NOTE: This post is intended towards more advanced media buyers and advertisers. This is a case study in general.
This week I did some deep research about auction types and one of the most fascinating and complicated one was VCG auction.
Why was I researching on this?
To learn some ins and outs of how facebook’s bidding strategy has been implemented so we can be more efficient at our media buying campaigns.
It is kinda fascinating because Facebook’s advertising chief has talked about this openly that this is the strategy they are using and also proudly says this is something that cannot be “gamed.”
Personally I believe that everything is hackable. However, the way facebook has implemented this is freaking genius in regards to how it works in long run.
See they want advertisers to keep advertising on their platform, that’s how they will win the long game, and the way they are doing it is by forcing them to put highly relevant ads, and they are using VCG auction strategy to make sure everybody only pays the price that desplaced the perceived value of the next highest bidder.
And they also want users to “stay on facebook.”
Which brings us back to the topic of this post – RELEVANT ADS
The way facebook has implemented this is kinda awesome for their users and advertisers who aren’t shady.
If you actively advertise on facebook, then you can see the relevance score of your ad varies as time goes by. If it was 7 today, it would be 5 tomorrow and 4 after that etc.
Why is that?
I think that relevance score has nothing to do with the “user” himself but what’s in his newsfeed at a particular time.
This news feed contains the content from the pages he “LIKED” as well as his interests based on websites that he is checking out etc.
It might be the case that facebook has an interest profile for each user based on these facts. A privacy advocate’s nightmare though…
How much relevant your ad is, at a given TIME in relation to the user’s newsfeed. This could be why we see variations in relevance scores of ads as time goes by.
The relevance score is how your ad fits in a user’s (custom) newsfeed at a particular time. This news feed might exclude content that are trending and facebook is able to figure out what’s trending and what a large number of people are talking about actively and part it out from the “custom newsfeed.”
They actually launched a dedicated tool for journalists to monitor what’s going on and trending so I’m sure they have this pretty nailed down.
“Custom interest feed” is the feed generated by the websites he like, pages he likes, the type of ads he is responding to, websites he is visiting actively and the type of people that he follows.
This is a combination of off the platform and on the platform user activity. (side note: Recently facebook was rewarding users who post videos on facebook along with video advertisers, why? Because they can’t track the user’s activity and add it to their user’s ad profile when the user is watching or consuming a certain kind of content on YouTube. Nah ah, the big G won’t allow that.)
This ensures that user doesn’t see ads that are out of place for him, and makes his experience more linear and stable with Facebook.
Advice: As far as bidding strategy goes, have your KPI in order and figure out how much a lead from facebook is really worth to you and then bid how much you’re willing to pay based on that.
It’s complicated stuff, but this is a brilliantly monetized platform, damn.