Omer Artun cannot predict the future. But he has a pretty good idea what it might look like.
Artun heads AgilOne, a cloud-based provider of predictive analytics for retail. He explained the basics of this craft at a late August launch party for his book, Predictive Marketing, co-written with AgilOne CMO Dominique Levin.
AgilOne is one of many firms that are surfing on the big data wave. But the approach differs, said Artun, in that AgilOne is relying on artificial intelligence to make a difference. The key is to take big data and use AI to “learn” from it, Artun explained. “It has to teach itself.”
“There is a lot of information in data, but a lot of it is junk,” Artun said. What’s important is to tune out spurious information that does not explain a buyer’s behavior, test correlations again, and seek out the data that is validated through repeated rechecking. That should build a predictive model that retailers will find useful.
“For every situation, you don’t need to predict customer behavior,” Artun said. He explained this by way of example. A retailer is trying to figure out how much foot traffic is passing through is store. The AI will note the weather and the stock market performance for a particular day. It will screen out the stock market closing figure, but correlate the weather to the store’s foot traffic. Why? If the weather is good, more people shop. If the weather is bad, fewer show up. The stock market may have closed down on busy days and up on nice days, so it gets screened out. But that information could correlate to boat sales. When the market closes up, stockholders may feel flush and are more likely to buy a boat. The AI may correlate these two factors while discarding others, Artun explained.
All this is happening against a backdrop of five trends.
- Consumers are more digitally connected than ever, thanks to social media and mobile computing.
- They are consuming information faster, so marketers have to pitch their brand experience in real time.
- Mobile applications are the bridge between on-line and off-line purchases (albeit this trend is still in its early stages).
- People are getting hit with information via many different channels; a predictive approach to marketing can cut through that clutter to make a connection.
- Consumer privacy is important.
Go too far in terms of initmacy and the regulatory backlash “will be bad for everybody,” Artun noted. Customers must be able to control how much information they give marketers, even having sufficient control to delete information if they change their minds.
AgilOne offers predictive marketing as a cloud-based service. The company is based in Sunnyvale, CA, employs more than 100 people. It is still privately held, having raised $41 million in venture capital from Sequoia, Tenaya, Mayfield and Next World Capital.