The Mechanisms Of Enshittification

Enshittification is the gradual, deliberate degradation of a product or service by whoever controls it. Cory Doctorow coined the term in 2022 for digital platforms, which decay in a recognizable sequence: good to their users first, then worse for those users to please business customers, then worse for everyone to feed the platform itself, and finally collapse. The word caught on fast, and Paul Krugman even formalized an economic model in a 2025 article.

Doctorow has already chased the rot off the screen and into physical things along one key dimension: products with a chip in them and a lock on the chip. The printer that rejects a cheaper cartridge, the tractor a farmer is barred from repairing, the refrigerator that learns to show ads. A thin layer of software, shielded by law, becomes the lever, and the maker can make the product worse because you can neither escape it nor fix it.

This essay is about the other physical goods, the ones with no chip to lock and no app to spoil. Your backpack, your mattress, the sandwich at the diner. They get worse too, on a slower clock, with nothing digital trapping you. A different mechanism is at work.

That mechanism is ownership. My central claim is that a brand's quality tracks who it answers to, and that this, rather than greed, is what explains decline. Founders answer to the product and to a name that is usually their own. Outside capital answers to a return. Both want money, so greed is the one trait founders and financiers reliably share, which is exactly why it explains nothing about why some brands keep their quality and others rot.

The variable is time horizon and audience. An owner who will be there to collect a reputation's future earnings protects it. One who plans to exit, or who answers only to a margin target, spends it down, because a reputation is a depletable asset and the consequences for depleting it arrive only after they are gone.

The purpose here is to set that process down once, in full, as a reference. What follows defines the lifecycle a brand moves through as it changes hands, catalogs the specific mechanisms by which a reputation gets converted into cash, and explains why consumers usually cannot react fast enough to stop it. It ends with a short field guide for telling, before you buy, which side of the line a brand sits on. The evidence for each claim is the set of investigations this newsletter has already run, linked throughout and gathered at the end.

This is a living document - it will grow as I uncover new reputation extraction tactics through my research.

A reputation behaves like capital. Benjamin Klein and Keith Leffler made the point formally in 1981: a good name is, in economic terms, a stream of future earnings, and a company keeps its quality up largely because cheating the people who trust it would forfeit that stream. The reputation has a present value, and it sits on the books as an asset whether or not anyone writes it down as one. An owner decides, in effect, whether to live off the interest or sell off the principal.

Spending the principal either amounts to deliberately lowering input costs (quality) or increasing the price of a given product. When Krugman set enshittification down as a formal economic model, he left quality out of it on purpose and built it on rising prices alone. His model neatly fits a technology platform lifting its take but is less useful for a physical good that gets worse while the price holds.

In either case, the key point to consider is that a shorter time horizon will inevitably increase the aggressiveness with which an owner is willing to sacrifice a brand's reputation in the interest of short-term earnings. A private-equity firm working on a three-to-seven-year clock spends a reputation down fastest, since it will be long gone before the name runs out.

The 'reputation extraction' pattern reliably moves through four stages.

A founder builds something good, and the trust accrues slowly, over years of the product earning it. Then, at some point, comes the handoff. The moment the company stops answering to a person and starts answering to capital: a sale, an IPO, a death, a hostile takeover. Then the harvest, where the new owner converts that trust into cash, cutting quality while the name still carries the old promise. Then the husk: a name with nothing behind it, kept alive because it can still be licensed out for whatever the logo alone is worth.

Scale is not necessarily the villain here. The power tool industry holds a good example. A Hong Kong conglomerate, TTI, bought Milwaukee in 2005. It poured money in: $206 million of R&D in a single year, with the founding family still holding about a quarter of the company. Milwaukee got better.

For an opposing example in the same industry, an American conglomerate, Stanley Black & Decker, bought Craftsman in 2017, carried billions in debt, and built a failed factory in Texas that could not reliably stamp its own name on a wrench and shut down within three years of opening.

Craftsman got worse. The same move in the same decade and the opposite result, because the owners answered to different things.

