accounting help needed

Careful mate… that privacy nerd wants your cookie!

anyone have the original source for this? First of all, they cite Digital Advertising and Market Structure: Implications for Privacy Regulation by Deisenroth et al. This paper is intended to influence privacy policymakers, but it’s really more interesting from an antitrust point of view. The claim the researchers make here is that basically

1. Apple ATT made the Meta ads less effective

2. As a result, small advertisers ended up spending more money on ads to make a sale, and many of them ceased advertising on Meta sites and apps, or went out of business entirely.

In a fair advertising market, the result of (1) above would be that the affected ad medium is forced to cut rates in order to keep customers. (If radio advertising were regulated, the radio stations would cut their ad rates or reduce ad load in order to stay competitive with newspaper ads and billboards.) But Meta might have a monopoly in this kind of small business advertising market, depending on how you define it. The Deisenroth et al. paper, showing that Meta was able to pass the costs of ATT along to small business advertisers instead of having to eat them, is remarkably good evidence that Meta has a monopoly somewhere.

Larger advertisers saw less impact, which makes sense—for now, at least, larger advertisers have more options for where to advertise. Larger advertisers can use adtech firms like The Trade Desk to put ads on small legit sites. But the problem of how can a small legit company put an ad on a (big or small) legit site is not well addressed by the adtech business. A small site can sign up with an ad services company to run ads from big advertisers, a big advertiser has their choice of big and small sites or Google or Meta, but a small advertiser is pretty much stuck. A better title for the paper might have been We Already Have A Monopoly in Small Business Advertising, So If You Try Any Privacy Stuff We Will Punish The Small Business Advertisers And It Won’t Hurt Us LOL.

The same paper makes one of the old privacy paradox arguments. (we have new and better paradoxes but they dig up an old one) claiming consumer behavior seems inconsistent with a strong preference for privacy. This paradox has been figured out by now. Of course people don’t choose to go where their family and friends aren’t, or dink with settings that are hard to find and easy for companies to reverse. When we want privacy we don’t go ask companies, that would be a waste of time. We have a strong enough preference for privacy that we generally escalate to effective options such as the ballot box (example: Prop 24) or the jury box (example: Meta v. Flo).

Meta has had sufficient market power to force small businesses to give up on organic social and switch to surveillance advertising. And, because big advertisers can move ad budgets to other places like TV and The Trade Desk and small advertisers can’t, Meta was able to increase customer acquisition costs (CACs) for small businesses much more than for large ones. Meta is a growth stock, expected to grow faster than the economy as a whole. Because the Meta/Google duopoly is already about 47% of the advertising business globally, they have no alternative but to keep increasing customer acquisition costs. Mr. Market needs to see Meta growing, or will freak out. So Meta has, as Ben Thompson points out, successfully adapted to the ATT environment on iOS.

But after its success in adapting to ATT, Meta did not—and to meet market expectations, could not—bring ad rates back down for the affected small businesses. CACs went up as ATT rolled out and Meta passed the costs along, but then stayed up and continued to go up. In a world without ATT, Meta would still have had to raise rates, but on its own initiative, and with the risk of catching more grief from competition regulators. With ATT in the picture, they were able to re-frame their squeeze on small businesses as just passing along the costs imposed by Apple.

Finding the accounting error

Meta and Deloitte seem to have an accounting error in here somewhere. The effects of surveillance advertising that we’re seeing in the real world don’t match up to Meta’s claims about it. With a little research, it should be possible to do some better math. We can start with two helpful facts about money.

• People don’t have infinite money. Any money that is taken in a scam, gambled away, or spent on a regrettable win-lose deal is not available to be spent in a win-win transaction for a legit product or service.

• Every deal has two sides. When a customer has a favorable outcome from buying something, a legitmate business gets income from the same transaction. Or, to look at it from the other side, when a consumer is experiencing some kind of economic harm, that means some money wasn’t available for a win-win deal, the deal didn’t happen, and some legitmate business is worse off than they would have been.

Meta claims it’s linked to €213 Billion in Value and Almost 1.5 Million Jobs, which is a pretty weak-sauce argument. The Soviet Union’s entire(ish) economy was linked to Gosplan and very few modern researchers would agree that was the best way to run things. The real question that keeps coming up for legislators is: is surveillance advertising really good for small legitmate businesses? Looking at the big picture, it would be hard to say that it is. In recent business history, the growth of surveillance advertising has coincided with years of buying more and more of our stuff from fewer and fewer companies. Is is just that bigger companies can advertise more efficiently because the ad market works differently? It’s easy to say that big brands can use TV, or the remaining non-duopoly adtech firms, or other media, and small businesses have fewer advertising options, so naturally Meta is going to squeeze small businesses for higher CACs (see the ATT saga above), giving big businesses an advantage over time. But is that really all that’s going on? Even after the Meta/Google duopoly manages to squeeze out all the other advertising, so that small and large businesses are in the same boat as far as lock-in goes?

