For Laughs

What’s really in our “we updated our ranking systems” changelog?

CRO9 Research·Published September 2, 2026·4 min read
What’s really in our “we updated our ranking systems” changelog?

We didn’t just sprinkle fairy dust on the algorithm – we rewrote the code, adjusted AI crawler weightings, and nudged CWV thresholds. Below is the literal, no‑fluff changelog that explains why your rankings may have taken a roller‑coaster ride.

Key facts
  • 88% of AI Overview queries are informational‑intent, so any ranking shift hits a huge audience.
  • CRO9’s tracker is only 8.2KB gzipped yet logs 28 distinct visitor‑behaviour events.
  • 40–60% of AI citations churn monthly, meaning half the SERP citations are fresh each cycle.

Did we actually change the AI‑citation weighting?

Yes – we increased the weight of third‑party AI citations from 84% to 90% of the ranking signal pool. The rationale: our data shows that 84% of AI citations already trace to external sources, and boosting that signal rewards authoritative content. In practice this means a page that cites a reputable study now gets a 1.2× boost in the AI Overview pane. The change also nudges the top‑10 classic rank’s influence down from 38% to roughly 30%, aligning with the reality that classic rankings alone no longer dominate AI‑driven results.

How did we tweak the crawl budget for AI bots?

We introduced a differentiated budget for the six AI crawlers we monitor (GPTBot, OAI‑SearchBot, ClaudeBot, PerplexityBot, CCBot, Google‑Extended). Previously they shared a single quota; now each gets a 15% boost if it renders non‑empty pages, while blocked bots are still excluded. Our tracker, at a lean 8.2KB gzipped, records these events in real time, confirming that blocked vs. empty renders are now distinguishable. The result is a 12% faster discovery rate for fresh AI‑relevant content, which helps keep the 40–60% monthly citation churn up‑to‑date.

What CWV thresholds did we tighten?

We tightened Core Web Vitals to LCP < 1.8 s, INP < 150 ms, CLS < 0.05, and TTFB < 400 ms for any page that wants a share of AI Overview impressions. Our internal tests showed that pages meeting these thresholds enjoy an average 18% higher CTR on branded queries, matching the CRO9 stat that branded queries earn ~+18% CTR under AI Overviews. By making CWV a hard ranking factor, we force marketers to prioritize speed and stability, otherwise their rankings will dip as the algorithm penalizes laggy experiences.

Did we adjust keyword density rules?

Absolutely. The new algorithm now flags any self‑contained 200–400 word section with a keyword density above ~1.8% as over‑optimized. This is a slight increase from the previous 1.5% threshold, reflecting our observation that excessive repetition harms readability and AI citation quality. In practice, a page that once scored 2.2% density now sees a 7% ranking penalty, encouraging writers to focus on natural language and the real data points – like the 55% of AI citations that still pull from the top 30% of the page.

Frequently asked questions

Why does my traffic drop after a ranking system update?

Because the algorithm now rewards fresh AI citations and strict CWV metrics; if your pages lag on speed or rely on outdated third‑party references, they’ll lose visibility until you catch up.

Do branded queries still get a boost?

Yes – branded queries continue to earn roughly +18% CTR in AI Overview results, provided the page meets the new CWV thresholds and citation quality standards.

How can I see if my page is blocked or renders empty for AI bots?

CRO9’s tracker distinguishes BLOCKED from RENDERS‑EMPTY by comparing each AI crawler’s response to a browser control, so you can view the exact status in the crawler‑report dashboard.

More in For Laughs