Why Doesn't ChatGPT Recommend Your Brand? An AI Visibility Audit Guide

ChatGPT skips your brand for one of three reasons: it can’t retrieve you, it can’t verify you, or it doesn’t trust you enough to name you. ChatGPT Search answers from web pages its OAI-SearchBot crawler was allowed to read, then cites sources backing the claim, and OpenAI’s own help center warns those citations “can be incomplete, outdated, or incorrect.” If your site blocks the bot, if independent sources barely mention you, or if your pages offer nothing quotable, the model falls back on competitors it can verify. An AI visibility audit: AI visibility audit is a structured check of whether, where, and how accurately AI assistants mention your brand: shows you which of the three is actually happening.

Nothing here guarantees citations, because nothing can.

Why doesn’t ChatGPT recommend your brand in the first place?

Because recommending a brand is the last step in a chain where you can fail at any earlier link. OpenAI’s help documentation describes the process: ChatGPT may search the web automatically when a question benefits from current information, it typically rewrites your question into one or more targeted queries, and results are ranked “using multiple factors intended to help users find relevant, reliable information.” No placement is guaranteed. OpenAI says so plainly.

The chain looks like this:

1. Retrieval. Your site has to be crawlable. Per OpenAI’s crawler documentation, OAI-SearchBot “surfaces websites in ChatGPT’s search results,” and blocking it means your site “will not be shown in ChatGPT search answers.” That’s the single most common silent failure discussed: a robots.txt rule meant to block AI training accidentally blocks search inclusion too.
2. Selection. ChatGPT picks which retrieved pages support the answer. OpenAI notes it may work with third-party search providers, and independent analyses through 2025-2026 found it citing pages well beyond any single index, so ranking in Bing alone doesn’t cover you.
3. Naming. This is where “cited” and “recommended” split. In her research on AI shopping answers, Lily Ray found that brands publishing their own “best of” listicles got the listicle cited as a source, but the publishing brand was left out of the actual recommendations about 69% of the time. The model used their list as evidence for recommending someone else.

One more wrinkle: answers aren’t stable. Sampling randomness, saved memories, and personalization shift what ChatGPT says between sessions. If you asked once, saw a competitor, and concluded you’re invisible: that’s not a measurement, it’s an anecdote.

How ChatGPT picks brands and sources: what the evidence shows

The strongest verified signal is corroboration. Ahrefs’ AI brand visibility study (December 2025) analyzed 75,000 brands across ChatGPT, Google AI Mode, and AI Overviews. Branded web mentions (your brand name appearing on third-party pages) correlated with AI visibility at 0.65-0.71 across all three engines. YouTube mentions correlated even harder (~0.737). Backlinks? Around 0.19, barely a blip. The same handful of brands dominated every engine, with 0.75-0.82 overlap between them.

Two honest caveats before you run off: these are correlations, not causes (Ahrefs says so itself) and “get mentioned more” isn’t a tactic you can execute directly. A related Ahrefs analysis found brands with 10x more web mentions earned about 3.7x more AI mentions, which tells you volume of third-party talk and AI visibility move together. The actionable version: the sources AI answers already cite are the pages you want to be mentioned on. That’s what step 5 of the audit finds.

For your own pages, the Princeton-led GEO study (published at KDD 2024, tested on 10,000 queries) remains the best controlled evidence: adding citations, quotations, and statistics to content boosted its visibility in generative-engine answers by up to 40%. Verifiable specifics beat adjectives.

Is AI search big enough to justify this yet?

Yes, and it’s worth being precise about why. The numbers cut both ways.

• 68% of US Google searches ended without a click in January-April 2026, up from roughly 58-60% in 2024, per SparkToro’s study built on Similarweb clickstream data (June 2026).
• AI referral traffic grew 9.7x year over year across ~82,000 sites: yet still accounted for only about 0.1% of total referral traffic, with Google sending 210x more (Ahrefs, June 2025).
• What little traffic does arrive is qualified: Semrush’s AI search study found AI-search visitors convert at about 4.4x the rate of traditional organic visitors. (Conversion rate, not volume. AI referrals remain a rounding error in most analytics.)
• ChatGPT reported around 900 million weekly users in early 2026, and OpenAI’s own usage research (September 2025) found roughly 21% of chats are information-seeking: exactly the queries where brands get named or skipped.