Meta makes a big claim:

Personalized ads enable growth because they work—they make it easier and faster for people to find products and services they’re looking for, and they make it possible for entrepreneurs to start, grow and market their businesses, significantly enhancing economic productivity.

The good news for customers here (and an inconvenient fact for Meta) is that the above claim is more testable than it looks, because of how money works. For every sale by a legitimate business, there is a positive outcome for some customer. The transaction is not just an Econ 101 win-win, it gives researchers a way to cross-check what’s going on. Meta makes it hard to evaluate their claims from the Meta side, largely because they insist on serving both legitimate and deceptive advertisers (and often give the deceptive advertisers an advantage). So we can probably best measure the impact of surveillance/personalized advertising from the customer side. And from the customer side, it doesn’t look good: advertising personalization: good for you?

So if the customer outcomes are negative, and positive outcomes for customers have to balance with successful sales for legit businesses, then how is Meta so often able to claim some surveillance advertising benefits? It looks like what’s going on is two effects: price discrimination and fraud.

• Do a Control-F on the full Deloitte/Meta report, and look for price. Not found. A big win from personalization from the seller side is capturing some of that sweet, sweet consumer surplus. Personalization enables sellers to price discriminate, I mean surprise and delight (some) consumers (classified by ML as bargain hunters) with personalized (lower priced) offers (not available to those classified as less price sensitive).
  
  Legit small businesses are less likely to benefit from price discrimination than large ones (look at what happened with Groupon—a lot of entrepreneurs got clobbered offering specially-priced deals to Groupon users.)

• Meta has trouble telling fraud operations and legit sellers apart (remember, this is by design, not just an unwanted consequence of an understaffed moderation team.) Some of the small business success measured from the Meta side is going to be more accurately categorized as fraud losses from the consumer side.

Implications for policy work

Because every deal has two sides it is possible to do a lot of research on benefits for legit small businesses by looking at the customer side.

• Compare people who made a high-personalization tool or settings choice 5-10 years ago to people who made a high-privacy choice. Which group has higher incomes and assets today?)

• Compare users and non-users of commonly used surveillance advertising apps. Do the users buy more or less from legit local businesses than the non-users?

• Are users of privacy tools less satisfied with the products and services they buy?

When Meta makes claims about small businesses their numbers are hard to trust because of all the deceptive advertisers they allow on their platform. Too many negative-sum deals get mixed into the data. Research on all of this stuff would be a lot easier if they chose to not just staff up on their existing anti-fraud efforts, but change the rules of the ad game on their platforms to stop giving so many advantages to the fraud side. But that’s another story.

An experiment somebody (not me) could do

Andrew Orlowski writes, in Trump’s TikTok deal sets a very dangerous precedent,

[T]his is a straightforward asset seizure, which has not gone unnoticed by American executives with investments in China and the BRICS.

If the administration in the USA can expropriate ByteDance’s service here, then some other government could take away one of Meta’s services in their country. And here’s how to turn it into an economics experiment: hand over the local Meta operation to a new company, and pack the Board of Directors with small business owners. This would give us all a chance to see the differences between actual pro-small-business advertising outcomes and Meta management’s claims about benefits to small business. A lot of anti-fraud measures are relatively cheap, and Craigslist has done them, while Meta persists in messing up all the numbers by allowing so much fraud. Anyway, definitely a publishable experiment here for somebody in some government somewhere.

More: the duopoly crunch from the brand side

Related

Some good news anyway. Any time that Professor Eric Goldman writes about a court disagreeing with his interpretation of Section 230, something positive is happening for the victims of one of Big Tech’s schemes. Meta didn’t just stand up a Usenet server, they built a system designed to facilitate fraud. (IMHO the big platform companies are not Usenet servers, newspapers, or the Pruneyard Shopping Center, and legislators would be better off regulating them as what they are: Big Tech platforms: mall, newspaper, or something else?)

(update 6 Oct 2025) Deloitte will refund Australian government for AI hallucination-filled report by Kyle Orland. The Australian Financial Review reports that Deloitte Australia will offer the Australian government a partial refund for a report that was littered with AI-hallucinated quotes and references to nonexistent research. (Don’t tell Meta to check their Deloitte report, all right?)