IBM’s Watson platform has another new side project: making holiday gift predictions. The company is releasing a new iPhone app called IBM Watson Trend, which they claim can predict which gifts are most likely to sell out at stores and e-commerce sites. While aimed at helping consumers, the app also shows the investment IBM has been putting into expanding Watson into the business world.
The app shows “trending” gifts in three categories–toys, consumer electronics, and health and fitness–alongside explanatory information about why those particular objects are trending. This includes analysis of hot purchases such as Lego City and the Nikon D-SLR, as well as products like Mattel’s “Hello Barbie,” which are declining in popularity (IBM says that, although popular with parents, children feel the interactive doll has an outdated style).
Watson Trend mines data from Twitter, which IBM has a formal partnership with, alongside ratings and reviews from more than 200 e-commerce websites, and thousands of blogs. The app then applies data mining and natural language processing techniques to approximately 10,000 sites and services in total to make its analysis.
According to Justin Norwood, a product strategist at IBM Commerce and a primary force behind the app, Watson Trend was his pitch in a Shark Tank-style internal competition for new ideas to bring Watson into the commerce space. Peers upvoted his idea, and IBM executives decided to fund the project. The whole process, from approval to deployment, took approximately 90 days.
“My thought process was that people make purchase-decision advice via social media,” Norwood tells Fast Company. “We have this great Watson service at IBM that understands natural language. We then decided to harness the power of collective conversation, of understanding what people like and dislike, and to bring the collective wisdom of millions of people to individual shoppers using Watson’s cognitive power.”
But there’s a reason IBM was crowdsourcing ideas on how to deploy Watson in the commerce world from employees. After being targeted at health care and tooled for financial purposes, IBM sees e-commerce as a growth area for the big data platform. IBM, which has a long competency in developing and selling complicated, sophisticated products to large corporate customers, thinks Watson holds the potential to help retailers make sense of the massive amounts of data they store.
This past May, IBM announced they were investing in two companies called Sellpoints and WayBlazer. The purpose of the investment, IBM said, was to develop new Watson applications for the e-commerce and travel sectors. One of the major problems IBM faces is that they have a distinct corporate culture that is great at technological development, but less great at understanding the needs of different industry sectors. This is a big reason why they have teamed up with third parties such as Apple, and are investing in outside companies in order to come up with new use cases for Watson. IBM’s own marketing cloud platform, in fact, was built through the acquisition of outside firm Silverpop.
It’s also no accident that the Watson Trend app is aimed at consumers rather than retailers. Norwood says the app is designed for a mass audience, but that IBM plans further uses of Watson for the commerce sector. In the first half of 2016, the company says, it plans to offer much more detailed insights to both retailers and consumer product manufacturers.
Commerce data–of both the brick-and-mortar and e-commerce varieties–is also something that’s of major interest to IBM and other large companies. Amazon has pioneered the act of turning e-commerce into a massive source for clues that can be deciphered and retooled to sell consumers more products. The switch from magnetic credit cards to chip-and-pin credit cards is also leading retailers to overhaul their entire IT systems, so apps like Target’s Cartwheel are generating reams of customer data, and iPhone-friendly beacons placed in main chain stores let corporate offices understand just who exactly is shopping there.
For IBM, Watson Trend is an easy way to show off to consumers what they later want to sell to retailers. Their hope, it seems, is that by giving customers a peek at how they can sift through endless product reviews and Twitter messages to get a better idea about product availability, they can also sell retailers on fancier big-data products months and years down the road.