The recovery-lag problem corrupts nearly every anchor case study

The recovery-lag problem corrupts nearly every anchor case study

A recurring claim: "I cleaned up exact-match anchors and rankings recovered in N weeks." These timelines anchor much of the field's practical advice. They're also the most confounded data we have.

The issue is lag and concurrency. Anchor changes propagate only as Google recrawls the referring pages — weeks to months, unevenly. During that window, the site keeps changing: new content, other links, algorithm updates, seasonal demand. By the time rankings move, dozens of variables have shifted. Attributing the recovery to the anchor cleanup specifically is a story, not a measurement.

On one hand, when the only deliberate change was anchor removal and recovery follows a plausible recrawl window, the inference has some force. On the other, the recrawl window overlaps so many uncontrolled changes that clean attribution is essentially impossible, and confirmation bias does the rest — we remember the cleanups that preceded recovery and forget those that didn't.

A meta-look at published recovery timelines found enormous variance (weeks to a year) for nominally similar interventions, which is what you'd expect if the intervention isn't the main driver.

Limitation: no case study controls for the recrawl-window confound, because you can't freeze a live site.

Practical read: treat anchor-recovery timelines as anecdotes, not dosage curves.

Open question: what fraction of attributed anchor recoveries would have happened anyway from concurrent changes?