How to design Google/FB Ads system?
Jun 21, 2020
6 Comments
I would like to know how to design an Ads system like Google and Facebook Ads engine, especially how AI/ML can be incorporated to recommend Ads accurately to the target audience.
For example, when I search for a specific product on Amazon, I get the same product as a recommendation on Instagram and/or FB. I think there is no AI/ML required for this one, but interested to know how these platforms interact with each other and how these things have been monetized.
Any pointers to resource/blog would be helpful.
Google search/Youtube didn't fetch me what I was looking for.
#engineering #software #swe #google #facebook #fb
comments
From my limited understanding ads as a service is mainly composed of:
- ads marketplace - platform that allows advertisers to make bids as buyers, and g/f/etc are the sellers
- user profile - this is honestly already very complex, all these ad companies build a profile of you through various identity matching, browser history, click-link-frequency, etc.
- relevance engines - a whole variety of services that tl;dr take your user profile and somehow figure out what they should be showing you
User profiles feed into these relevancy engines that then gets fed to the marketplace and advertisers can decide whether or not to buy an ad for that profile
The relevancy engines are where it gets more interesting as these things run both nonstop in the background, and also on request. Basically you do a bit of processing as a user is browsing to prepare to match them to relevant targets and then when they user actually requests an ad (in the sense that htey open fb and fb requests an ad to serve them) their relevant targets go through more lightweight post processing and get matched within marketplace and that's how you see the final ad
A lot of oversimplification here haha but hope that helps. Would love to hear from more ads experts. I'm no longer in the field but thought it was fascinating when I was there