Less than a third of Google searches now send a click. When the answer happens inside an AI chat instead of a results page, your brand either shows up in that answer or it doesn’t exist.

The volume is small; the intent is high. For most businesses that means AI visibility is a compounding asset you build cheaply now, not a fire alarm.

What actually moves AI visibility: and what doesn’t

Here’s the testable-versus-speculative split: the part worth screenshotting.

Backed by evidence you can check:

• Allow OAI-SearchBot (and decide deliberately on GPTBot, which is training-only: blocking it doesn’t affect search inclusion). This is straight from OpenAI’s docs, and robots.txt changes take effect in about 24 hours.
• Make your pages quotable. The GEO study’s up-to-40% lift came from citations, quotations, and statistics: specifics a model can verify and repeat.
• Exist on the sources that already get cited. Independent reviews, comparison articles, YouTube videos, industry roundups. Correlation-heavy but directionally solid across three engines.
• Be unambiguous. One consistent brand name, one clear description of what you do and for whom. Entity confusion is a fixable crawl-and-training problem.

Not backed: treat sales pitches about these skeptically:

• llms.txt. It’s a real proposal, but no major AI company has confirmed any system reads it. Google’s John Mueller said publicly in June 2025 that no AI system currently uses it, adoption hovers around 10% of domains, and it’s become the new meta-keywords tag. Harmless to add; worthless to buy.
• Schema markup as a citation lever. Ahrefs tracked 1,885 pages that added schema between August 2025 and March 2026 and found only marginal citation changes (+2.2% on ChatGPT, +2.4% on AI Mode). Keep schema for classic SEO; don’t expect it to bend AI answers.
• Guaranteed citations. OpenAI’s docs explicitly state there’s no guaranteed placement. OpenAI is experimenting with advertising in ChatGPT (the crawler docs even list an ads-related bot) but ads are separate from answers, and nobody can sell you a citation.
• Reddit spam. Note how fast citation mixes shift: Promptwatch data reported Reddit’s share of ChatGPT citations collapsed from ~3.8% to ~0.5% in mid-August 2026. Betting a strategy on any single platform’s citation share is building on sand.

The AI visibility audit, step by step

Here’s the whole procedure in one table, then unpacked.

Audit step	What to check	Tool	Cost
1. Crawl access	OAI-SearchBot (and GPTBot) rules in robots.txt	Free robots.txt testers / your CMS	Free
2. Prompt set	15-25 buyer-shaped questions in three types	Spreadsheet	Free
3. Run + record	Mentions, citations, position per run	ChatGPT (logged-out) + spreadsheet	Free
4. Share of voice	You vs. 3 competitors across the set	Same sheet; Peec AI free GEO report	Free
5. Source audit	Which pages ChatGPT cites per prompt	Manual review of citations	Free
6. Accuracy check	How ChatGPT describes your product/pricing	Fresh session, logged out	Free
7. Off-site presence	Whether cited sources mention you	Ahrefs site explorer or manual	Free-$129+/mo
8. On-site fixes	Quotable stats, comparisons, entity clarity	Your CMS	Free
9. Re-run monthly	Same prompts, same sheet, trend line	Spreadsheet discipline	Free
10. Automate (optional)	Daily tracking, alerts, dashboards	Peec AI / Brand Radar / Semrush / Otterly	$29-$495/mo

