NowDiscover Uses Machine Learning in Video Recommendation Engine
Berlin-based startup NowDiscover has launched a video content recommendation engine for enterprises. It blends algorithms able to capture explicit and implicit purchasing intent signals, advanced filtering and machine learning and then hands the findings off to a human curator.
In short, NowDiscover‘s engine has just evaluated and ranked hundreds of consumer-created YouTube videos for appropriate content and presented the best options to the corporate user for review and ultimately selection.
In real life, that process would look something like this: A global brand name cosmetics company decides its wants to augment its own marketing with YouTube videos of real people using its products. Real people expertly applying eye shadow, blush, mascara and — that bane of almost all wearers of makeup — liquid eyeliner.
Real people earnestly demonstrating just how beautiful the product can make you look. Real people who do not use a competitor’s brand or has an obnoxious dog barking in the background or possibly offensive material hanging on the wall.
YouTube Video Stars
YouTube, by the way, is where CoverGirl found its latest star and its first CoverBoy: 17-year-old social media star James Charles, who rose to fame with his videos showing men and women how to apply makeup.
CoverGirl is not the company in the pilot, company founder Vishal Kawatra told CMSWire. Just about every cosmetic company, from global brand to local start up, has turned to YouTube tutorials for promotion, probably because their use case in the customer journey is so obvious and simple.
Taking the Use Case to Other Industries
What Kawatra and other video recommendation engine providers are counting on is that this use case will spread to other sectors and industries where the tutorial model isn’t so obvious for marketers.
Kawatra uses Apple as an example. A customer is planning to purchase a new smartphone and the new Apple is on his shortlist. This person is new to the Apple brand and wants to know how easy it would be to access apps that he uses for work. He also wants to know if the phone can serve as a mobile hotspot — and if so, how well does it work — as well as a few other issues.
There are YouTube videos from actual users that can answer all of those specific points, Kawatra said. By displaying them to this prospective customer in advance on the Apple website Apple has shortened the customer journey plus added some authenticity to the brand’s marketing.
A New Approach
“To date, most recommendation engines have been based around products or services and use approaches that are rudimentary such as collaborative filtering or content-based filtering,” Kawatra said. Approaches, in other words, that try to match individuals with like-minded individuals and their choices.
“With the maturity of big data, we can now take more of a cognitive approach and align explicit and implicit buying signals from consumers and use them in more precise models,” he said.
“That is our use case: recommending content based on an explicit signal of purchase intent.”
The platform shortens the purchase decision by placing the YouTube videos on the company’s branded site — YouTube videos that the person would have likely found when researching the product.
“The goal is to inject video content to whoever needs it at the exact time they need it.”