The Elusive Definition of Agile Analytics
It's about time for an authentic definition of "agile analytics" based on analysts' understanding.
- By Ted Cuzzillo
What is agile analytics? Answering such a simple question was like stalking a definition in the wild.
Software vendor marketers like the term "agile analytics." Lacking the anchors of its cousin "agile development," it can morph to please. At first, it seems to mean speedy analytics, but the more you look, the more the term grows to become one of those rapid-deployment, take-it-anywhere term.
A few experts have offered definitions, most of which seem intended for other experts or the anxious eyes of senior management. To find an authentic definition, I went into the wild to find one that everyday analysts might identify with.
At first when I went searching for a definition of agile analytics, it seemed as if I wanted something that wasn't there. One analyst who's always good for a strong opinion replied, "I have no idea." Two industry observers begged off. My questions on Quora and Linkedin prompted zero responses. Was this an omen?
My first insightful answer came out of a large, established BI vendor from a longtime source who, as always, asked for anonymity: "This has always been the promise of BI," he wrote in an e-mail message, "It was just a lot more difficult than previously thought."
The data models weren't sophisticated enough to handle the many aspects of business, he explained, and people who designed the models couldn't see far enough ahead to anticipate problems. Agile, by contrast, goes where no one has gone before. It's "a sort of Lewis and Clark affair."
I tried the "Lewis and Clark" metaphor on a longtime Wall Street analyst who has since also become a vendor. He rejected it. Sandy Steier, vice president of 1010 Data, offered a rapid succession of alternatives starting with the Interstate Highway System. He wasn't happy with any by the time he had to end our conversation, but he did seem to grasp agile thinking: "It's an open road. ... It's like flying without a flight plan ... freedom, lack of organizational structure, planning, implementation, processes, all that stuff. ... It's true, Lewis and Clark didn't have that. No one planned the route. They just followed the lay of the land. ... Freewheeling ... Turn on a dime, go your own way ..."
I asked a series of analysts who, at least the last time I talked to them, had no direct stake in any tool.
As if to advertise Amtrak, Andy Cotgreave cited his blog post, The Journey is the Destination. Until recently, he was a senior data analyst at Oxford University and has now joined Tableau Software. He wrote in an e-mail message, "It is the ability to explore data without a specific end point in mind." You can't decide beforehand where data's going to take you, so you can't tolerate any tool or process that artificially restricts the conclusions.
Several others also emphasized tools that encourage experimentation, and one-after-the-other cycles of questions and answers.
Then I heard from an analyst with a formal view. Agile analytics "is not to conform to any particular process," writes Breanne Cameron, product manager at Cambio Technologies, LLC and holder of a master's degree from the Institute for Advanced Analytics at North Carolina State University. "Once a problem has been defined, there is a usual, or typical, broad path the analyst can follow to solve that problem," she wrote me. "However, data is dirty and the world is not perfect." Analysts have to stay on their toes and recognize challenges that can alter the direction of the analysis.
Nimble as all this may be, well done agile is careful, too. For analysts, it means staying aware of assumptions, knowing what violates them, and what danger may result. Some analyze data "blindly," she writes, and "that scares me."
Joe Mako, renowned for his technical wizardry among fellow analysts, mentioned exploring data with a client alongside. Mako creates viewpoints guided by the client's knowledge of the subject.
Finally, someone invoked culture. "Agile analytics starts with the corporate culture," wrote a data architect at a Bay Area firm founded in the '90s. It's not just about tools, modeling, good data, and bad old IT butting out. Agile minds have set up a centralized analytics platform, he says, and have a "wide variety of mechanisms" to makes sense of the data "in a couple of hours" even for a brand new idea or question.
So what is agile analytics? Over a week, several themes emerged:
- This may sound like marketing, but it's true: Agile analysis lets people analyze data the way they actually think. They try out one idea after another, recalling one chunk of data after another as needed. Sometimes they're confused, sometimes they're clear. The process is inefficient and nonlinear, but they're more likely to figure something out this way than any other way.
- Quick iterations. It's like agile development in this way.
- Data granular enough to answer unanticipated questions.
- New importance of personal skill, knowledge of the subject, and skill managing data.
- It thrives in organizations that encourage it.
Agile analytics is a style of data analysis that uses quick, repeated, and uninhibited experimentation with granular data, either subsets or whole sets, usually with tools designed for this style. It often involves collaboration, such as between subject expert and analyst or between two or more analysts. Speed to insight is often but not necessarily a byproduct.
I still want to know more about the effects of organizational culture. Can data analysis be truly agile, for example, within an organization whose style conflicts with it, such as when IT's parental grip won't loosen?
For now, here's a clue: "You can't buy it," said Gartner vice president of research Merv Adrian last August about a different type of agile. Another clue: CONNECT founder and TDWI faculty member Maureen Clarry's astute observations that either/or dilemmas within organizations thwart agility.
If you can't buy agile, can't command it, and can't fake it, then you have to be it. Socrates must have known all about looking around for what we could have had all along, observing that "you always take yourself with you." Be agile to know agile.
||Ted Cuzzillo is a journalist and industry analyst focused on analysts' tools and needs as well as the environments in which they work. You can contact him directly at
firstname.lastname@example.org. If you analyze data, he'd appreciate your participation in his survey; you'll receive a free preview of his report when it's complete. |