1. Confirm ChatGPT can crawl you. Open your robots.txt (yoursite.com/robots.txt) and check the rules for OAI-SearchBot, GPTBot, and ChatGPT-User. If OAI-SearchBot is disallowed, ChatGPT Search can’t show you as a source. Fix it, then wait about a day. OpenAI says changes take roughly 24 hours to propagate.
2. Build the prompt set. Write 15-25 prompts a real buyer would type: category questions (“best lightweight CRM for freelancers”), comparisons (“Notion vs Coda for teams”), and problem questions (“how do I stop cart abandonment”). Mix in your brand name on a few, to test what the model already knows.
3. Run each prompt 3-5 times in clean sessions. Logged-out, incognito, no custom instructions. Record every run: did your brand appear, which competitors did, and which sources were cited? Averages, not single answers.
4. Score share of voice. For each prompt set, calculate the percentage of runs where you’re named versus each competitor. This number is your baseline: it’s what you’re trying to move month over month.
5. Audit the cited sources. This is the highest-leverage step. List every page ChatGPT cites across your set. You now have a concrete target list: these are the pages the model already trusts for your category, and where mentions of you would carry weight.
6. Check factual accuracy. Ask ChatGPT what it knows about your brand, pricing, and products. Errors here trace back to stale third-party pages or thin on-site information, and per OpenAI’s own warning, its citations can be outdated or wrong, so verify against the sources.
7. Assess your off-site footprint. For the top cited sources, check whether they mention you. If the answer is “rarely,” your worklist is: pitch the relevant comparison posts, get listed in the roundups, and consider YouTube, which correlated most strongly in the Ahrefs study. This is outreach work, not SEO tricks.
8. Fix on-site quotability. Add the specifics the GEO study validated: real numbers, named comparisons, clearly attributed claims, a crisp one-sentence definition of what you are. Make the homepage and key product pages say “X is a Y for Z” in plain text a model can lift.
9. Re-run monthly. Same prompts, same sheet. Three months of trend lines tell you whether retrieval, naming, or corroboration is improving.
10. Automate only when the manual version works. If monthly re-runs get tedious or you manage multiple brands, paid trackers remove the busywork. Peec AI’s current pricing starts at $95/month for 50 prompts across ChatGPT, Gemini, AI Mode, AI Overviews, and Copilot (verified November 2026, with a free trial). Ahrefs Brand Radar adds prompt tracking from $50/month on an Ahrefs plan (small free daily quota included), or its AI Visibility Index at $199/month across 542 million tracked prompts. Semrush sells an AI Visibility toolkit add-on, reported at $99/month per domain billed annually. Otterly’s Lite plan is $29/month for 15 prompts. Profound is the enterprise pick: third-party reviews report tiers around $99 and $399/month plus custom contracts, but it doesn’t publish pricing, so treat those as reported figures.

Three mistakes that waste an audit

• Publishing self-congratulatory listicles. Ray’s data again: cited ~69% of the time without the self-promoter being recommended. Write comparisons that include competitors honestly, or get into someone else’s.
• Buying magic fixes. llms.txt packages, “schema for GEO” upsells, guaranteed-citation retainers. Verified evidence says none of these move AI answers on their own, and OpenAI states no guaranteed placement exists.
• Measuring with a dirty session. Saved memories and personalization skew results. Fresh, logged-out sessions, or you’re auditing a funhouse mirror.

What to do next

If you do nothing else this week: check robots.txt for OAI-SearchBot, run ten prompts three times each, and list every source cited. That’s an afternoon, it’s free, and it converts a vague anxiety (“AI ignores us”) into a worklist (“these five pages need to know we exist”). If your share of voice is near zero, the bottleneck is almost certainly third-party corroboration: start with the sources your competitors already get cited on. Re-run monthly before spending a dollar on tooling, and add a tracker only when the spreadsheet becomes the bottleneck. If legal or financial claims about your business circulate in AI answers, involve counsel before responding publicly: corrections through your own verified pages come first.

FAQ

Why doesn’t ChatGPT recommend my brand?

Usually one of three things: your site blocks OAI-SearchBot so ChatGPT can’t retrieve you in search answers, independent web sources rarely mention you so the model has no corroboration, or your pages don’t give it clear, verifiable facts to quote. OpenAI’s help center also warns citations can be incomplete, outdated, or incorrect, so weak or stale sources can crowd you out. A 10-step audit pinpoints which cause applies.

How do I check if ChatGPT mentions my brand?

