How Facebook is cheating advertisers to make up for a day of lost revenue

Originally published at: How Facebook is cheating advertisers to make up for a day of lost revenue | Boing Boing


If they don’t already know that the advertising business model is a cheat already they’ll swallow Zuckerberg’s latest slimeball move without giving it a second thought.


Essentially two thieves stealing from one another.


Does this mean Trump-publican ad spending has also been sped up? In which case there may be a silver lining.


CYA :roll_eyes:


These are called “pacing algorithms”, and they are, generally, terrible across the board. We regularly see partners who blow their entire load of ads in the first day of the month, or that end up doing 50% of their month’s spend in the last 72 hours of the month which is also terrible for the advertiser.

Here’s hoping that FB shining a light on this sort of practice will cause these ad providers to be better at this, because it would make our lives on the publisher infrastructure side easier and smooth out these huge swings in supply.

I would much rather that advertisers hold their demand partners to account for uniform pacing of their campaigns.

Disclaimer: I am a 17-year veteran of the adtech industry and the food on my table comes directly from managing adtech.


About a year ago Cory did a review of Tim Hwang’s book “Subprime Attention Crisis”, also mentioning a Wired article on the topic by Gilad Edelman:

It would seem that these broken (by design in FB’s case) “pacing algorithms” are only one small aspect of a severely flawed and scammy technology and business model that seems fundamentally resistant to attempted improvements by competent and ethical developers like yourself.

To give an anecdotal example, the ad-tech that’s had enough effort put into it to get past my blockers still tends to serve up ads trying to sell me durable goods and vacations I was shopping for and bought months ago, along with random things that would have no apparent interest to whatever my demographic/behavioural profile is. That doesn’t seem to deliver a whole lot of value to most parties involved.

I suppose, though, that I should be grateful that there are enough gullible advertisers willing to give enough money to slimy ad brokers and dodgy operations like StackSocial to keep the lights on at great sites like this. However, if the ad-tech market is a bubble as Tim Hwang suggests (and is increasingly exposed as such by slimy actions like this) then I hope there’s a fallback position in place.


There is a very good reason that, for my entire career, I have paid 0 attention to the payloads of what my networks carry. :slight_smile:

My job is to make sure those payloads get moved around quickly (so we don’t slow down your browsers!), reliably (again, so we don’t timeout and delay your pageloads) and that our inter-partner communications are as reliable as possible, since real-time-bidding (RTB) means that your pageloads are delayed by slowdowns between bidders and buyers.

Pacing, at the technical level, messes with that system because 1) more responses means we have to plan for that capacity, and 2) many supply sources try to “predict” how many ad responses we will get from the demand (advertising) side (again, so we don’t wait around for a bunch of requests that likely won’t result in an ad anyway), and broken pacing algorithms make our data scientists cry as we try to solve that.

Thing is, most people (advertisers included) don’t care about my part of the world, they care about the part you mention - what happens in those ad payloads, and how often they are “effective”.

I don’t think there’s a bubble to burst here, primarily because the “effectiveness of advertising” has been a nubulous thing since looooooong before there was an internet. We know it works because nearly everyone has bought or been swayed by an ad at some point in their lives, myself included. The fact that we can measure it now to the granularity we can provides massively more data to advertisers than we’ve ever had before, and most of them still follow the same process that advertisers did in the 50’s: “Hey, we have a thing, we ran ads, we sold more of the thing”. They don’t really care about these specifics - they want to catch the person that didn’t already book their vacation or buy the thing, and they don’t care about blanketing hundreds or thousands of people with ads for things they’ve already bought any more than they care about Honda car owners seeing ads for Honda cars on TV. They aren’t the market for the ad, and who knows, maybe being reminded of your vacation or Honda owners being reminded of Honda will influence your next purchase, too.

But again, not in my wheelhouse. :slight_smile: It passes my mythbusters-esque “plausible” sniff test though, and as a result, I don’t think any bubble is going to burst around advertising as a whole, just maybe some of the shady ways people like FB bill for that advertising (like cost-per-click or cost-per-conversion).


Something seems fishy here. I’m not saying that facebook doesn’t or isn’t doing underhanded things, but it doesn’t seem to me like the results presented could be caused by facebook doing what it is alleged to do based on a ~6 hour outage unless that is extremely cherry picked data or caused by some other effect.

In particular, a more than 50% drop in conversions? Over how long of a period? If facebook were doubling the ad delivery rate for a specific customer over a 6 hour period, I still wouldn’t expect a 50% reduction in conversion unless 100% of the “bonus” ads were ineffective. Even then, when looked at over a day or so the impact should be less. A > 50% reduction in conversion rate means something else happened, essentially either your ad or your audience changed.

Maybe the interruption could have disrupted the auction economy and caused the bidding to go up so people are spending more for the same # of slots? I could believe that maybe. I still don’t think it would be long lasting from a relatively short outage.

1 Like

When is it enough money?

1 Like

Grifters gonna grift, no matter what.


On Twitter, I keep getting ads for Helium-3 isotopes.

Assuming that they’re not broadcasting it to everyone, what the heck is in my user profile that makes them think that I’d need some?


Are you trying to get Mark Zuckerberg caught in an infinite loop?


I have this vision of our choices directly change what you eat.
We click on a ramen ad, you get a case of ramen show up.


Hmm check your user profile data. Is “Mad Scientist” set to “True”?


That would explain why every other item on my feed is an ad for the past couple days, at least.

1 Like

I think all these tech types are infinitely loopy.


That is all…

1 Like

Clearly we need more lobster and banquet ads, stat!


They know about your doomsday device side project.
Sorry, autocorrect. I meant “cold fusion research”.

1 Like