Analytics on offense: How to build a data strategy
A big Wall Street bank’s Chief Digital Officer and an analytics thought leader walk into a bar.
The analytics person says, “So, tell me about your data strategy.” The CDO nods. After some careful consideration, he replies, “Yeah, we still need to decide what we’re going to do with our data.”
This is the anti-punchline of analytics: Every big enterprise is collecting troves of data, but few have agreed on what to do with it or how to operationalize that information.
In recounting this story for me (which is not a bad joke, but rather a recent conversation) Tom Davenport said, “This Chief Digital Officer? He’s a very smart guy. He was right, but one would think all that would have happened earlier. Basic questions like, how are we going to make money on our data, would have already been answered.”
Tom has conversations like this all the time. The President’s Distinguished Professor in Management and Information Technology at Babson College, he wrote the book on it (literally, it’s called Competing on Analytics). Having consulted with dozens of enterprise companies fine-tuning their data schemes, he’s identified a couple areas for improvement.
Enterprise companies need to define their data strategies upfront and decide on the right balance of defense and offense. And then they need to put that plan in motion. Here’s how.
What is a data strategy?
To many, data strategy probably sounds like a mashup of the two most overused words in 21st century business. But it’s a real thing. Going beyond mere collection, a data strategy also considers purpose and, as a result, use.
“Data has been dealt with in largely technical approaches thus far,” Tom said. “A lot of work around how we capture it, store it, analyze it, and so on, has been done, but without a huge amount of thought as to what kind of data really matters to us and what we’re going to do with it.”
Part of this is first-mover anxiety. Suddenly, Big Data was a thing everyone had to invest in. Enterprise businesses couldn’t spare an opportunity to fix inefficiencies in their own processes or find opportunities for their growth. Except they didn’t always tie it back to their efficiency or growth.
It was somehow enough to know that somewhere things were being tracked.
In Tom’s mind, management theorist Peter Drucker set up the dichotomy companies are living through perfectly: “Information is data with relevance and purpose.” Most companies are data-rich these days. And yet, in the data revolution, they’ve become information-poor.
So, Tom’s advice is for companies to actually assemble a stated data strategy. Technology teams and business unit leaders need to come together to decide what the priorities are for data usage, where it can have the biggest impact, and how they’re going to operationalize those priorities.
“Technology people alone can’t address this,” Tom said. “They’ve got to engage with their senior executives. Once they’ve done that, one key issue is to focus on either controlling data or freeing it up.”
Naturally, there’s friction here, but it’s healthy friction. It can clarify where data is already being used efficiently and where analytics aren’t fully operationalized.
“If you’re on the user side, you tend to focus on the “freeing data up” side,” Tom said. “If you’re on the IT side, you tend to focus on the controlling dimension of it. The balance you maintain is a dynamic thing, but currently the conversations between those two poles aren’t very enlightened.”
An important stepping stone to a mature data strategy is deciding how much you play on defense and how much you play on offense. There’s no categorical answer, but you should have one for your own business.
Defense is the default
Offense and defense are both integral parts of winning any game. The decision to emphasize one or the other depends on your strengths and goals. Like any shrewd business decision, you need to think about how your data corresponds to your goals.
“One big question is do we really focus on offense or defense, in terms of data,” Tom said. “That seems like an obvious question, but most organizations don’t really have any sort of prioritization of offense versus defense.”
A lot of companies fall into defense by default. Defense involvess things like cybersecurity, preventing breaches and hacks, following regulations, and data integrity.
The all-defense, all-the-time status quo comes from a relatable sentiment: Don’t screw anything up. Analytics professionals have been trusted with a resource that, to their colleagues, is largely mysterious. They don’t want it to be liability.
But not screwing anything up has hamstrung businesses. It’s precluded the intense kind of experimentation data allows.
“Banks are a great illustration of this problem,” Tom said. “They’ve collected massive amounts of data, but most have not really done much with it to help their customers. In the vast majority of cases, they’re not sure how to make money with it.”
