Twitter data + weather data = better customer insights, IBM finds

A partnership between IBM and Twitter that might have seemed strange five years ago now seems to make perfect sense.

IBM has just announced new cloud-based services for analyzing Twitter. These tools promise to combine Big Blue’s expertise with massive sets of unstructured data — plus its decades of experience serving big organizations — with Twitter’s firehose of real-time, unstructured (and sometimes nonsensical) tweets.

The result, ideally, will be insights that enterprise marketing departments can use to better hone their approaches to the market.

“This is going to be an important business for IBM, especially around combining Twitter data with other types of data and also enterprise data,” said Dave Schubmehl, a research director at IDC.

But it will be good for Twitter too, said Charles King, the principal analyst at Pund-IT.

“IBM carries weight with enterprise customers that other vendors only dream of. This deal and the seriousness of the collaboration should provide Twitter numerous large-scale prospects,” King said.

IBM will not only analyze the Twitter data, but will also combine it with a wealth of data from outside in the “real” (aka non-Twitter) world, “such as weather forecasts, sales information, and product inventory stats,” IBM wrote.

The goal: “To uncover powerful correlations that drive more actionable insights,” the press release stated.

IBM said that more than 4,000 of its employees are now trained on working with Twitter data, and can help customers with implementation.

Additionally, more than 100 customers are already working with the service through “early engagements” — presumably not high-dollar, long-term contracts in most cases. However, those customers have already implemented the solution enough to deliver some “actionable insights.”

Some of the kinds of insights IBM has already helped its customers discover include, for three sample industries:

• Telcos: Correlations between nasty weather, angry tweets, and cable or wireless customers canceling their subscriptions. In some cases, that insight helped reduce customer churn by as much as 5 percent.
• Retail: Correlations between employee turnover and customer turnover. When your most valuable customers have a relationship with an employee they trust — which is often the case — they’re more likely to defect if that employee leaves.
• Fashion: Why some products sell and others don’t. IBM used psycholinguistic analysis of tweets, combined with sales and market share, to help figure out why those chartreuse mohair sweaters just aren’t flying off the shelves this season.

IBM’s offering competes with a host of other social analytics tools, including offerings from Adobe, Salesforce, and Oracle, to some extent. Other companies exist that integrate public data (like weather and market data) with enterprise data, including Via, ClearStory, and Treasure Data.

Its most direct competitor, said Schubmehl, is DataSift.

“What is different here is that IBM is using Twitter to provide insights around topics like churn analysis or talent management, which is not something that most social listening tools are doing,” Schubmehl added.

“Correlating Twitter data to other enterprise data is probably at least 5 years old. IBM’s behind,” noted Jon Cifuentes, an analyst for VB Insight, VentureBeat’s research division. “They do, though, have the advantage of 400k+ employees on which to conduct these data experiments.
“It will particularly interesting to see how they apply this to talent management and the work they do with Smarter Workforce. They have a clear advantage there,” Cifuentes said.