Software Predicts Which Companies Are an Easy Sell
A former Yahoo search engineer raises funds to bring sophisticated data mining and modeling to the business world.
A startup called Infer, led by a former Yahoo search engineer, plans to help salespeople identify potential business customers by gathering useful information from news sites and the Web. For example, marketing department job postings online might be one clue of a company’s readiness to buy marketing software.
Once the domain of high-tech hedge funds and companies like Google or Facebook, tools that make predictions by mining and analyzing large quantities of data are slowly trickling into the mainstream business world, says Infer CEO Vik Singh, who was recognized by MIT Technology Review in 2009 for his work at Yahoo (see “Innovators under 35: Vik Singh”).
Singh left Yahoo in 2009 to build Infer, and has operated it quietly since. His company is announcing $10 million in financing today from four venture capital firms, led by Redpoint Ventures and including the high-profile firm Andreessen Horowitz. By assessing a simple score to each customer lead that a sales department receives, the software attempts to give clearer direction about which efforts will have the most payoff.
This story is only available to subscribers.
Don’t settle for half the story.
Get paywall-free access to technology news for the here and now.
Subscribe now
Already a subscriber?
Sign in
You’ve read all your free stories.
MIT Technology Review provides an
intelligent and independent filter for the
flood of information about technology.
Subscribe now
Already a subscriber?
Sign in
The underlying technology builds on ideas Singh developed at Yahoo, where he created the company’s “build your own search service.” The program allowed developers to customize Yahoo’s search formula to their own needs and also incorporate their own private data sources.
Similarly, in predicting which potential customers that sales departments should woo, Infer uses machine-learning methods to quickly build a model based on a company’s individual sales data.
The software looks at a company’s internal sales databases, analyzing each entry and whether it led to a new paying customer, and then correlates each with at least 150 outside signals that may have had even some small bearing on whether a salesperson closed the deal. It can then look at these same signals about new leads to guess at whether they’re likely to become a customer.
The predictive factors that Infer pulls in about companies range from the obvious, such as news reports, SEC filings, and legal dockets, to the quite subtle. From job listing sites, for example, it looks at the hiring strategy, and social media activity can help it gauge how savvy its brand is. Infer also considers factors about the main contact in the sales database—say, a vice president or an IT manager—such as how active they are on Twitter. “We’re trying to crawl as much information as we can,” says Singh. The software then computes a score, from 1 to 100, that ranks where they should expend their energies. It’s displayed within existing software tools, such as Salesforce, for managing this kind of sales work.
Zack Urlocker, chief operating officer of the customer service software provider Zendesk, says that using Infer for the last year has significantly improved the productivity of his 60 salespeople, as measured by the percentage of deals they close from thousands of leads they receive each month from people who visit the website and sign up for free trials. Typical scoring tools use at most a dozen factors to evaluate such leads, and wouldn’t have improved over time by gleaning new insights from Zendesk’s continued growth, he says.
Already profitable, Infer only serves companies that sell to other businesses, not consumers—though Singh says that, by mining and buying information about people, the same general concepts could help predict whether someone might be amenable to a sales pitch.