There are two ways to run the harvest, and the veterinary industry demonstrates both. Mars, the company behind Snickers and Pedigree, buys clinics and keeps them, milking the margin year after year under names like Banfield and VCA. KKR buys clinics, borrows against them (a single $2.3 billion loan the clinics themselves have to service), folds in more, and aims to sell the whole thing inside a few years. One holds and squeezes, the other flips. Either way the clinic is run as a financial instrument first and a place that treats your dog second.

Whole Foods is the arc end to end. John Mackey spent thirty-seven years building a grocery chain that answered to a mission and to the neighbors who once turned up with mops to bail it out of a flood. He took it public in 1992, which let Wall Street into the room. In 2017 an activist fund bought in and told him, by his own account, that it would replace his board, fire him, and sell to the highest bidder. He sold to Amazon for $13.7 billion. The fund walked off about $300 million richer two months later. Mackey was the founder the entire time. What changed was the audience he answered to, and the suppliers and the workers and the quality of products on the shelves all followed it down.

When researchers studied private equity buying up nursing homes, they found that ownership by a buyout firm raised patient mortality against comparable non-PE homes, differing only in who held it and for how long. The owner type was the variable that killed people. And when the harvest is done, the brand does not really die. The name detaches from anyone who ever made the product and becomes a thing to rent.

These are the specific ways a reputation gets spent. The list is not complete, and it is not meant to be. New ones surface every time I look into a new category, and I'll add them here as they do.

Buy a loved brand out of bankruptcy, fire the designers and shut the factories, then rent the name to whoever will pay to stamp it on their product. Authentic Brands Group has run this playbook on Brooks Brothers, Eddie Bauer, and Sports Illustrated, among fifty-odd others, and admitted the whole thing in its own SEC filing: it does not make the products and has limited control over their quality. The trust the original company earned becomes a royalty check, cashed quarter by quarter until nothing is left, and then they go find the next name.

Fill the shelf with brands that look like rivals and quietly own all of them. One Italian company, EssilorLuxottica, owns Ray-Ban and Oakley and Persol, the stores you would buy them in, and the vision insurance that pays for them. The three premium names on the mattress wall are a single manufacturer. Whirlpool sells you Amana at the bottom of the ladder and JennAir at the top and builds them in the same plants. The choice is an illusion.

Swap the materials for worse ones and leave the label crisp. Backpacks lost their denier count and their YKK zippers while the logo stayed sharp. A Pennsylvania diner's all-beef patties turned into beef cut with soy filler across a few quiet contract renewals, and the owner only found out when he finally read the label.

Keep advertising the guarantee that built the loyalty, then rewrite it so it never pays. JanSport's lifetime warranty now excludes "normal wear and tear" on a bag engineered to wear out. Eddie Bauer killed its lifetime guarantee outright in 2019. A twenty-year mattress warranty that voids over a single stain is a sales tactic wearing a warranty's clothes.

Engineer the product to fail right around the end of the warranty so you sell another one. Smart appliances break roughly 40% more often than the plain versions, and a dead control board can cost a third of a new machine, so the broken one goes to the landfill on cue. John Deere built tractors a farmer can no longer legally fix without crawling back to the dealer, the practice that got the company sued by the FTC in 2025.

Borrow against the company you just bought, pay yourself the loan as a dividend, and leave the debt on the books of the thing you intend to sell. George Akerlof and Paul Romer named this one in 1993: under the right conditions it becomes rational to bankrupt a company for profit, looting it on the way out the door. It is why the Instant Pot, a genuinely great product with a devoted following, ended up bankrupt in 2023 after a private-equity owner loaded it with debt.

