New York-based technology company Eyeview raised $20 million in combined equity and debt from existing and new investors to build out its dynamic video ad-targeting business to support brands such as Bayer, Caesars, Hyundai, Honda, Lowe’s, and Uber, according to the company.

The company’s video targeting platform dynamically inserts  personalized messages with help from data seeped in machine learning and AI, along with about $70 million in total funding from Qumra Capital, Innovation Endeavors, Nauta Capital, Gemini Israel Ventures and Israel Secondary Fund.

In fact the “creative personalized” platform relies on hundreds of thousands of attributes per campaign to support ads, mostly pre-roll, running in real time on mobile, desktop, connected and addressable TV per campaign.

Eyeview CEO and co-founder Oren Harnevo said it began after brands began asking to send all different types of data. Eyeview would factor the data into the ad targeting, so marketers from the brands could learn more about their customers.

The platform is built so each campaign receives performance feedback, making it easier to tell, for example, if women between the ages of 25 to 40 are more interested in safety features for a car.

Personalized data gets factored into the video ads and measured to help Eyeview’s platform learn the engagement rates for each piece of information.

Internal data tells Eyeview that local content in the ad — information about where to go to find a product — typically gets about 20% lift on average, compared with ads that do not offer that info.

When retargeting, the lift grows much higher. Say a consumer searches for and then views a shoe on Zappos. She doesn’t purchase the shoes, but closes the browser to leave the site. A retargeted video ad with information on where to buy the same shoe viewed on Zappos’s website would produce about 250% lift, according to Harnevo.

Harnevo said the company, with about 130 employees, runs more than 100 campaigns per quarter, and some 50,000 ads per campaign. “We see about 20 billion impressions per day,” he said.  

About 45% of the company’s research and development team works in research and development. About a dozen are data scientists. Others have been working to tag videos so the platform can textually understand the content. Harnevo calls this part of campaign optimization.

“We start with 50,000 ads per campaign, but we work on not targeting the same message to all,” he said. “The campaigns may start at a 2:1 ROI, but as it matures the lift changes with help from data.”