News Stay informed about the latest enterprise technology news and product updates.

An analytics team how-to for small business, from Google

Experimentation has become trendy. The test-and-learn philosophy — a variation on the fail fast approach — has been embraced by companies for building products or adding new features to a website. CIOs should also be incorporating the same approach in how they structure IT teams, hire new employees and even work with the business. That was the big takeaway from Deepak Tiwari’s short presentation at the recent Useful Business Analytics Summit.

As the head of strategic analytics and insights for Google’s consumer operations division, Tiwari has lived through his share of trials, errors and successes. Three years ago when he joined Google, he oversaw a team of two. Today, he’s leading 30-plus employees to provide support for and user insights in to Google consumer products. Here are five “lessons learned” he shared with attendees.

Remain independent. Arguments have been made for a decentralized approach to analytics, but for Tiwari and his team, centralization is crucial. While they serve and support the lines of business, they don’t report to them.

The buffer ensures a consistency across the organization rather than having to bend to the different dynamics of each business group. “Even if you’re a small analytics team, that’s how you’ll succeed and make your mark,” Tiwari said. “Make sure you remain independent in some way shape or form.”

Build a solid team culture. Tiwari’s “hire the best people” and “pay them well” are two obvious pieces of advice, but it’s important to remember that “best people” is subjective. A good fit at Google could be a bad fit at a non-Google company. And, while “salary matters,” Tiwari said, “that’s not the only thing that will keep them there.” Music to a small businesses’ ears, no doubt.

Tiwari made three additional suggestions: Hire slowly and take time to find the right people, give incoming employees interesting business problems to solve and build a solid team culture. Tiwari said his team “loves hanging out,” frequently grabbing lunch and dinner together. While that’s easy to do at Google (hello, free food), it’s also a possibility for smaller businesses to plan group down time.

Hire philosophers. Tiwari encourages businesses to avoid getting caught up in a data scientist game of hide and seek. The reality is, there is no simple method for finding the elusive data scientist or even a consistent job description to refer to.

In other words, “there is no secret sauce,” Tiwari said. That might mean thinking about the role of a data scientist as a team of people rather than a single individual or expanding your search to include candidates who don’t call themselves data scientists.

That’s good news for small businesses, which may not have the money to hire someone with a data scientist title. Instead, find people who are “hungry to look for patterns or hungry to do this kind work,” he said. Those qualities don’t begin and end with engineering, programming or computer science.

“You’re almost hiring philosophers,” he said. “And a philosopher, you can find in anybody.”

Find places to automate. As the team goes on to solve bigger — and harder — problems, the tediousness of, say, churning out reports can present a hurdle. One solution? Find ways to automate, Tiwari said.

Attribution modeling is an example. Businesses use this to pinpoint what events contributed to a certain outcome, such as a sale. So if revenue went up by 10% instead of the predicted 9%, “somebody would have to go and do an analysis,” Tiwari said. The good news? “If you have the data structure in place, if you have the data in place, attribution can be achieved with the click of a button,” he said. That frees up the team for exploring the data. “If you’re going to hire good people,” he said. “Make sure they’re not spending a lot of time on reporting – that they’re doing more insightful and strategic work.”

Not everyone needs to code. It’s also worth thinking about where to implement self-service technology so that it can, as Tiwari put it, “give power back to the people.”

He and his team are doing just that by “building experimentation platforms so even someone who is not very technical, someone, maybe, who is writing content, can actually go in and look at the impact of their content,” he said.

That kind of a project might be out of a small business’ scope, but figuring out ways to open up data and build processes the business can operate without IT can be a time saver. As, Tiwari pointed out, “not everyone has to learn how to code in Python.”

Start the conversation

Send me notifications when other members comment.

Please create a username to comment.