A data scientist points out that when it comes to making innovation a scientific process, it’s important to keep human nature in mind.

If data can enable a successful innovation process to be repeated over and over again, it’s Anna Gordon’s job to find out how.

Gordon is a data scientist at San Francisco-based Mindjet, a company that develops collaboration software. Mindjet‘s clients, including American Express, Merck and Cisco, use a product called SpigitEngage to present challenges to their employees in the hopes of crowdsourcing a solution. This technology is also at the core of products from Mindjet’s competitors like Waltham, Massachusetts-based InnoCentive and Boston-based InnovationCast.

It’s easy to understand why large organizations with hundreds to thousands of employees are eager to utilize their collective brainpower. But it’s only relatively recently that they’ve been able to use technology to do it in an efficacious way.

In a recent post on VentureBeat, Gordon described how she’s constantly trying to refine that technology. She revealed that in order to pull it off, she needs to give almost as much thought to human psychology as she does to algorithms and data–or understanding “crowd science.”

How It Works

To understand why, you first need to understand how companies use this innovation software. First, a leader at the company poses a problem. Everyone who has a login — from janitors to managers — has a a chance to weigh in, Gordon told Inc. The ultimate goal is to use the platform to float the best — and potentially million dollar — ideas up to the top leadership.

Employees help to do this essentially by “up-voting” their favorite ideas. They’re presented with two submissions, and out of those, they vote for their favorite. Here’s where the human psychology part comes in.

Because there is usually some kind of prize involved to incentivize employees to participate, some users have a tendency to try to game they system by asking their friends to up-vote all of their ideas. For that reason, Gordon and her team wrote an algorithm to identify these people in order to flag their submissions.

Future Applications

Gordon and her team’s analysis of this hard data has also uncovered some pretty positive human insights.

The system has begun to reveal different employees’ personas and roles in the innovation process.

“We have persona called a ‘discerner,’ which is someone who up votes a lot of ideas,” Gordon said. “They don’t necessarily create the good idea, but they can pick out the good idea.” And, of course, those who consistently earn up-votes for their ideas are labeled with the “innovator” persona.

“Right now this isn’t in the product, it’s just sort of a one-off thing that we’ve done with a couple of customers, and we’re thinking about including in the future,” Gordon said.

But in the meantime, it’s allowed some of those customers to learn a lot about where good ideas bubble up from. For example, Gordon noted, one organization found that the top innovator at the company was an employee working at one of their call centers.