Every SEO tactic dies the same way. We just keep forgetting.

Joost de Valk put the diagnosis cleanly in May 2026: "anything that can be spammed in SEO, will be spammed." That sentence is the entire model. Every SEO tactic that has been retired in the last fifteen years followed the same five-step arc, and the FAQ schema deprecation is just the most recent example.

Step one: a platform introduces a feature that gives publishers a small visibility advantage. Step two: a few practitioners use it correctly and post case studies. Step three: tools and agencies productize the tactic, and adoption becomes mass-market. Step four: the tactic is applied indiscriminately, the signal gets noisy, and the platform sees abuse outweighing utility. Step five: the platform restricts or kills the feature.

The five-step pattern is not about morality. The publishers who add FAQ schema to product pages aren't villains. They are responding rationally to an incentive a platform created. The cycle is structural: if a feature gives an edge, it will be applied everywhere it could plausibly apply, including everywhere it shouldn't.

The only thing that breaks the cycle is the platform retiring the reward. Which is also the only thing that gets attention — by then the damage is done, the tooling is built, and a new tactic is already rising to replace the dead one.

Search engines aren't withdrawing rewards out of spite. They're withdrawing them because the signal-to-noise ratio collapses. When 1% of sites used FAQ schema correctly, it was a useful hint. When 80% of sites used it — including product pages with three fake Q&A pairs at the bottom — it stopped telling Google anything about page type.

The retirement is forced by economics. Showing rich results to junk pages erodes user trust. Showing rich results to legitimate pages costs the platform nothing extra. But the platform can't reliably distinguish the two at scale once abuse is widespread, so the safe move is to remove the rich result entirely. That's what happened with FAQ. It's what's happening, more slowly, with several other markup types.

There's a strange honesty to the pattern: platforms tell you, in real time, which tactics are reaching the end of their useful life. Restrictions in 2023 were the canary. Full removal in 2026 was inevitable from the day FAQ schema started getting applied to recipe pages.

You can't stop the industry from running the cycle. You can opt out of the part that gets you punished. The rule set is short and not novel — it's just rarely followed because it doesn't feel like working hard.

Rule one: only adopt a tactic when you can describe what it does honestly on your specific page. Schema that lies, content rewritten for nobody, files no one reads — these are the early symptoms of cargo culting. If the tactic only makes sense as "because someone said it might help with AI," you are at the wrong end of the cycle.

Rule two: track tactics back to primary sources. If the trail goes blog → blog → blog without ever landing on a Google doc, a peer-reviewed paper, or a vendor with a methodology section, treat the tactic as fiction until proven otherwise. The FAQ "3.2x more citations" claim is the canonical example — citation chain leading to a broken link.

Rule three: weight signals by evidence quality. Causal studies and platform documentation beat correlation studies. Correlation studies beat anecdotes. Anecdotes beat speculation. Most AI SEO advice today is speculation dressed as anecdote dressed as data. Pricing your time accordingly is the cheapest insurance against the next deprecation.