Is this data scientist a consumer’s best friend?

In Oren Etzioni’s world, telling you where to buy a product is so 20 years ago. He did that with his first startup, Netbot, in 1996. Today, Etzioni wants to tell you when to buy — that ideal moment when the price won’t fall for a while and you won’t get burned by the release of a new model a week later. Tomorrow, well, maybe he can let you know when you’re in the vicinity of a great deal.

Etzioni, who spends his days as a computer science professor at the University of Washington, is probably best known as the co-founder of Farecast. That company, which Microsoft (s msft) bought for $115 million in 2008 and incorporated into Bing, helped even the playing field between travelers and airlines by predicting the best times for travelers to purchase their tickets. Airlines practice yield management by regularly changing their prices to maximize profit, but Etzioni was able to find a fair amount of predictability once Farecast was able to get the data it needed from the airline industry.

Lately, though, Etzioni has been focused on his latest startup,, which launched in 2010 and applies the premise behind Farecast to consumer electronics and appliances. At some point, Etzioni told me, the practice of yield management had migrated to “pretty much every non-trivial good.” Consumer electronics was the natural place to start, he said, because they tend to be highly considered purchases and enough of them are made online to generate lots of data and justify the existence of a web application.

I explained Decide’s methodology in some detail recently, but the gist is that it analyze thousands of data sources — from pricing history to model history to rumors of new models — to tell consumers whether to buy or wait on their electronics purchases. It’s similar to some degree with Retrevo (see disclosure), but Decide focuses more on predicting price than on telling consumers what’s a good value now.

There’s a notable difference between predicting airlines prices and predicting consumer-goods prices, though — while Farecast had to fight to get airlines to release their data, there’s an overload of it on the web. Between news sites, blogs, product profiles and price-comparison sites, Decide has all the data it needs. “The hardest is still separating the wheat from the chaff,” Etzioni said.

Decide’s prediction models have become so accurate, though — about 77 percent — that the company recently decided to put its money where it’s mouth is. Its new Got Your Back program provides members with a number of deals each day, and if a member buys and the price drops, Decide will make up the difference.

Aside from expanding price prediction into other areas with highly volatile prices, such as hotel rooms, Etzioni appears to have his eyes set on the mobile world.¬†There is, of course, location data, which is the holy grail of mobile analytics. Etzioni thinks it will be about three years before location-based analytics really reaches its stride in terms of usage, but it’s coming. Being able to get signals as to where app users are at any given time is a powerful proposition in terms of personalization, and already there are stealth-mode startups such as Placed trying to solve the problem for developers.

Etzioni also thinks there’s work to be done in condensing big data to fit onto small screens. Companies trying to improve the consumer experience will keep incorporating new analytic techniques, such as sentiment analysis, into their products, but the output of those efforts has to go somewhere. Producing and analyzing big data is one thing, he said, but you need to summarize it in a manner that people can consume on their mobile devices.

Disclosure: Retrevo is backed by Alloy Ventures, a venture capital firm that is an investor in the parent company of this blog.