03 Aug 2025 Search

Previously: tires, myths, and reality

Just read the Google and Meta quarterly earnings reports, and there sure is money in the advertising duopoly business.

Add those together, multiply by 4, and the ad duopoly is pulling in about $472 billion per year. This is just their advertising businesses, not other lines of business like cloud services or hardware. Meanwhile, all advertising, total, worldwide is about a trillion-dollar business. It seems like the duopoly has already colonized about half of their Petri dish, and will max out in a few years.

But the duopoly are still growth stocks. The market expects Google and Meta market cap to increase faster than the number of people in the world increases, and also faster than the rate at which people get more money to buy more stuff. On average, that would mean, as Myles Younger wrote, the share of GDP that goes into advertising will have to increase. The duopoly can’t keep their status as growth stocks by just picking off the rest of the advertising business. (More US Ad Spend Comes Direct From Brands Than Through HoldCos as AI Reshapes Media Planning and Buying.) They have to grow advertising.

A fast-growing ad duopoly in a more slowly growing economy has a big obvious consequence for any business that uses advertising. On average, each successful customer acquisition will need to result in more and more money to Google and Meta. That even applies to remaining non-duopoly advertising. If Meta and Google have to keep driving up customer acquisition cost (CAC) on their own platforms, the whole advertising business will keep changing.

  • The costs of the remaining non-duopoly media will be bid up. Good news for landlords who own the space occupied by billboards.

  • The money goes to the duopoly anyway (Google often makes more money from ads on a legit site than the site does)

  • Non-duopoly ad media? In this economy? Now that the real antitrust enforcers are out of work here in the USA, and the legal environment still rewards consolidation and lock-in, the duopoly will get back to buying up any advertising-related business that they can’t squeeze the profit out of.

  • Even if governments successfully break up either or both of the duopoly firms, or if ByteDance makes it a big three, most of the problems that a duopoly causes will still be there with an oligopoly.

Besides driving up the cost of making a sale using its own ads, each duopoly company also needs to justify those increased costs by controlling the methods and tools used to report on how well the ads work. Paul Worthington explains:

The disruption of marketing wasn’t about software or data; it was a new truth. Digital Natives. Machine learning. First-party data. Predictive analytics. Real-time bidding. Performance Marketing. Product Marketing. Account-Based Marketing. Growth Hacking. The promise wasn’t just to increase reach or target more precisely—it was a whole new reality of how people consume, presented in dashboard form. Click here, spend there, get this many conversions. Never mind that the metrics were largely self-reported by the platforms themselves. Or that most of the promises being made—like last-click attribution—were a lie. Or that the platforms won’t allow you to run an independent audit. Or that fraudulent impressions are a feature rather than a bug. Marketers, while pursuing what they believed to be rational value-maximizing behaviors, instead became useful idiots for tech firms hellbent on industry domination. At some point, they stopped trying to hold the machine accountable and instead started thanking it for the scraps.

Controlling the numbers means giving advertisers training and tools. Both companies are releasing open-source marketing mix modeling software to help marketing departments justify ever-increasing budgets for the same results. But advertisers and agencies are already noticing higher CACs. Sam Tomlinson, an executive vice president at search-focused marketing agency Warschawski, told Catherine Perloff of The Information that his clients are spending more money to reach the same number of customers, and It’s not like you’ve got this wonderful assortment of good alternatives.

Increasing CACs will change the facts that feed into any decisions about customer retention. A company that had chosen not to do some customer retention feature or service might re-visit that decision as CACs keep being driven up. It looks like companies will have a variety of options for approaches to the CAC squeeze.

  • Keep cutting quality and service to pay increasing CACs while maintaining margins. This already seems to work and will probably keep working for some vendors.

  • Rely on lock-in to retain customers. This currently works for some categories of product but not others. Product categories where lock-in is hard will see innovation that makes it easier. (Yes, Lawrence G. Townsend predicted a lot of this stuff back in 2001, in Secrets of the Wholly Grill. Did the Juicero founders read this book?)

Those two are the dismal options, but as an Internet optimist I have to add a few more.

