Search engines don’t agree on who ranks.

Across 60 evergreen keywords sampled on July 22, 2026, Google and Bing shared an average of just 2.3 page-one result domains (31.3% of the shorter result set), and all three engines shared 2 (26.9%). Only 24.6% of the 57 keywords every engine answered produced the same #1 result on all three, and 31.6% produced three different winners. Brave is much closer to Google (71.3% overlap) than Bing is.

SERP volatility is how much a search engine's results change over time. This index measures it as the share of an engine's page-one result domains that are different from the previous day's sample of the same keyword: two identical result sets score 0%, two completely different ones score 100%. It is measured at the registrable-domain level, so a page moving from #4 to #2 is not counted as churn — only a domain entering or leaving page one is. The daily panel behind this page starts on 2026-07-22; the volatility series fills in as samples accumulate.

Because the engines mostly index different winners. 58.1% of the 666 domains that appeared anywhere in our snapshot appeared on exactly one of the three engines, and only 99 domains showed up across all three. Bing feeds Microsoft Copilot and a meaningful share of ChatGPT's web citations, and Brave runs an independent index — so a Google-only rank tracker cannot see where you stand in either.

A frozen basket of 60 evergreen head and mid-tail terms spread evenly across 6 categories: SaaS & software, Developer & API, Consumer & retail, Finance & money, Health & wellness, Travel & local. They are deliberately non-newsy — no people, events or breaking entities — so that changes in the results reflect ranking movement rather than the news cycle. The basket is frozen: a volatility index is only comparable over time if the keyword set does not change.

Yes. The raw daily observations are committed as open data (CC BY 4.0) and the collection script is public: every keyword's page-one domains, per engine, per day. Bing and Brave come from Crawlora's own search endpoints — the same API anyone can call with a key. Google's column in this snapshot was captured from page one of a real browser session rather than the API, and the panel records that difference per day so no statistic ever compares a browser-collected sample against an API-collected one. Country is US, language en.