Is there an observable exact-match anchor threshold before things go wrong?

Is there an observable exact-match anchor threshold before things go wrong?

Everyone warns about over-optimized anchor text, but at what ratio does it actually correlate with trouble? I tried to find a waterline.

— Across 21 sites with known anchor profiles, I plotted the share of exact-match commercial anchors against whether the site had a visible ranking collapse in the prior 18 months.
— Sites with exact-match shares above roughly 25-30% were disproportionately represented among the collapses.
— Below ~15%, collapses were rare; the 15-25% band was mixed and noisy.

The nuance: this is a fragile correlation, not a Google-confirmed threshold, and reverse causation lurks (aggressive operators build aggressive anchors and take other risks). The safest read isn't a magic number but a direction: branded and naked-URL anchors dominating, exact-match as a minority, tracks with durability.

Caveat: anchor data from third-party tools is incomplete; "collapse" timing is fuzzy.

Method note: third-party anchor profiles + manual collapse-history coding, 21 sites.

Confidence: low