If I tweet my feelings about an artificial sweetener, Coca-Cola wants to know about it. Am I discussing new scientific findings about sweeteners? Praising the taste of one while maligning another? Talking about how it’s helped me lose weight? Marketers typically pay big money for research into what people are thinking about–to gauge success, identify threats, ferret out misinformation, and pick up on themes that resonate with consumers.
These days, ample clues to the future direction that products should take are hidden in the fields and streams of the Web. “Brands don’t become great by monitoring the past,” says Stan Sthanunathan, vice president of marketing strategy and insights for the Atlanta-based Coca-Cola Company. “The challenge is to have a point of view on the future.” Tens of millions of new blog entries, status updates, and tweets appear online every day–so information is out there for the taking, on everything from tech product launches to new movies to soft drinks to brands of soap. “Consumers know what they want and are giving their opinions in an unconstrained fashion,” he says.
The trouble is that there’s far too much data for any team of human readers to get through. “Very often, you end up boiling the ocean,” Sthanunathan says. Machines can help, but peering into human emotion is a more complicated task than traditional search and analytics. After all, feelings often come in the form of abbreviated, slang-ridden tweets that computers cannot translate into anything meaningful. Marketers today aren’t mining simply for information on click-throughs and page views–they want to mine the secrets of the human heart and come up with hard data on soft concepts such as “mood” and “passion.”
A new crop of companies say they can deliver this to brands. Jonathan Spier, CEO and cofounder of Netbase, a startup based in Mountain View, California, bills his company’s technology as able to “read and understand the English language.” He means that to include emotion and nuance.
Coca-Cola has been testing Netbase’s platform at its corporate headquarters since September, using it to watch the response to a new advertising campaign–and to monitor the discussion around artificial sweeteners. Sthanunathan says Coca-Cola plans to roll out the platform within the company globally starting in November. Netbase has impressed him, he says, with “its ability to understand the context as opposed to just the content.”
Spier explains how Netbase works by citing a sample tweet: “The iPhone has never been this good.” Some systems that purport to tease out users’ moods search for keywords, such as “iPhone,” “never,” and “good.” Most of these would be fooled by the sample sentence, interpreting its sentiment as negative because of the word “never.” The simple word “this,” however, alters not only the mood of the sentence but also its intensity. “The iPhone has never been good” is a negative statement, Spier explains, while “The iPhone has never been this good” is strongly positive.
Although Netbase’s consumer marketing and research product is new, it’s based on R&D that goes back to 2004. The company has eight patents pending, some from its early work building a platform to identify and understand patterns in English sentences. More recently, Netbase has added statistical learning capabilities that allow the system to improve as it encounters more and more text.
It took time for Netbase to find its current focus. Its first products were tools that allowed experts in a field to perform fine-grained searches within specific subject areas. NetBase has a major partnership with publisher Reed Elsevier, for example, to crawl and categorize its archives of journal articles in order to make them easy to search from a variety of angles.
The new consumer marketing product was born at the request of consumer products giant Procter & Gamble, which already used Netbase for its scientists. In late 2008, the head of P&>’s market research division asked if the company could adapt its platform to allow fine-grained searches of the data being posted about brands online. Netbase complied, working with an advisory council of customers, including P&> and Coca-Cola, to factor in all the ways the companies wanted to be able to slice and dice their data.
Today, Spier says, consumer marketing has become Netbase’s leading product. The company has acquired 50 new customers in the past 90 days, he says, and is processing more than 50,000 sentences a minute.
Other companies working on tools for mood mining include Viralheat, a startup based in San Jose, California, and Jodange, based in Yonkers, New York.
As the tools proliferate, the companies developing them must make them part of a broader package if they are to appeal to businesses, says Ed Chi, area manager for the Palo Alto Research Center’s Augmented Social Cognition team. Chi’s group has been researching the technologies that businesses need to analyze and respond to social media effectively, including tools that classify topics being discussed online, perform sentiment analysis, suggest possible responses, and analyze a company’s message to determine how likely it is to go viral.
Spier believes companies can satisfy their customers by putting data in context. “What’s needed are systems that can view the world through many lenses, with the kind of accuracy businesses really need to make decisions,” he says.
Netbase’s dashboard provides context by automatically surfacing terms that make for interesting comparisons. For example, Spier says, it may not be all that interesting to know that response to the iPhone 4 is positive. More interesting is whether it’s positive when compared with attitudes toward the Motorola Droid, the HTC Evo, or the Samsung Epic.
Netbase lets users choose what terms to compare as well as offering automatic suggestions. And its metrics go beyond simple positive and negative. For example, its passion index tries to determine how powerful sentiments are for consumers. Spier says that Netbase has found that a product will sometimes pop out on the passion index while it gets lost on simpler measures. What’s more, he says that the loyal, engaged responses that indicate passion do not typically correlate with the amount of buzz around a product.
Since these tools are so new, it remains to be seen whether any approach will provide the vision that brands are looking for–but these startups are betting that mood mining is here to stay.
“For 40 years now, the tech industry has been digitizing everything in sight,” Spier says. “The next 40 years, businesses will be focused on how to make sense of all that information.”
The dark secret behind those cute AI-generated animal images
Google Brain has revealed its own image-making AI, called Imagen. But don't expect to see anything that isn't wholesome.
The hype around DeepMind’s new AI model misses what’s actually cool about it
Some worry that the chatter about these tools is doing the whole field a disservice.
The walls are closing in on Clearview AI
The controversial face recognition company was just fined $10 million for scraping UK faces from the web. That might not be the end of it.
This horse-riding astronaut is a milestone in AI’s journey to make sense of the world
OpenAI’s latest picture-making AI is amazing—but raises questions about what we mean by intelligence.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.