  • Reward shoppers for changing their habits. Amazon.com stopped paying for Google Shopping ads, probably because people who shop online are in the habit of putting an item in their cart and then checking if they can get the same thing on Amazon.com but with free or faster shipping. I don’t think that a company has to reach Amazon’s scale to do this. Habits can also change as a result of the next item.

  • Make quality and service improvements to keep customers, including some improvements that might have been harder to justify when CACs were lower. The more that CACs go up, the more it costs to lose a customer, so the better that doing what it takes to keep a customer is going to pencil out. Shifting money from marketing to quality and service will involve a lot of cultural change at a lot of companies. Money is hard to move across department boundaries, and marketing departments are better at getting budgets approved than customer service departments are. (This is a place where hiring Rory Sutherland to help figure out how to look at marketing and customer experience together, not separately, is going to be a win for a lot of companies.)Why are marketing employees considered talent while customer service employees are labor?

  • In political debates about the issues of surveillance and privacy, switch sides. Some possible privacy laws would limit value extraction by the duopoly, and benefit most other companies along with consumers. The duopoly’s policy people are used to being able to line up entrepreneurs and business groups to testify against proposed privacy laws. As the duopoly’s value extraction has to increase, that’s going to be harder and harder.

The post-surveillance economy will look like the pre-surveillance economy in some ways. This is already too long, but big picture the trend that needs to happen, for everyone except the central planners at the ad duopoly, is for reputation graphs to be distributed, partly public, and commoditized while value moves into the nodes of the graph. The duopoly’s goal is the other way around: keep proprietary reputation graphs confidential and under their control, and commoditize everything else. More: the Linux device driver hacker’s guide to giant Internet monopoly dominance

answers I would have posted in the comment section if this blog had a comment section

But it’s not really a duopoly, because Amazon.com ads! Amazon.com imposes a bunch of bullshit fees on sellers. Some of those fees are labeled as advertising. For example, if a seller is offering computer mice, they can pay Amazon not to show the AmazonBasics mouse at the top of the page when a shopper searches for Logitech mouse. It’s convenient for Amazon.com to call some of their bullshit fees advertising because it makes their bullshit fees look lower, and it’s convenient for the duopoly to call some of Amazon.com’s bullshit fees advertising because it makes their market share look smaller. But they’re not advertising by any normal definition.

(update 21 Aug 2026) Seth Godin explains The Amazon tax.

At first glance, this is compelling. Someone who isn’t sure what they’re looking for, who is looking for a book or a kitchen appliance, might find one if the right ad showed up at the right time.

But of course, that’s not what yields, or what most of the ads you see on Amazon do.

If you’re searching for an air fryer, Amazon already knows quite a bit. They know the best-reviewed, least-returned, best-priced model. The only purpose of the ads is to get you to pick an air fryer that isn’t that one (or for the best air fryer, to keep you on track to buy the one you wanted in the first place). The ads make the search worse.

The duopoly doesn’t need to grow ad revenue, they’ll get rapidly increasing revenue from AI! Maybe, in the future, people will be paying monthly AI bills, or paying for AI as part of other services. But that’s unlikely to mean all the money flows to the duopoly. Big incumbent companies have sustainable disadvantages in offering AI services. See My 2.5 year old laptop can write Space Invaders in JavaScript now, using GLM-4.5 Air and MLX by Simon Willison. Cheap independent AI keeps catching up. (If they get to say that AI is inevitable I get to say that Smaller, Cheaper, Faster is inevitable, too.) More: what if Ed Zitron is an optimist?

The duopoly really only needs one customer. Now that Trust And Safety is not so much a thing, the duopoly can pivot to propaganda, continue the direct mail/database/surveillance advertising tradition of building power and extracting wealth, and put pressure on less surveillance-friendly countries to accept the situation. They just need whoever ends up as dictator to sign one check. This is a tough problem. (Mike Elgan asks, Where’s a ‘Red Scare’ when you need one?) In general the duopoly tends to consistently underestimate the value that people place on not living under an evil dictator. Meta is fine with putting its users on the Republic of Gilead timeline (they agreed to the terms of service!) but at least 12 people in California are not. We could still end up stuck with a long-term duopoly situation as part of being stuck in an evil dictatorship. But in that case worrying about CACs is way down on the priority list anyway.

More: the duopoly crunch from the brand side