YouTube Shows Shopping Ads Based On Video Content

Karthick Prabu
Product Locus
Published in
2 min readJul 13, 2017

Best use of context + machine learning to monetise YouTube

YouTube has been monetising the platform by various layout / delivery combinations of image ads and video ads, both have been greatly successful.

The recent experiment by YouTube is to show a full-width ad card loaded with three products based on the video content that’s being played.

I play YouTube videos almost every day, certainly I’m a power user of YouTube, but till date I have seen this new ad layout only once. I’m not able to see this ad layout again.

When I was watching a video on Artificial Intelligence, a nicely animated card opened up to show books from Amazon that are about Artificial Intelligence.

Clicking on each of the book took me to Amazon.

I believe this is a first step (MVP) implementation of machine learning driven contextual advertisement by YouTube. Currently, it is able to show relevant product ads based on the video title.

In the near future, we shouldn’t be surprised if YouTube could understand the semantic of the video being played and shows direct or indirectly relevant products to purchase in ad slot (provided there is an advertiser for the brand).

For example:

  1. In a MKBHD’s video review comparison of iPhone 7 Plus and Samsung S8, Samsung could decide to show its S8 phone and its accessories in the ad slot.
  2. A video by Travel + Leisure shows top 5 luxury hotels to stay in Switzerland. YouTube has tremendous revenue potential in this case. Because, there is more than one product (brand) to showcase to the user to click to view/purchase. Any of the five luxury hotels in the video could advertise or Switzerland Tourism board could advertise or Swiss Air could do. The highest bidder gets the ad slot.

With the advent of products like Google Lens and machine learning advancements by behemoths like Facebook, Amazon and Apple, seeing the examples in this article in real-life is very near. That day will come soon. Very soon.

Godspeed.

--

--