Case Study 17.2 — The Rise and Fall of AMP: A Bet on Rented Land

A real, public arc. The product history and Google's announced changes are Tier-1 fact. The antitrust material is explicitly labeled as allegations in ongoing litigation — not established findings — and we present it as exactly that. No statistics are invented.

Background

AMP — Accelerated Mobile Pages — was announced by Google in October 2015 and launched in early 2016 as an open-source framework for building very fast mobile pages. It was, on its face, a good-faith answer to a genuine problem: in 2015 the mobile web was slow, bloated with heavy ad scripts, and miserable to use on a phone. AMP fixed that by constraining the developer — a restricted subset of HTML, no arbitrary custom JavaScript, and pages that Google could pre-cache and serve from its own infrastructure so they appeared almost instantly when tapped from search.

For publishers, AMP arrived with a carrot big enough that it functioned as a stick. To appear in the Top Stories carousel — the image-topped, top-of-screen news unit that drives an enormous share of mobile news traffic — a page effectively had to be AMP. For any organization whose business depended on that placement, AMP was not a technology to evaluate. It was a toll to pay. And so they paid it, building and maintaining a parallel AMP version of essentially every article, alongside the "canonical" version on their own site.

The issue: a coercive incentive and its hidden costs

From the start, publishers had grievances that the Top Stories carrot kept them from acting on:

  • The URL problem. Because AMP pages were served from Google's AMP Cache, readers often saw a Google-hosted address (a google.com/amp/...-style URL) in the address bar rather than the publisher's own domain. To a news brand, the URL is part of the brand and the trust signal; handing it to Google felt like erasure. (Google later introduced Signed Exchanges to let AMP pages display the publisher's real URL — a partial technical patch that did not resolve the deeper discomfort.)
  • Loss of control. AMP's constraints limited page design, analytics implementations, and — critically for a media business — advertising and monetization setups. Publishers felt they were ceding control of their own product to a framework governed by their largest source of traffic.
  • The maintenance tax. Every article now existed twice. Two versions to build, test, keep in content parity, and debug — a permanent drag on engineering.
  • Governance optics. A public "AMP Letter" signed by web developers and publishers in 2018 objected that a dominant search engine using preferential search treatment to push its own technical format was bad for the open web, regardless of AMP's technical merits.

The contested part (labeled: allegations in litigation, not findings). AMP also surfaced in a multi-state antitrust complaint led by Texas concerning Google's advertising-technology business. The complaint alleged, among many other things, that Google used AMP in ways that advantaged its own ad exchange and disadvantaged rival ad-tech and "header bidding," including by affecting how non-AMP ads loaded. These are allegations in ongoing legal proceedings, not proven facts, and Google has disputed them. We include them only to note that AMP's story became entangled with a broader debate about platform power — a debate a strategist should be aware of without pretending it is settled.

Hold every one of these costs in mind, because they were all being paid the whole time. The Top Stories requirement was the only thing making the ledger balance.

What it shows: the limits of a platform-specific bet

Then Google removed the requirement.

With the Page Experience update — announced in 2020 and rolling out on mobile through 2021 — Google opened Top Stories eligibility to any page that met its news policies and provided a good page experience (Core Web Vitals and the rest). AMP was no longer the price of admission. The distinguishing AMP badge in results was retired. In one announced change, the single incentive that had justified all of AMP's costs simply evaporated.

The instant the requirement lifted, the cost-benefit math inverted for almost everyone. Why maintain a second, constrained, Google-hosted version of every page when a well-built responsive page that passes Core Web Vitals is fast enough to compete and keeps your URL, your control, and your single codebase? The answer, for a great many publishers, was "no reason at all." A wave of major publishers dropped AMP in the following period, and new sites largely never adopted it.

Crucially — and this is the honest nuance — AMP was never a ranking factor in itself. Its SEO value was always indirect: speed (a modest page-experience input) plus the now-removed Top Stories gate. Take away the gate, and there was no ranking rationale left to defend.

Outcome

AMP the technology did not vanish — it still exists, with niche roles in email (AMP for Email) and some ad formats, and sites that already run it and are happy can keep it. But as an SEO necessity, it is over. The framework that once compelled the world's newsrooms to rebuild their sites became, within a few years of losing its privileged placement, an optional legacy technology in decline. Chapter 16 measured the thing that replaced it — Core Web Vitals — as a neutral, standards-based way to reward the same speed AMP had enforced, without demanding a specific vendor's framework or a Google-hosted URL.

The lesson

  1. Beware building on rented land. AMP's entire value rested on a search perk that a single company controlled and could revoke — and did. Any strategy whose payoff depends on a platform-specific format, a preferential placement, or a proprietary technology carries that revocation risk. Prefer fast, standards-based pages you own. You never have to unwind a bet you didn't make.
  2. Distinguish a real benefit from a coerced one. AMP genuinely made pages fast. But much of its adoption was not because speed was worth the trade-offs — it was because Top Stories left no choice. When you evaluate any tactic, ask: would I do this if the platform weren't forcing my hand? If the only answer is "the placement requires it," you are exposed the day the requirement changes.
  3. "Indirect ranking value" is fragile value. AMP wasn't a ranking factor; it was a gate. Gates move. Durable SEO value comes from being genuinely the best, fastest, most useful answer — the things Google can't easily take back because they're not perks it grants but qualities you possess.
  4. This is why "you still need AMP" is a myth (§17.7). The requirement is gone; the costs never were. Chasing it now is paying a toll for a road that's been demolished.

Discussion questions

  1. AMP solved a real problem (a slow mobile web) with a coercive mechanism (Top Stories exclusivity). Can a technically good solution be strategically bad for the ecosystem it's imposed on? Where's the line?
  2. Publishers maintained AMP for years despite disliking it. What does that tell you about the power of a single, dominant traffic source to shape how independent businesses build their products?
  3. The antitrust claims about AMP are allegations, not findings. How should a professional talk about contested matters like this — to a client, in a report, on social media — responsibly?
  4. Core Web Vitals replaced AMP's role with a neutral, standards-based metric. Is that a better model for the open web? What new risks, if any, does a metric-based gate carry?
  5. Apply the "rented land" lesson beyond AMP. Name two other SEO or marketing bets that depend on a platform perk that could be revoked — and how you would hedge each.