The Google Graveyard Isn't Carelessness. It's a Company That Only Knows How to Count One Thing.

There is a website that exists only to count the dead. It is called killedbygoogle.com, it was built by an outside developer, and as of 2024 it lists more than 250 discontinued Google products — 57 apps, 209 services, and 22 pieces of hardware that the company once shipped, promised to support, and then quietly turned off.⁸ No company maintains an obituary page for its own products. So a volunteer built one, and it never stops growing. That single fact — that the corpse count needs a public tracker — tells you the killing is not occasional. It is structural.

The official story, repeated every time another beloved tool goes dark, is that Google has a kind of institutional restlessness — bored engineers, shifting priorities, a company too rich to finish anything. That explanation feels true and is almost entirely wrong. Google does not kill products because it is careless. It kills them because, by its own internal arithmetic, they were never going to count.

The company can only see one number

Start with the math that runs everything. In fiscal 2024, Alphabet reported total revenue of $402.8 billion, up 15% year over year, and the overwhelming weight of it sits inside Google Services — Search, YouTube, Android, Maps, Play — with Google Cloud as the other real engine. Everything else, including the moonshot lab X, lives in a small, loss-making segment called Other Bets.⁷ That structure is the whole story. When a single advertising-fed monoculture produces hundreds of billions of dollars, anything that does not feed that monoculture — or grow into a comparable franchise of its own — is not a business. It is a hobby with a server bill.

Here is the thesis a smart friend can repeat at dinner: Google's graveyard is not evidence of innovation failure. It is evidence of a deliberately narrow definition of what 'success' means at this company. A product survives only if it can plausibly become a billion-dollar line of its own, or fortify Search, YouTube, or Android. Fail both tests — be merely good, merely loved, merely profitable at a modest scale — and you are not safe. You are on the list.

Read the death certificates and they all say the same thing

Look at what Google actually wrote when it pulled the trigger, and the pattern is not chaos — it is a single sentence, repeated. Google Reader, killed in 2013, was explained away with 'a loyal but declining following' and a stated desire to focus on fewer products.¹ Notice the words: not 'unprofitable,' not 'broken' — declining, and a desire to focus. Reader had devotion; the petition to save it gathered more than 100,000 signatures. Reader had devotion; the petition to save it ultimately drew more than 150,000 signatures.¹⁰ It simply had no path to the only number that matters, and devotion is not a line item.

Stadia, the cloud gaming service, got the same treatment with a more revealing twist. When Google announced it would shut the servers off on January 18, 2023, the executive in charge said plainly that the service 'hasn't gained the traction with users that we expected' — and then said the quiet part: the underlying streaming technology would be reused inside YouTube, Google Play, and the company's AR work.³ The product died; the capability got harvested and folded back into the franchises that count. That is the monoculture digesting one of its own experiments. And the company refunded hardware and games — but explicitly refused to refund Stadia Pro subscriptions, a small, cold detail that tells you exactly how much sentiment factors into the accounting.²

Google+ looks at first like the exception that breaks the rule — the famous case of a product killed for losing to Facebook. But the death certificate there reads differently than the legend. The proximate trigger for the accelerated shutdown was not the social scoreboard at all: a 2018 API bug exposed the private data of roughly 500,000 users, a flaw Google had noticed in March 2018 but did not disclose, and a second breach later that year exposed around 52.5 million profiles, directly speeding up the timeline.⁴ The 'it lost to Facebook' story is the comfortable one. The real one is that a product with no path to franchise scale and a live legal liability is the easiest thing in the world for this company to let go.

The lab where killing is the job

If you doubt that the killing is a strategy rather than an accident, look at the one place inside Alphabet where it is the openly stated method. X, the moonshot lab founded in 2010 with a mission to 'invent and launch moonshot technologies,' is run on the principle of trying to kill projects early — tackling the hardest part first and rewarding teams when a project is shut down.⁵ Its leader has said that only about 2% of the lab's experiments ever graduate, and that most of what they try doesn't work out — and that's okay.⁹ That figure is a quote, not an audited metric, so hold it lightly. But the philosophy is the point: at X, death is not failure. It is the product. The graveyard is just that same logic applied to things that already shipped.

Isn't this just discipline? The honest counter.

The fair objection is that this is good management, not tragedy. Every large company prunes; a firm that kept every product alive out of sentiment would drown in maintenance debt, and a willingness to abandon sunk costs is exactly what most companies lack. By that reading, the graveyard is a feature — proof of an organization unsentimental enough to stop spending on things that won't matter. There is real truth in that. The discipline is genuine, and the refusal to defend a dying franchise out of pride is rarer and healthier than critics admit.

But discipline answers the wrong question. The issue was never whether Google prunes — it's the threshold it prunes against. When the only test of life is 'can this become a billion-dollar franchise or feed one,' you systematically euthanize the entire category of things that are merely excellent: the loved tool with a modest, stable audience, the product people would happily pay for at a scale that means nothing against $400 billion. The cost is not borne on the income statement. It is borne in trust. Every dead product teaches developers and users a lesson — that building a life around a Google product is a bet against the house — and that learned reluctance is the real, uncounted line item the monoculture never has to book.

So the graveyard is not a story about a company that can't finish things. It is a story about a company that finishes precisely the things it values — and values almost nothing that doesn't pour back into the one machine that pays for everything. Google didn't lose interest in Reader, or Stadia, or the rest. It simply ran each of them through the only filter it owns and found, as it almost always does, that the world's love and a billion dollars are not the same number. The lesson isn't that Google is reckless with what it builds. It's that a company can be ruthlessly rational and still leave a trail of bodies — when 'rational' has only ever meant one thing.