Hacker News Reader

Comment by puranjay | Hacker News Reader

Is there any independent audit of any advertising platform's data at all? Like Facebook tells me that 10,000 people saw my ads and I have no real way of verifying that. And I'm supposed to believe that Facebook is being honest even though Facebook has all the incentive to falsify numbers

View on HN

Game companies that buy ads for mobile installs-- a huge advertising market-- pay per install. Of course fraudulent companies create fake installs from fake accounts. But when you get an app installed a thousand times it is possible to judge the quality of those installs. Where they are, if they are real people, etc. It is clear that the results vary wildly and from my subjective perspective ad fraud is at least as bad as it has ever been.

Yeah, I always find it bizarre when people make these naive claims. Of course number of impressions are a proxy for reality, but when the rubber hits the road advertisers know what value they are getting. They are leveraging the microtargeting and the cookie tracking an God knows what else zuck has cooked up. Facebook, Google etc are fundamentally profitable places to advertise -when done intelligently. Smart, repeat advertisers pay Facebook a lot of money for this privilege.

Arguments like this always strike me the same way as the arguments about real-estate around 2007: “Smart bankers would never invest billions in bad loans”.

Maybe advertisers know what they’re getting. Or maybe they think they know but are actually willfully ignorant of something important because it goes against “common knowledge”.

Without some kind of third party auditing it’s impossible to tell, and I haven’t seen anything like that from Facebook.

The problem is you cannot audit as much as you would like. The important part is a real human sees the ad. (human with the desirable characteristic is a secondary factor). However you cannot measure the effects of an ad easily because there are too many confounding factors: people who buy because of a previous ad, people who would buy anyway, people who will buy in a few years.

We do know the companies that have cut advertising too much have lost market share. However when you have a lot of ads it is hard to figure out which ad made the difference.

Isn’t that basically the point of Nielsen? They go to a lot of trouble to get trustworthy stats about TV viewership so that advertisers can make informed purchase decisions wrt. both stations and programs.

Several companies have tried to fill that role online, but not to much success.

The basic attention token (https://basicattentiontoken.org/) seems to have the right characteristics and incentives to fill this role. Any ideas why the others have failed so far? It’s hard to insert yourself as a value-adding middleman in an existing marketplace unless consensus agrees the current configuration is in crisis. It’d be interesting if anyone could find a way to do this for Amazon’s commingled counterfeit inventory, for example.

I don’t know much about the current state of adtech, so I’d be hard-pressed to make specific recommendations. As a broadcast medium, TV stations had basically no data about the number of viewers that they had; Nielsen instrumented a representative sample of consumer TVs and the sold that data to both stations and advertisers.

That is the issue with closed ad inventory. On ad exchanges (which are slowly dying), you can use a host of third party companies (such as Integral Ad Science) to run more or less sophisticated verifications on the quality of the inventory.

-even though Facebook has all the incentive to falsify numbers

It's worse than that: Facebook has little short-term financial incentive to reject third parties who proliferate fake FB profiles.

Reply on news.ycombinator.com