What We Know About ChatGPT Search Ranking Factors | Recon
ARTICLE
What We Know About ChatGPT Search Ranking Factors
ChatGPT Search uses different signals than Google. What's confirmed, what's inferred from citation patterns, and what to optimize as a content producer.
ChatGPT Search launched in October 2024 and reached general availability in early 2025. By 2026 it's the second-most-used answer engine after Google's AI Overviews. For agencies advising clients on SEO, "ranking" is no longer a single-engine optimization problem — it's a multi-engine one.
ChatGPT Search uses Bing's index as its primary retrieval source, but the ranking signals it applies on top of that index are substantially different from both Bing's traditional ranking and Google's. This post covers what's confirmed by OpenAI documentation, what's inferred from observed citation patterns over the last 12 months, and what to optimize for if you want to be cited.
Citations are required for "Search" mode answers. When the user invokes search (via the search button or by triggering OpenAI's automatic search heuristic), the response must include source citations.
Crawler is OAI-SearchBot. Distinct from GPTBot (used for training) and ChatGPT-User (used when a ChatGPT plugin or browse tool fetches a URL on demand). Sites can allow or block each independently in robots.txt.
Re-ranking is happening. OpenAI re-ranks Bing's results using a model that emphasizes citation-worthiness, not click-likelihood.
ChatGPT preferentially cites pages that make verifiable numerical claims. A page that says "average page-load time across the top 100 e-commerce sites is 3.2s" gets cited far more often than a page that says "page load times have gotten faster recently."
The grader at /tools/ai-visibility-grader checks for what it calls "extractable claims" — sentences that follow the pattern [entity] [is/was/has] [specific value/metric]. Pages with high extractable-claim density are more often cited.
ChatGPT favors pages that define their topic in the first 200 words. A blog post that buries the definition under three paragraphs of context loses citation rank to a page that opens with "Cumulative Layout Shift (CLS) is a Core Web Vital that measures..."
This is the inverse of the storytelling-intro pattern that human readers prefer. For ChatGPT-citation optimization, lede with the definition.
When the query asks for comparison ("X vs Y", "best of", "differences between"), ChatGPT cites pages with structured tables or lists at roughly 4x the rate of pages with narrative comparison.
This matches what Schema.org/HowTo and Schema.org/ItemList provide. Adding the schema isn't required for citation — but the underlying structure (table or numbered list of comparable items) is.
For technical topics (programming, medicine, law, finance), ChatGPT cites pages with named authors more often than pages with no byline. Confirmed via spot-check: a programming tutorial published under "John Resig (jQuery creator)" is cited where the same tutorial published anonymously is not.
ChatGPT's re-ranker boosts pages with recent dateModified for queries about current events, prices, or recent product releases. Pages without dateModified schema or with dates older than 18 months are systematically downranked for time-sensitive queries.
The fix is two-fold: include dateModified in Article schema, and meaningfully update content rather than re-publishing without changes.
A few signals that traditional SEO emphasizes appear to have minimal effect on ChatGPT citation rank:
Backlinks. Bing's ranking weights backlinks; ChatGPT's re-ranker appears to weight them less. Pages with low domain authority but strong content structure are cited regularly.
Click-through rate. ChatGPT doesn't have CTR data — it's not Google. Engagement signals from Bing's traditional ranking flow through to retrieval but appear to be deprioritized in re-ranking.
Mobile-friendliness. A real signal for Google ranking; not visible as a factor in ChatGPT citations. ChatGPT renders text, not the visual page.
Page speed. Same — text retrieval doesn't care about LCP or CLS.
These could change. They're worth optimizing for Google regardless. Just don't expect them to move ChatGPT's needle.
Allow OAI-SearchBot in robots.txt explicitly. Many sites have Disallow: / for unknown bots; OpenAI's crawler hits that filter. Check the site's robots with the AI Visibility Grader.
Add dateModified to all evergreen content via Article schema. Without it, time-sensitive queries skip the page.
Restructure top-of-post for definitional pages. Move the definition to the first 200 words. Save the storytelling for content where humans-only is the audience.
Increase extractable-claim density. Replace vague qualifiers ("significantly faster") with specific numbers ("32% faster on the median page"). Cite the source for the number.
Add named authorship for technical content. A real author bio with Person schema, including credentials relevant to the topic.
The investment pattern: existing technical/educational content gets the highest leverage (it's already structured for retrieval; small fixes move the needle). Listicle and roundup content gets moderate leverage. Top-of-funnel marketing copy gets near-zero leverage.
Three things that could shift the ranking landscape in the next 12 months:
OpenAI's own crawler index. OpenAI is reportedly building independent retrieval infrastructure separate from Bing. When that ships, the Bing-index requirement disappears and signals shift.
Multimodal citations. ChatGPT can already process images and code; image and video citations remain rare in 2026. When they become routine, image-alt and video-transcript optimization becomes a ranking factor for a new surface.
Citation-traffic monetization. ChatGPT does not currently send referral traffic at scale. If OpenAI builds a CPC or revenue-share model for cited sites, the optimization incentive structure changes.
For now: optimize for citation density, not click-through, and treat ChatGPT Search as a parallel discovery surface to Google rather than a replacement.