This one is newer, and it belongs to the direct-to-consumer era. When you sell straight to people online, the cost of acquiring each customer can rival the cost of making the product itself. The money that could have gone into developing a better product goes to the ad auction instead. Casper spent 43% of its revenue on sales and marketing in 2017, against a cross-industry norm of 10 to 12%, and earned about three dollars back for every dollar it spent. Allbirds put a quarter of its revenue into marketing and still lost more money every year, $101 million in 2022 alone. Then Apple's 2021 privacy change made acquiring a customer more expensive overnight, costing Meta around $10 billion and hitting the smallest brands hardest. A company that has to rent its customers from an ad platform has less left for the product. When the platform raises the rent, the product is the first thing cut.

Once they have you, leaving gets expensive. Banfield sells a wellness plan that renews on its own and front-loads the value, so cancelling means owing them money, and grieving owners have been billed for plans attached to pets that were already dead.

None of this would survive if the people paying for it could walk away quickly enough to matter. They usually cannot, for three reasons.

The lag. A brand's reputation outlives its quality, sometimes by years, because you keep buying on the strength of what the company used to be. How long the lag runs comes down to how often you buy the thing and how easily you can judge it once you do. A mattress is the perfect target: you buy one maybe twice a decade, and you cannot tell a good one from a bad one in a showroom. A t-shirt is a terrible target, since you buy them constantly and you know inside three washes. So the extraction runs longest and richest in the durable, hard-to-judge categories: mattresses and tools and appliances and eyewear, which is not a coincidence and is most of what I end up writing about here.

The fog. You cannot judge what you cannot see, and George Akerlof won a Nobel for working out where that leads. His 1970 paper showed that when buyers cannot tell quality before they pay, the good stuff gets driven off the shelf by the cheap stuff that looks identical. The brands lean into the fog and thicken it on purpose. The mattress industry sells one bed under ten different names so that no two stores carry a comparable model and the price-match guarantee never has to pay out. Half the time you cannot tell the well-made version of a licensed brand from the gutted one, because they wear the same label.

Nowhere to run. The surest way to stop you leaving is to own everywhere you would go. Tim Wu, the antitrust scholar, put the eyewear monopoly plainly: it is as if Hermès and Louis Vuitton were secretly the same company, mimicking competition while pocketing both sales. In food distribution, the one place an independent restaurant could check a fair wholesale price, Restaurant Depot, is now being bought by Sysco, the giant they were checking it against. This is the physical-world version of the lock-in Doctorow described in software.

The defense is knowing who you are buying from before you pay, and a handful of pointed questions will help you make an informed decision before your next big purchase.

Find out who owns the brand now, and whether the founder is still there and still in control, because that one fact predicts more than any review. Check whether the company went through bankruptcy in the last decade, and who bought the name out of it. If the brand is public, check what share of its revenue goes to marketing. Notice when the same brand turns up at a flagship store and at Costco and at TJ Maxx all at once, a sign the name has been split across tiers and licensees. Ask how the person selling to you gets paid, because commission turns every recommendation into a sale. And run the math that marketing is built to keep you from running: divide the price by the years the thing will genuinely last, and compare across fifteen years instead of one weekend, because the cheap option you replace twice over is usually the expensive one. If you cannot find out who actually makes a product, that is your answer.

I keep the running tally on The Brand Ledger, a searchable record of every brand I have taken apart: who owns it now, what it used to be, and whether it is still worth your money. It changes as the brands change hands, because they will. This essay works the same way. It is one page meant to stay current, and it will be updated every time I discover a new mechanism.

A dollar spent with a company that still makes its own things is a dollar the roll-up did not get, and enough of those dollars is the only language this machine understands.

Strip it to the bone and the story is simple. The decline you keep noticing is a decision. Somebody chose it, on a clock you could not see, to collect a return that arrives before the bill does, and then they moved on to the next name on the list. The product got worse because the people who own it stopped answering to it.

That is the machine. It runs in tools and groceries and eyewear and your dog's vet, and I’m certain it will run in whatever category I pull apart next.

Every claim above traces back to one of these. Each is a single category, taken apart.

If you have a category you’d like me to dissect next, or know of an enshittification mechanism I have not named here, reply and tell me.

If someone forwarded this to you, subscribe here for the next one.