A complete picture of how AI assistants see your site.
Reffed runs the same set of checks on your website that ChatGPT, Claude, Perplexity, Google AI Overviews, Gemini, and Microsoft Copilot use to decide who to recommend. Here is everything included.
All six AI assistants, scored individually.
Most AI search tools check two or three assistants. Reffed checks all six. Each one gets its own score and its own list of fixes, because each one weighs signals differently.
Weighted toward clear business identity, content depth, and original facts. Your score reflects how likely ChatGPT is to cite your site when a buyer asks for a recommendation in your category.
Favors thorough content with clear structure. Claude recently added live web search, so we check whether the ClaudeBot crawler can reach you and whether your content matches Claude's depth-over-breadth preferences.
Citation-heavy by design. Loves FAQ schema, HowTo content, and recent dates. The crawler is PerplexityBot. Recent content with fresh stats outperforms older evergreen pages.
Uses standard Googlebot, so traditional SEO signals carry the most weight here. Structured data (FAQ, HowTo, Article) matters more in AI Overviews than anywhere else.
Draws from the same index as AI Overviews but cites differently. Authority signals and entity identity matter most. Strong overlap with how Google's Knowledge Graph evaluates your business.
Uses Bingbot plus Microsoft's own model. Bing weights breadcrumbs and image alt coverage more heavily than other engines. Strong in LinkedIn-integrated B2B contexts.
AI crawler access check, across every major bot.
If your robots.txt blocks the AI crawlers, none of the other optimization matters — those assistants literally cannot see your site. This is the single most common catastrophic mistake we find.
Plus OAI-SearchBot — OpenAI's two crawlers, one for ChatGPT training and one for live search. Both must be unblocked for ChatGPT to surface your site.
Anthropic's crawler for Claude. Required for Claude to see your site, both in training and in live web-search mode.
Perplexity's citation crawler. Blocking this means zero chance of being cited in Perplexity answers, regardless of how strong your content is.
The AI training opt-out token Google honors for Gemini and AI Overviews influence. Allowing this lets your content shape how Google's AI models understand your category.
Feeds both Bing and Microsoft Copilot. Also drives ChatGPT's live web search results, since ChatGPT runs on Bing's index for fresh information.
Common Crawl, which most AI assistants learn from at training time. Blocking CCBot has a long tail — it limits how future model generations see your business.
Many website builders (Wix, Squarespace, certain Shopify themes) block AI crawlers by default. We flag this in red on your audit so you can fix it in one line.
llms.txt detection, the new AI standard.
In September 2024 a developer at Answer.AI proposed llms.txt — a single file at the root of your site that tells AI assistants which pages are most important. Think of it as a sitemap built for AI. As of mid-2026 only a small percentage of sites have one, so it is an easy way to stand out.
- Detection — we check the standard
/llms.txtlocation and confirm it is a real text file, not a 404 page served as HTML - Format check — your llms.txt should be Markdown-formatted; we verify it parses correctly
- Coverage check — we look at whether your highest-value pages are listed
- Recommendation — if you do not have one, we explain what to put in it
Mention check, would AI actually cite you?
All the technical signals in the world do not matter if no AI assistant would ever pick you as the answer. The Mention check identifies your category, runs a typical category query, and gives an honest assessment of whether you would be cited versus larger competitors.
- Category detection — what is your business actually selling, and to whom
- Typical query — the most likely AI-assistant question someone in your category would ask
- Verdict — likely, possible, or unlikely to be cited
- Reasoning — the specific signals you are missing (or that you have working in your favor)
- Competitors — the established players AI assistants would cite instead, so you know what you are up against
Structured data and content audit, end to end.
Beyond the AI-specific checks, Reffed runs a full technical audit on every signal that matters for getting cited.
Does your site tell AI assistants what your business is and how to contact you, in structured form. The foundation signal for entity recognition.
Required for AI assistants to surface specific content or items. Without it, your pages exist but aren't categorized for AI use.
Heavily favored by Perplexity and Google AI Overviews. Pages with FAQ schema get cited 2-3x more in question-answer contexts than pages without.
Bing and Microsoft Copilot weight this more than other engines. Also helps AI engines understand your site hierarchy when citing internal pages.
Clear H1, H2, H3 makes content quotable. AI engines pull short answers from headings and use them as the surface signal for what each page covers.
Word count, paragraph structure, presence of specific facts and numbers. Thin pages don't get cited regardless of how well-structured they are.
How many of your images have descriptive alt text. Bingbot in particular treats alt coverage as a quality signal for the whole page.
Duplicate-content cleanup. AI engines reject pages with conflicting canonical signals, so this is a baseline correctness check.
Length, clarity, click-worthiness. These often become the first text an AI engine sees when deciding whether your page is worth indexing further.
Reffed Watch — the full audit, every Monday.
The free audit runs once. Reffed Watch runs the full deep audit on your site automatically every week and emails you the score change with the prioritized fix list. $59 a month, cancel in one click.
- Full deep audit every Monday — not just the score, the whole report
- Per-engine breakdown — see which AI assistants are improving and which are slipping
- Mention check rerun — track whether AI citation likelihood is moving in your favor
- Score change vs last week — know exactly what is working and what is not
- Top three fixes ranked by impact — fresh every week as your site evolves
- All historic reports in one dashboard — see your 90-day trajectory
- Cancel from any email — one click, no friction, no questions
Reffed Watch+Fix — automated fixes.
Watch+Fix is the Reffed tier that does not just find the problems — when it opens, it fixes them automatically on your WordPress, Shopify, or Webflow site. It's in early access at a planned $249/mo or $2,490/yr, hand-onboarded — not self-serve yet. Here's what it will do.
- Auto-deploy schema markup — Organization, Article, FAQ, HowTo, Breadcrumb all written directly to your site files
- Meta title and description fixes — applied across every indexed page
- llms.txt generation and maintenance — created if missing, updated as your site grows
- robots.txt fixes — AI crawler blocks unblocked with one click
- Image alt text — written via vision model for every image on your site
- One-click rollback — undo any individual change or the whole batch
- Everything stays on your site if you cancel — we write to your CMS, not a JavaScript overlay
Watch+Fix is in early access at a planned $249/mo. Join the waitlist to be hand-onboarded.
Run a free audit and see what Reffed would find on your site.
No signup. Full report in 60 seconds. Works on any website.