<b>Counting nofollow and UGC anchors as if they were dofollow</b>

<b>Counting nofollow and UGC anchors as if they were dofollow</b>

A classification failure that skews every ratio. Many audits lump all anchors into one distribution regardless of <code>rel</code> attribute. But since 2019, <code>nofollow</code>, <code>ugc</code>, and <code>sponsored</code> became hints, and these links almost certainly carry different anchor weighting than clean follow links. Mixing them inflates or deflates your computed exact-match ratio depending on where the keyword anchors sit.

On one hand, Google now says it <i>may</i> use nofollow links for ranking and crawling, so they aren't pure noise. On the other, treating a forum-profile UGC keyword anchor as equivalent to an editorial follow anchor misstates your real distribution.

The mistake is computing one undifferentiated ratio across signal classes that are almost certainly weighted differently.

The fix:
— Segment the distribution by <code>rel</code>: follow, nofollow, ugc, sponsored
— Compute your over-optimization ratio primarily on the follow segment, where anchor weight is highest
— Watch sponsored-tagged exact-match anchors specifically; correct tagging is what keeps paid placements compliant

Limitation: the "hint" model means we no longer know the exact weight of nofollow anchors — it's deliberately opaque.

Open question: are exact-match anchors on ugc/sponsored links included in over-optimization assessment at a reduced weight, or excluded from the commercial-anchor count entirely?

—
Хочешь больше link velocity penalties? @TooFastTooLinked