adept we are with them in practice.
The same is true when building charts. Most tools ask us to do things like “map dimensions to axes”, “define the chart series” and “set the category axis labels”. All very simple concepts to the initiated, but total gibberish to ordinary people despite them knowing their data very well and having a clear picture in their mind of how they want to chart it.
People that have studied the way spreadsheet users make charts have told me that for most users the process is really “clicking around until the chart resembles what I had in mind” rather than methodically setting all the options in a deliberate and informed manner. I’m convinced the same is equally true of other data solutions.
This may sound almost caveman-like to those of us that understand data models and some sort of “grammar of graphics” (although not necessarily The Grammar of Graphics). But is it really such a strange behaviour given the tools they are provided with?
As my friend and mentor Donald Farmer put this in our recent conversation: “You don’t have to be a linguist to write great text, why should you have to be a data expert to communicate with data?”
At GRID, our vision is that working with data and numbers should be as easy as working with the written word.
When it comes to data preparation we believe in prep-less data work: bring in the data in the format that makes the most sense to you, and we’ll ask questions as needed along the way to get your task done.
And when it comes to charting, our focus is on you describing the chart you want to make and our system deducting the data model, rather than asking you to define an abstract data model that the system then uses to build a chart.
We’re by no means there yet, but we seem to be off to a decent start.
This article has also been published on Medium.