Data platform Quantifind’s new Analysis application lets marketers ask their own questions
Menlo Park, California-based firm has used similar approaches to discover which social conversations connect to terrorism.
As it turns out, finding out which conversations on social media are connected to terrorism is not that different from finding out which conversations lead to a sale.
Menlo Park, California-based Quantifind, founded in 2009, was involved at one point in the former task, but now it is solely devoted to the latter.
But, unlike terrorists, marketers like to ask questions. In order to give marketers more control over getting insights from conversations that predict their business outcomes, the company released this week an Analysis application for its existing SIGNUM platform.
Previously, co-founder and president Ari Tuchman told me, a brand like a car company might ask Quantifind to search for clues in customer conversations and feedback indicating, say, what kinds of advertising approaches are most effective.
Quantifind would create the analytical model, onboard the data — including zillions of Twitter posts, Tumblr blogs, forum comments, customer surveys, call center transcripts and more — and then run the analysis against point-of-sale records or other results relating to the desired outcomes. The company likens this to conducting “the world’s largest unaided focus group.”
The results might indicate, for instance, that potential customers respond better to ads showing the interior of a new car.
But the marketer might then want to drill down, asking for details about, say, whether this means a complete inside tour of the car or just some images, whether this applies across all demographics and so on.
In the previous incarnation of the platform, Quantifind had to conduct the drill-down follow-up questions as well, which Tuchman said slowed down and limited marketers’ inquiries. Now, with the Analysis release, marketers can essentially conduct their own drill-downs after the initial analyses by selecting and comparing an assortment of the most frequently found terms.
Beyond sentiment analysis
“Once we’ve built the model and shown how to use the product, they can use it themselves,” he told me. A beta phase of Analysis over the last several months has involved half a dozen brands, including Pepsi and Heineken, although Quantifind said it does not yet have any data on how effective the new tool is. The company has about two dozen customers for its platform.
Tuchman said that his company goes beyond sentiment analysis, in that knowing how many positive comments are posted about a product doesn’t necessarily tell a brand what factors helped drive a sale.
He recalled that the Disney Company approached Quantifind in 2011 after the moviemaker discovered most of their films “had similar sentiment metrics on social media,” whether they were hits or flops. In other words, there was no consistent relationship between the perceived sentiment and engagement of analyzed social conversations and ticket sales.
Quantifind, by contrast, specializes in finding and analyzing the conversations that are predictive of the desired metrics.
For instance, it might be the case that social posts with some variation of “need to find a baby sitter” could indicate the commenter wants to see a movie they’ve heard good things about. Just as, it turns out, there might be certain phrases across some social media channels that correspond to terrorist activity.
While there is no shortage of data analysis tools, Tuchman said his company’s platform is different from, say, social analytics tool Socialbakers or big data analysis platform Palantir, in that Quantifind finds and focuses on the conversations that matter to your key performance indicators.
“You’re looking at a million conversations,” he said, but Quantifind “is the only one [that automatically finds] the 200,000 you want to pay attention to.”
About The Author
Branch launches cross-channel cohort analysis for app marketers
Teach yourself about machine learning and artificial intelligence
GDPR: A 10-step action plan
Budget allocation question: Is there such a thing as too much martech?