Data visualisation success hinges on solid storytelling skills

Tableau's Jock Mackinlay explains why data is inert and worthless without the twin practices of visualisation and storytelling.

Data by itself is fairly useless without solid storytelling and narrative skills, according to Jock Mackinlay. Mackinlay is director of visual analysis at Tableau. He's an alumnus of UC Berkeley and Stanford University and a veteran of Xerox PARC. He is also known for coining the term "information visualization" in his 1999 book Readings in Information Visualization: Using Vision to Think.

He recently took time out from a family holiday in the UK to talk to What follows is an edited version of that interview.

Why do you say that businesses "must" use visualisation tools to allow data storytelling and to aid decision-making? It seems like a strong statement.

Jock Mackinlay: Data by itself is not worth much; it's only worth something in respect to what people are doing. And if you make the data visual and simple to work with, [it's] more accessible. The challenge is that data is very relevant to lots of endeavours, and it is people who have to take advantage of it. Take cognitive biases, for example. Smart people are susceptible to those because they are good at jumping to a conclusion very fast. One of the valuable things about data is that it stops smart people from falling prey to their cognitive biases [or] jumping the wrong conclusion.

For more on data visualisation and storytelling

Analyst views on how visualization proves the value of big data analytics

How data scientists impose narrative form to bridge analysis gap

Data scientists mining data from transaction data

So that is the moment of analysis. But what about storytelling? Why that emphasis?

Mackinlay: Once you have some understanding of your data you have to tell another human. First there is analysis and then there is storytelling. And they blend back together. The classic way of doing this is traditional business intelligence (BI), but that is slow for organisations to do. There is the whole chain of data warehousing [and] getting IT to build reports. Business users should not have to find a statistician or programmer when data is becoming democratised within organisations -- who are increasingly encouraging employees at all levels to use and interpret data.

How does storytelling balance out the requirement for a new cadre of data scientists?

Mackinlay: Data scientists are valuable but if you become dependent, then it is like being dependent on IT again. Our focus is on the data enthusiast, partly because of the nature of our technology. We want to empower the data enthusiasts so they can get along by themselves. The Tableau technology that Pat Hanrahan, formerly of Pixar, and Chris Stolte developed out of Stanford is partly about drag-and-drop and partly about direct connection to databases. So, it can be used on all kinds of databases by all kinds of people.

Data by itself is not worth much.

Jock Mackinlay, director of visual analysis, Tableau Software

Can you give an example of a business decision that data visualisation has enabled that would otherwise not have been taken?

Mackinlay: It's more about making the decision with more certainty and much more quickly. Speed is essential in the modern world. The biggest return on value for our kind of technology is that. If you want to do lightweight equations, Microsoft Excel is perfect, but once you've got bigger data, it's not good. We are very human oriented. My title is director of visual analysis; we are not a data visualisation company, as such, we are a visual analysis company.

What is your take on all the attention being given to "big data" of late?

Mackinlay: Big data is a fad. It is essentially parallel computing. The MapReduce technology developed at Google, and the development of that into Hadoop makes parallel computing less expensive. The weaknesses of Hadoop is that the computation still takes time, there is still latency. You have to write a programme, let it run, and wait. Of course, there are researchers working on reducing latency. In the meantime you have architectures like eBay's -- MapReduce/Teradata/Tableau. But I come back to the main thing, which is that data by itself is not valuable. Our data enthusiasts love the speed and storytelling made possible by data visualisation.

Read more on Business intelligence and analytics