Build a set of 15-25 prompts buyers would actually ask (category, comparison, and problem-framing questions) then run each one 3-5 times in fresh, logged-out sessions. Log whether your brand is named, which competitors appear, and which sources get cited. Free to do in a spreadsheet; Peec AI also offers a free single-brand GEO report.

How does ChatGPT choose which sources to cite?

It rewrites your question into search queries, retrieves pages (sites must allow OAI-SearchBot to be included), then ranks results using what OpenAI describes as multiple factors favoring relevant, reliable information. No placement is guaranteed. OpenAI warns results and citations can be incomplete, outdated, or incorrect, so citation counts measure surface area, not accuracy.

Does llms.txt improve AI visibility?

No evidence says so. llms.txt is a proposal for a file that summarizes your site for LLMs, but no major AI company has confirmed using it. Google’s John Mueller said publicly in June 2025 that no AI system currently uses it, and adoption sits around 10% of domains. Adding it is harmless, but treat any tool selling llms.txt optimization as a red flag.

Does schema markup get you cited by ChatGPT?

Not directly, per current evidence. Ahrefs tracked 1,885 pages that added JSON-LD schema between August 2025 and March 2026 and found only marginal changes in AI citations (about +2.2% on ChatGPT). Schema still helps classic SEO and makes content unambiguous, so keep it: just don’t expect it to move AI answers by itself.

How much does an AI visibility audit cost?

Free if you do it manually: a spreadsheet, 20 prompts, and a few hours per month. Paid trackers automate the re-running: Otterly Lite is $29/mo for 15 prompts, Semrush’s AI Visibility toolkit is reported at $99/mo per domain, Peec AI starts at $95/mo for 50 prompts, and Ahrefs Brand Radar runs from $50/mo as an add-on. Enterprise tools like Profound run higher.

Which AI visibility tool is best?

For a single brand on a budget, Otterly ($29/mo) or Peec AI ($95/mo) cover the essentials. If you already use Ahrefs, Brand Radar’s free daily prompt quota and $199/mo AI Visibility Index are efficient. Teams wanting AI data inside an SEO suite suit Semrush’s $99/mo add-on, while enterprises with multi-engine reporting needs look at Profound.

How many prompts should I test in an audit?

Fifteen to twenty-five is the practical sweet spot: enough to see patterns, few enough to re-run monthly. Split them across question types: category, comparison, and problem-framing. Run each prompt several times, since answers vary between sessions and across accounts.

Can I pay ChatGPT to recommend my brand?

No. OpenAI’s help docs state results are ranked by relevance and reliability factors with no guaranteed placement, and there’s no paid inclusion for citations. OpenAI is testing advertising in ChatGPT, but ads are labeled separately from answers. Anyone selling guaranteed ChatGPT citations is selling something they can’t deliver.

How long until AI visibility improvements show up?

Crawl fixes are fast. OpenAI says robots.txt changes take effect in about 24 hours. Content and mention-building changes take longer, since models need to re-retrieve and publishers need to update. Re-run your prompt set monthly and judge on two to three months of trend lines, not single sessions. No source publishes a guaranteed timeline.

Sources

• Searching the web with ChatGPT: OpenAI Help Center
• Overview of OpenAI crawlers: OpenAI Developers
• In 2026, Less Than One Third of Google Searches Still Send a Click: SparkToro
• AI Brand Visibility Correlations: 75,000 Brands Analyzed: Ahrefs
• Which Types of Mentions Correlate With AI Visibility: Ahrefs
• We Tracked 1,885 Pages Adding Schema. AI Citations Barely Moved: Ahrefs
• AI Traffic Increased 9.7x in the Past Year. Ahrefs
• GEO: Generative Engine Optimization: arXiv (KDD 2024)
• Why Calling Yourself the “Best” Could Be Helping Your Competitors. Lily Ray
• AI Search SEO Traffic Study: Semrush
• Peec AI Pricing
• Ahrefs Brand Radar
• Otterly.ai Pricing
• Semrush AI Visibility Toolkit
• How People Are Using ChatGPT. OpenAI Economic Research