Re-reading a 2012 Penguin casualty: anchor velocity, not just ratio

Re-reading a 2012 Penguin casualty: anchor velocity, not just ratio

One of the more instructive early Penguin cases involved a finance affiliate that lost ~70% of organic traffic in the April 2012 rollout. The post-mortem (reconstructed from archived rank data) is useful because ratio alone undersells it.

The profile at impact:
— 58% of anchors were two exact-match phrases ('best X loans', 'X loans online')
— ~6,000 of ~8,400 those anchors had been built in a single 90-day sprint

So the site combined extreme exact-match concentration with a near-vertical velocity curve. Sites with similar ratios but flatter, multi-year acquisition curves in the same comparison set were hit less severely or not at all.

This hints that Penguin's signal was multivariate — ratio interacting with the temporal distribution of when those anchors appeared. The 'unnatural spike' framing in Google's own communications supports reading velocity as a co-factor.

Limitation: archived rank data is noisy, the comparison set was small (n≈12), and we have no algorithmic ground truth. We're inferring structure from outcomes.

On one hand, ratio and velocity co-moved here. On the other, manipulated profiles tend to be both spiky and over-optimized, so the two are hard to separate observationally.

Open question: would the same 58% ratio, accumulated organically over five years, have survived?