Affine Rebrands to SET, Uses Machine Learning to Help Place Ads in Online Video

Today, online video ad targeting company Affine announced new branding, as it becomes SET. Accompanying the branding change is a new classification technology for online video the company is debuting today, which is designed to make it easy for advertisers to buy and sell online video ads by category easily and at high volumes, in the same way they do for TV.

The problem with online video ads up until now, according to SET VP of Product Matt Tillman, is that it’s been hard to know exactly what’s in a video in a reliable and consistent way; unlike web pages, where Google can easily index all the necessary content and provide a very good idea of what kind of site and what kinds of ads should appear there, videos have been relatively opaque, at least without manual viewing and categorization.

“When you’re buying display, it’s somewhat easy, because Google goes around and indexes a bunch of pages,” Tillman said. “With video it’s a lot harder. With video the unit of inventory is the video itself, so in order to give advertisers control over what they’re sponsoring, you have to do a lot of work. And it’s actual technology, it’s not just starting an ad server, hiring a sales team and a dev team and going to town.”

Instead, to analyze the content of video, Tillman says companies have to spend significantly in engineering in order to parse millions of videos efficiently and quickly. That’s what SET’s done, using a combination of image and motion recognition, machine learning and natural language processing to come up with a way to sort video automatically into various vertical buckets, quickly and at scale.

That should help everyone involved come out better. For publishers, it means that they don’t have to go around individually pitching and courting brands directly, and for advertisers, it means not having to make sure that content is a good fit before setting up a campaign; the risks inherent in volume purchasing of video ad inventory are lessened considerably, since SET can ensure that family friendly brands won’t be advertised in any kind of adult-oriented videos, as just one simple example.

SET’s tech can get much more specific, including targeting basketball and baseball, but not soccer or cricket, for example. And Tillman noted that one of the most important things, from a brand perspective, is that its tech actually allows them to drop ads into video inventory from their competition, putting their products right in front of the audiences they want to attract.

Late last year, one study from and Digiday suggested that video ad spending would increase sharply this year, with a projected 47 percent increase in online video budgets among those who are already buying. While Tillman admitted that SET, which charges based on a CPM model, does indeed ask for a fairly high rate because of what it sees as unique positioning, the appetite appears to be there to support continued growth.

0 replies on “Affine Rebrands to SET, Uses Machine Learning to Help Place Ads in Online Video”