AdTech is hard.

I worked for Facebook’s Audience Network team from April 2014 to August 2016. I had zero experience with ads or AdTech and all I know I learned “on the job” as a member of the Marketing Science team. This write-up will have some inaccuracies and little AdTech speak.

Facebook is good at two things:

• Scaling — the sheer size of the Facebook network almost guarantees success (when the expectations are somewhat reasonable). There has been little innovation from Facebook in the past years but they’re really good at extracting what is good of another product, testing it, iterating and then launching it as a small feature in one of their core products. Good examples of these “products” are Payments within Messenger (vs. re-building Venmo/Square) and more recently Stories on Instagram (vs. a new version of Snapchat).
• Measurement — the ubiquitousness of Facebook Login and how we are logged into Facebook pretty much everywhere make the ‘measurement’ game the place where Facebook thrives. Measuring conversions and funnels is really hard in an actionable way. It’s easy(ish/er) to know that out of X impressions then there were Y clicks. It’s much more powerful and difficult to know that person A saw X on desktop but clicked on it on mobile web and then purchased something on the iOS app. Facebook (and Google) are getting scarily good at this and not only in advertising.

These strengths matter because when you combine scale and measurement is how Facebook and Google can succeed and crush the competition with less effort and little risk.

An ad network is essentially the connector between those that pay for ads and people that show ads. Audience Network is Facebook’s mobile ad network. It is an auction based ad network that targets users that are using apps or websites off of Facebook based on their Facebook account. Facebook does not specifically choose what ads to show you, neither where or when to show these ads; a model does. It expands the options of where advertisers can show ads without having to find inventory themselves and it provides publishers and developers an option to monetize their product without having to find the advertisers. Facebook takes a cut of the revenue these publishers generate.

Audience Network launched in 2014 only on mobile apps and expanded to mobile web in early 2016. The problems faced early on for both mobile apps and mobile web were all on the measurement side of the products. These problems are inherent to the world of AdTech: how do we guarantee that real human beings see ads at the right time and with minimal fraud?

AdTech is a middleman business. It’s trying to balance people that want to spend as little as possible with people that want to make as much money as possible. It’s hard because the middle man is the one that has to decide what is value and how to price and measure that value in a way that pleases both sides. Most AdTech companies are shady for the simple reason that their value proposition has been to make the most money possible for as long as advertisers will pay them and publishers will use them. This stance has made impressions (CPM), brand advertising (CPM) and even clicks (CPC) the gold standard for measuring performance.

An AdTech company will promise 10K impressions and charge $100 ($1 CPM) but will pay publishers to deliver those 10K impressions $50 ($0.5 RPM) and bank those $50 for themselves. The $50 is set with little measuring of actual performance of 10K impressions (did people click? convert?) and with little guarantee that they were even seen by a real human being. This world described is an oversimplified version just to highlight the idea that middle men have all the power and their only incentive to fight for fair pricing is competition with other networks that deliver the same ads for cheaper and with more “effective” fraud checks. Essentially, checking that those $100 deliver $100 worth of value.

Facebook and Google with their identity machines have had the luxury to question this proposition and go beyond maximizing revenue as a goal and started introducing the concept of “value”. Facebook introduced a metric called Advertiser Outcome Score to let publishers (developers) know how their ads were doing. This metric has its flaws (I was part of the team helping implement it and understand it) but it signaled a shift in AdTech that was unprecedented. It tells publishers that impressions are worthless, clicks are worthless and if you don’t start worrying about driving installs or purchases (down-the-funnel measurement) then you won’t make as much money. And guess what? It was starting to work. I saw the shift in conversations with publishers and advertisers and the requests stopped being about helping to optimize for CPM and started being about ROAS (return on advertising spend).

Facebook is counting on winning the AdTech space or at least make it a two player game against Google. The narrative around value is not going to go away and slowly but surely the big advertisers and publishers (the ones with real $$$) will demand proof of value. There are some small players that will keep on making money in the short term but the long term will be value-driven. Advertisers will demand proof that a dollar spent is a dollar of returned value and publishers will be paid only for real value added.

This shift is important because a world where Google and Facebook define value to advertisers it inherently means defining value to their users. Do I really want a world where the value of the services I use are determined by those who pay and get paid by Google and Facebook? No, thank you.