The homepage-dilution myth: branded links don't cancel keyword links

The homepage-dilution myth: branded links don't cancel keyword links

A mechanistic misconception drives a lot of bad spend: the belief that buying branded anchors "dilutes" risky keyword anchors, as if anchors net against each other like charges. Distribution analysis doesn't obviously work that way. Adding 200 branded anchors lowers your percentage, but if the underlying keyword anchors came from a detectable paid scheme, the scheme is still there — you've buried it in a denominator, not removed it.

On one hand, a healthier-looking ratio is not nothing; distribution genuinely matters. On the other, if penalty logic also inspects link quality and scheme membership independent of ratio, dilution addresses a symptom while leaving the diagnosis intact.

The mistake is modeling anchor risk as a single scalar that branded links can offset, rather than as multiple independent signals.

The fix:
— Fix the source problem (remove/disavow scheme links) before cosmetically adjusting the ratio
— Don't manufacture branded links in bulk to dilute; bulk branded acquisition is its own unnatural velocity pattern
— Treat ratio and link-quality as separate axes; passing one doesn't clear the other

Limitation: we can't confirm whether ratio and quality are scored independently or jointly; the dilution critique assumes some independence.

Open question: can a sufficiently natural branded-anchor base genuinely neutralize a small exact-match scheme, or does scheme detection operate on absolute link signatures regardless of surrounding distribution?