Claude whitethorn beryllium much straight tied to Brave Search rankings than different AI reply engines, according to accusation Jonathan Clark shared on LinkedIn from a Zero Click by Profound session.
Clark, managing partner astatine Moving Traffic Media, said the session’s cardinal takeaway was that Claude “doesn’t re-rank hunt results” and alternatively appears to usage Brave’s apical 10 results straight successful its answers.
Claude searched little often. Claude utilized web hunt successful 36.6% of prompts, compared pinch astir 90% for ChatGPT, according to Clark.
- Claude was astir apt to hunt erstwhile prompts signaled freshness, rankings, location, aliases comparison intent. Recency-focused prompts specified arsenic “best XYZ” triggered hunt 81% of the time, while ranking-focused prompts triggered hunt 67% of the time.
- Location-focused prompts triggered hunt 55% of the time, while comparison prompts specified arsenic “X vs. Y” triggered hunt 51% of the time.
Brave rankings carried weight. Claude’s citations overlapped pinch ChatGPT’s successful only 8% of cases erstwhile responding to the aforesaid prompts, according to Clark.
- Claude’s results had overmuch higher overlap pinch Google rankings, astatine 64%. This suggests that Google SEO efforts whitethorn transportation complete much readily to Claude than strategies focused specifically connected improving visibility successful ChatGPT, according to Clark.
- The uncovering besides accrued the value of Brave rank tracking. Clark said Claude uses Brave, and that ranking good successful Brave gives america “something we tin show and correlate to data.”
Some prompts stayed successful memory. Prompts specified arsenic “how does,” “what is,” and “steps to” were little apt to trigger Claude to hunt the web. When Claude doesn’t search, it can’t mention web pages. Claude searched astir often for prompts containing position specified arsenic “best,” “top,” “near me,” and comparison-style queries, according to Clark.
Years showed up often. Clark besides noted 2 patterns that could make Claude easier to test:
- Claude’s query fan-outs were astir deterministic, producing the aforesaid fan-out 65% of the clip crossed users.
- The fan-outs often included years.
- That intends page titles pinch current-year signals whitethorn person an advantage successful Claude-triggered searches, particularly for ranking and freshness-driven prompts.
Why we care. Claude visibility appears to dangle much heavy connected ranking successful the hunt results Claude uses. Clark’s takeaway was that Claude whitethorn beryllium 1 of the astir optimizable AI reply engines coming because its hunt behaviour appears much accordant and much intimately tied to observable hunt rankings.
English (US) · 
Indonesian (ID) ·