<b>Distribution drift: the profile you audited no longer exists</b>

<b>Distribution drift: the profile you audited no longer exists</b>

Anchor audits are snapshots, but link profiles are living systems with constant churn. Links decay — pages get deleted, sites die, redirects break. Crucially, the decay is not uniform across anchor types, which means your distribution drifts even if you acquire nothing.

Consider the asymmetry. Editorial, branded mentions tend to live on stable, maintained publications. Manipulated exact-match links tend to live on cheaper, more disposable properties. Over time the disposable tier dies faster, which can actually purify a ratio without any cleanup work — or, if your good links were on a defunct host, corrupt it.

— A clean January audit can become a skewed July profile through differential decay alone.
— Velocity analysis built on acquisition dates ignores the deletion dates entirely.

On one hand, this suggests profiles are partly self-correcting and one-time disavow campaigns may be re-fighting an old snapshot. On the other, decay is unpredictable per-link, so you cannot rely on it to fix anything specific.

Limitation: link-decay rates vary wildly by niche and are poorly documented; the differential-decay claim is plausible but I have no clean dataset segmenting decay by anchor type.

Open question: should anchor distribution be modeled as a flow — births and deaths over time — rather than the static pie chart every tool still draws?