Big Data

Facebook's Automated Insights wants to give you an AI-powered team of virtual data scientists

Facebook launched Automated Insights at F8 2017 as an ambitious new machine learning service that can alert businesses to anomalies in behavior and trends and provide omnichannel data to act on.

fb-auto-insights.jpg
Image: Facebook


Data scientist is the hottest job in tech, which means it's rough if you're a company trying to hire one. Still, a lot of businesses can't afford them, and even the companies that can usually find that data scientists end up spending way too much of their time cleaning data than doing useful analysis.

Those are the kinds of problems Facebook wants to tackle with its new Automated Insights service launched on Wednesday at the F8 developer conference.

The main idea is that Facebook wants to use deep learning to surface insights you didn't even know you should be looking for—and can help your business. And, this is especially aimed at commerce and shopping data.

SEE: Facebook Workplace could replace Slack, Jira, Hangouts, and a lot more

Facebook software engineer Luke Sernau said Facebook can accomplish this by throwing a ton of computing power at the overwhelming amount of data now coming into Facebook about your company.

"Using machine learning we can search deeper and more thoroughly than a human," said Sernau at F8.

Automated Insights is looking for two things:

  • Dimension-based insights (how certain users behave differently than the whole)
  • Trend-based insights (how patterns change and shift in your market)

The other problem Automated Insights wants to tackle is the omnichannel nature of today's shopping data—now that buyers have multiple ways to research products and interact with brands before making a purchase.

For example, Facebook data scientist Lisa Jones explained that a typical consumer might hear about a brand on Facebook—from a friend or an ad—then research the company online, download the company's mobile app to research more about its products and then walk into a store to make the purchase. In the analytics systems of the past, this behavior looked like four different users.

Facebook wants to connect the dots on that so that companies can better understand customer journeys. This is especially critical in an omnichannel world.

The old model of people interacting with your business, said Jones, was typically very 1:1, but now there are almost always multiple touch points that bring a customer to a brand and a purchase. Analytics systems have to adapt to this reality in order to provide a true picture of what motivates customers and how they are making decisions.

The challenge is that data is spread across all kinds of different places and systems, and so in order to capture it a system needs very broad reach. Facebook is one of the few companies with enough data and connections to people and companies to make it happen. The average business, said Sernau, now triggers over 1 million events to Facebook's data repository. That's what will help Automated Insights delve deep and make connections that would be easy to overlook.

Sernau said, "If that insight is out there, we'll find it."

Automated Insights has officially launched as an open beta. You can try it for free, and a Facebook login is not required. At F8, Facebook repackaged and rebranded all of its analytics products under the new umbrella of Facebook Analytics. You can go to analytics.facebook.com to try Automated Analytics and you can read more in Facebook's official post about it.

Also see

About Jason Hiner

Jason Hiner is Global Editor in Chief of TechRepublic and Global Long Form Editor of ZDNet. He's co-author of the book, Follow the Geeks.

Editor's Picks

Free Newsletters, In your Inbox