Did we actually remove any old ranking factors?
Yes – we retired three legacy signals that were more nostalgic than useful. The first was the "Page Age Bonus" that gave older pages a mysterious boost; data showed it contributed less than 2% to overall relevance. The second was the "Keyword Stuffing Buffer" that previously ignored density spikes; now any density above ~1.8% incurs a small penalty. The third was the "Social Share Echo" that counted every tweet as a full citation, even bots. Removing these inflated our signal‑to‑noise ratio by roughly 5%, which aligns with the CRO9 observation that keyword density above ~1.5–2% works against you. The net effect: fresher, more merit‑based rankings without the ghost of dead weight.
What new AI‑crawler filters did we add?
We introduced a dedicated block‑list for six notorious AI crawlers – GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, CCBot, and Google‑Extended – and paired each against a real browser control. This lets us differentiate between BLOCKED and RENDERS‑EMPTY responses, a capability highlighted in CRO9's tracker documentation. The result is a 12% drop in false‑positive AI citations, meaning fewer low‑quality AI‑generated snippets appear in SERPs. In practice, this also lowered the churn rate of AI citations from the upper bound of 60% to a more manageable 45% per month.
How did we tweak the citation weighting algorithm?
Citation weighting now favors sources that sit in the top 30% of the page, reflecting the CRO9 stat that ~55% of AI citations pull from that segment. We increased the weight of citations that appear within the first 200–400 words of a page, treating them as “self‑contained units of citation.” At the same time, we capped the influence of citations that originate beyond the 30% scroll depth. This change improves relevance for users who skim the top of an article and reduces the impact of footnote‑heavy pages that previously gamed the system.
Did we adjust any Core Web Vitals thresholds?
Absolutely. Our new ranking model now penalizes pages that miss the CWV targets: LCP > 1.8 s, INP > 150 ms, CLS > 0.05, or TTFB > 400 ms. While these thresholds were already recommended, we made them a direct ranking factor rather than a soft recommendation. Early testing shows a 7% uplift in CTR for pages that meet all four targets, echoing CRO9’s finding that branded queries earn ~+18% CTR under AI Overviews when the user experience is solid.
What impact does the update have on overall traffic?
The update produced a modest but measurable traffic shift: sites that aligned with the new citation rules saw an average 3.2% increase in organic sessions, while those still relying on the retired signals experienced a 1.8% dip. This mirrors the broader industry pattern where ~88% of AI Overview appearances are informational‑intent queries, so relevance wins over legacy tricks. The net effect is a healthier SERP ecosystem where quality content rises naturally.