He believes a mixed approach is the right approach. In formulating a ratio of offense-to-defense, analytics professionals and business leaders should know what offense has to offer.
“I’ve tended to be more of an offense person throughout my career,” Tom said. “Because I’m primarily interested in how companies can use information to make more money, build new products, and improve their relationships with customers. Analytics, which I’ve done a lot of research and writing on, is mostly an offense activity.”
For companies that are interested in incorporating those themes into their data strategies, getting serious about offense is important.
Quarterbacking
“There has been too much focus on control and engineering from the top down of data,” Tom said. “Not enough focus on really identifying what data the organization has available and how you get it.”
Instead of holding onto the ball for dear life, analytics professionals need to quarterback. They need to define the play and lean on tactical business players to execute it. Rather than gatekeeping, they need to open doors.
“In terms of analytics organizations, it’s suboptimal if the people who are doing this kind of work are not coordinating and sharing their experience,” Tom said.
As a result, it might be time to restructure teams, or at least lines of communication. The way we share information in organizations has become outdated. “Centers of excellence” were a natural evolution, but they may be a relic of the past.
“For a while, I thought the ideal situation was one big centralized analytics group,” Tom said. “I don’t necessarily feel that way anymore, just because there are so many different things you can do with data and analytics. There’s more and more specialization, and more and more business unit executives are starting to understand things. As long as there’s still some collaboration and communication, I think a more decentralized approach can work out quite well.”
In some cases, non-analytics professionals may be able to take some of the lower-level work off their colleagues’ plates. With the technology available, there’s no reason organizations can’t make their business units more analytical, leaving the centers of excellence to focus on more strategic initiatives.
“Try to work with your IT organization if you can, but waiting around until it becomes your turn is not a viable response these days,” Tom said.
If this sounds like it’s more concerned with operations than analytics themselves, there might be some truth to that. The key to making your business more data-driven often lies in what Tom calls “operational analytics.”
Operational analytics
By necessity, analytics once had to be focused narrowly. It was an arcane science. The technologies were limited, and so was its business power.
“Analytics were sort of an artisanal activity at one point,” Tom said. “They took a while to develop.”
Now, what we can do with analytics seems limitless, and many business decision makers are bought into it, too.
“With operational analytics, you no longer have the problem of is this executive going to use my analytics to make a decision, or is he going to fall back on his gut, because it’s built into the process,” Tom said.
But your analytics are only as strong as that process. You need to define a good data strategy and execute it. That means operationalizing analytics.
“It makes sense that we’d start moving toward more operational analytics, which means embedding analytics into our systems and processes that we use to run the business,” Tom said.
There are lots of questions that accompany this shift like, What tools should I use? But analytics professionals and the executives that oversee them should first focus on data strategy. If collecting data, companies should define why that data is meaningful to business questions, as opposed to just diverting that information into a data lake.
Instead of accepting someone else’s playbook, data strategists should decide for themselves how much offense to play versus how much defense. When dollars and cents are on the line, getting business value from data should be as important as guaranteeing its integrity.
And once you’ve got the strategy outlined, you should commit to it. That might mean tweaking parts of your business model or your organization to accommodate it. Does that sound onerous? Then don’t call yourself data-driven. The ones who truly are will devise a strategy for their data, and not just tactics.
Nobody is expected to have their analytics all figured out. It’s okay to have your analytics be a work-in-progress. Even the industry leaders are a work-in-progress.
“To remain successful with this stuff, you have to really keep working at it and not rest on your laurels,” he said.
Analytics might not be the backbone of your business tomorrow. But you’ll never find yourself confessing, “We need to figure out what to do with our data.”
***
Be sure to check out Tom’s new book, Only Humans Need Apply, which draws on his years of expertise to reframes the conversation around artificial intelligence. In it, he explains why the future of automation will push human-computer collaboration forward, instead of displacing us.
