Michael Jones

Thanks

When I wrote "My quiet-leaf-blower company can find out: Who likes quiet leaf blowers?" I meant that precisely, my company can find out. One way is with an app using API query and traversal and one way is with a campaign if 2^n micro-targeted ads.

... There are more than 29,000 categories including "total liquid investible assets $1-$24,999" or "Individuals that are frequent transactor at lower cost department or dollar stores. By having an ad for each choice (2^N ads) then as soon as they click the one they are served...

...I don't mean thousands of choices, of course, just ten say and then 1024 versions of the ad with yes/no targeting for 10 categories. Each click tells an advertiser ten facts about age, health, gender, income, credit, or any of the thousands of categories. Thus it leaks.

But that click happens at the <1% level, and is the same deal for Google. Sorry, you have no idea how either the FB or Google ads systems work, and I say this as FB's first ads targeting PM, and as the person who copied Google's ad exchange to build a Facebook copy (FBX).

Let me just highlight some of the things you got wrong. Facebook doesn't expose user PII, or even much in the way of FB targeting data to advertisers. In fact, it's the total *opposite*. Facebook struggles to *ingest* outside data via the Custom Audiences tool.

You get the directionality of data flow precisely *wrong*. FB's targeting data (for DR advertisers) isn't terribly valuable. FB doesn't sell data, it *buys* it by convincing advertisers to match data with FB, and allow FB to optimize over that matched ad delivery.