You balanced the ratio and broke the relevance
An overcorrection failure. Terrified of exact-match, a team scrubs descriptive anchors entirely and ends up with a profile that's 95% branded and naked-URL. The exact-match ratio looks pristine — and the page now has almost no anchor-level relevance signal telling search engines what it's actually about.
On one hand, anchor text is a real relevance input; descriptive anchors help a page rank for its topic. On the other, hammering one descriptive phrase invites over-optimization scrutiny. The two pressures are in genuine tension, and the mistake is optimizing one to zero.
Natural editorial profiles are neither all-branded nor all-keyword; they're a messy blend including partial-match, topical, and generic anchors. Erasing descriptive anchors is as unnatural as flooding them.
The fix:
— Target a blend, not a corner solution: branded and naked-URL as the base, with a meaningful minority of varied descriptive and partial-match anchors
— Use partial-match anchors ("a guide to X") to carry relevance with lower over-optimization signal than exact-match
— Judge anchors by whether the mix resembles organic editorial behavior, not by minimizing one category
Limitation: the optimal blend is not a published constant; it's inferred from correlation studies with wide variance.
Open question: how much page-level ranking relevance is genuinely carried by anchor text post-BERT, given improved on-page semantic understanding — and is the descriptive-anchor relevance gain shrinking?
Anchor Theory
@AnchorTheory
You balanced the ratio and broke the relevance
Этот пост опубликован в Telegram-канале Anchor Theory. Подписаться можно по ссылке: @AnchorTheory.