Case Study 1 — The AMP Bargain: How Google Made a Format Mandatory, Then Made It Optional

Type: Real, public, Tier-1 event analysis. Every factual claim here comes from Google's own public announcements and documentation and the widely-reported public record of the AMP project and the 2021 Page Experience update. No traffic statistic for any named publisher is invented; where the industry-wide pattern is described, it is labeled as a pattern, not a precise figure.

Background: a fast-web bargain, offered from a position of power

In 2015 Google announced AMP — Accelerated Mobile Pages — an open-source framework for building stripped-down, very fast mobile web pages. The pitch was genuinely appealing: the mobile web was slow and bloated, readers abandoned sluggish articles, and AMP promised near-instant loading, partly by constraining what a page could do and partly by letting Google pre-cache and serve the page from its own infrastructure. For a news reader on a phone, AMP pages did load strikingly fast.

But the bargain had a second half that made it much more than a performance option. Google tied AMP to the most valuable real estate in mobile news: the Top Stories carousel — the swipeable strip of news articles at the top of mobile search results for newsy queries. In practice, through the late 2010s, being in that carousel effectively required AMP. For a publisher whose business depended on news traffic, this turned "should we adopt AMP?" from an engineering preference into a survival question. If your competitors were in Top Stories and you weren't because you hadn't built AMP versions of your articles, you were losing the most visible placement in mobile news.

So publishers built AMP — at real cost. Maintaining an AMP version of every article often meant a second rendering path, a constrained subset of HTML and JavaScript, workarounds for analytics and ads, and ongoing engineering effort to keep the AMP and canonical versions in sync. Many did it not because they loved the framework but because the alternative was invisibility in Top Stories.

The issue: platform dependence, wearing a performance badge

AMP drew sustained criticism, and the criticism is the SEO lesson. Three threads mattered:

  1. Centralization of control. AMP was, at launch, a Google-led project, and AMP pages were frequently served from Google's own cache under Google URLs — which meant a reader on an "AMP page" was often not on the publisher's own domain in the usual sense. Critics argued this concentrated control over the mobile web in one company's format and infrastructure. In 2019, partly in response, AMP's governance was moved to an open foundation (the OpenJS Foundation) rather than Google alone — an acknowledgment that single-vendor control of a de facto web requirement was a problem worth fixing.
  2. A requirement, not a choice. Because Top Stories eligibility was tied to AMP, publishers experienced AMP less as an optional speed tool and more as a tax on visibility. The framework's technical merits became almost beside the point; the incentive was the carousel.
  3. The dependence was invisible until it moved. Publishers had reorganized real engineering resources around a Google placement rule. That is a classic single-point-of-failure — and, as the whole chapter argues, single points of failure controlled by a platform are exactly the risk a publisher must watch.

📄 Read the Report

text FIGURE C34.1 — "The AMP bargain, and its reversal" [after public Google announcements] THE SOURCE Google's public AMP announcements (2015–2016), the AMP-for-Top-Stories era, and the June 2021 Page Experience update announcement + documentation. WHAT'S THERE 2015–2016: AMP launches; Top Stories on mobile effectively requires it. Late 2010s: broad publisher adoption under that incentive; 2019 governance opened to a foundation. June 2021: the Page Experience update makes ANY page meeting the page-experience / Core Web Vitals criteria eligible for Top Stories — AMP no longer required; the AMP badge is retired from the relevant surfaces. WHAT IT SHOWS A platform created a de facto requirement, publishers invested heavily to meet it, and then the platform removed the requirement. The investment's central rationale evaporated by decision, not by any failure of the publishers. WHAT IT DOESN'T It does NOT say AMP was worthless (fast pages are still good) or that Google acted in bad faith — only that the *incentive* publishers built around was Google's to revoke. THE MOVE Adopt platform-specific formats for their genuine merits and portable benefits (speed, Core Web Vitals), never solely to satisfy a placement rule that the platform can change. THE LESSON When a single company can make a format mandatory and then optional, building your house on that mandate is building on borrowed ground. Optimize for durable fundamentals.

What it shows

The AMP era is the clearest real-world illustration of this chapter's spine — platform dependence — and it teaches three things precisely:

  1. A platform can manufacture a requirement and then dissolve it. The June 2021 Page Experience update did exactly that: it decoupled Top Stories from AMP and made eligibility depend on general page-experience signals (including Core Web Vitals, Chapter 16) that any fast, good page could meet. Overnight, the core reason to maintain AMP — carousel eligibility — was gone.
  2. Investment made to satisfy a placement rule is fragile investment. Publishers who had adopted AMP only for Top Stories now had a maintenance burden with a vanished rationale; many deprecated or removed AMP over the following years. Publishers who valued the portable benefit — genuinely fast pages — kept what served users and dropped what didn't. The difference is whether you optimized for the platform's rule or for the underlying good (speed).
  3. "Google giveth and taketh" is not only about Discover. The chapter names Discover as the volatile surface, but the AMP story shows the same dynamic in a slower, structural form: the terms of visibility are set by one company and can change. The durable response is identical — build on fundamentals you control and formats whose value survives a rule change.

Outcome

By the mid-2020s, AMP had receded from its central role. It still exists and still produces fast pages, but it is no longer the toll road to Top Stories; the carousel is open to any page that delivers a good, fast experience. The publishers who navigated this best were those who had treated speed as the real goal all along — they already had fast canonical pages, so losing the AMP requirement cost them nothing and freed up engineering time. The ones who had treated AMP as a magic ticket discovered that the ticket had an expiration date set by someone else.

Crucially, note what did not happen: Google did not "penalize" anyone, and there was no scandal in the ordinary sense. The rules simply changed, as platform rules do. That undramatic quality is the point — the risk isn't malice, it's dependence.

The lesson

Optimize for durable fundamentals, not for a platform's current rule. AMP's genuine benefit — speed — survived the Page Experience update, because Core Web Vitals rewards fast pages no matter how they're built. AMP's incentive benefit — Top Stories eligibility — did not survive, because it was a rule one company owned. A working publisher reads every "you must adopt X to appear in Y" mandate through this lens: is X good for my readers and portable across changes, or am I only doing it because Y requires it today? If it's the latter, treat the resulting visibility as rented, and make sure it is not the only floor you stand on — which is the diversification argument of §34.7, taught by a real event.

Discussion questions

  1. Google tied Top Stories eligibility to AMP and later untied it. Was adopting AMP a mistake for publishers, a reasonable bet, or the only rational choice at the time? Argue it, distinguishing the speed benefit from the placement benefit.
  2. The chapter says "Google giveth and taketh" about Discover. Explain how the AMP story is the same dynamic in a structural, slower form — and what a publisher can do differently knowing both can happen.
  3. In 2019 AMP's governance was moved from Google alone to an open foundation. Why does who governs a required web format matter to a publisher, even one that finds the format technically fine?
  4. A vendor tells you "the next Top Stories requirement is coming — adopt our format now to be safe." Using this case, write the two questions you would ask before investing a single engineer-day.
  5. After 2021, some publishers kept AMP and some dropped it. What single question best predicts which choice was right for a given publisher — and how does that question generalize to any platform-specific optimization?