Being mentioned by ChatGPT is a threshold. Being named first, second or third is the outcome that moves revenue. ChatGPT answers to buying questions usually list three to seven options, and the first three get the fuller descriptions, the “if I had to pick” framing, and the buyer’s next Google search. Everything below that is context the buyer skims. If your visibility program reports mentions without position, it is measuring the wrong thing.
What a ChatGPT answer actually looks like to a buyer
Ask ChatGPT “what is the best project management tool for a 15-person design agency” and you will typically get an introductory sentence, a numbered or bulleted list of options with one to three sentences each, and a closing recommendation that often picks one or two. The structure is remarkably consistent across categories.
Within that structure, order carries meaning. The first option is usually the one the model is most confident fits the exact question. It gets the longest description and is most likely to be repeated in the closing sentence. Options five and six get a clause each and a qualifier like “also worth considering.” A buyer who asked a specific question reads this as a ranking, whatever the model intended.
Why the first three names take almost all the action
Buyers do not shortlist six brands from a chat answer. They take one or two names to the next step, which for most people is a Google search. A 2026 Idea Grove survey of 1,000 US consumers found that 45% immediately Google a brand recommended by AI, and only 2% would buy from an unfamiliar brand on the AI recommendation alone. The verification step is where the recommendation becomes a lead, and buyers verify the names at the top.
The same survey found 18% go to review sites next. Again, they look up the one or two brands that stood out, not the full list. A brand in position six might as well be absent for that buyer. This is why the ChatGPT buyer journey differs from the Google journey: the shortlist is formed inside the answer, before any click happens.
How ChatGPT decides the order
There is no index and no ranking algorithm in the Google sense. The order emerges from generation, and the model tends to produce first the brand it associates most strongly with the specific prompt. Several factors push a brand toward the top of that association:
- Specificity of fit. Sources describe the brand as the choice for exactly this buyer, size or constraint, not just as a member of the category.
- Consensus. Many independent sources agree. A brand named by one listicle sits lower than one named by twelve.
- Recency. When the answer is grounded in search, recent coverage dominates. A 2026 University of Toronto audit found cited content skewed recent, with a median age of 62 days for Claude in consumer electronics against 130 days for Google.
- Entity clarity. The model names confidently what it understands clearly. Blurry entities get hedged mentions lower down.
- Constraint matching. “Under $50 a month,” “HIPAA compliant” or “for UK businesses” reorders the list toward brands documented with that attribute.
For the deeper explanation, see the mechanics of a ChatGPT recommendation. For position specifically, the practical point is that generic category presence gets you mentioned; specific, consistent, recent descriptions for the exact prompt get you to the top.
Position is a distribution, not a number
Because each answer is generated fresh, your position varies between runs of the same prompt. Reading one answer and concluding “we are number two” is like checking one visitor’s screen and declaring your Google rank. The honest measurement is a distribution across many answers.
Illustration: suppose a prompt is run 100 times. Your brand might appear in position 1 in 18 answers, positions 2 to 3 in 34, position 4 or lower in 13, and be absent in 35. Your top-three rate is 52%, your mention rate is 65%, and your average position when mentioned is about 2.6. Each of those numbers tells a different story, and the top-three rate is the one that tracks commercial outcome most closely. Our methodology explains why around 100 answers gives a usable band and why fewer does not.
A table of what each position tends to mean
| Position in answer | How the model typically frames it | What the buyer typically does |
|---|---|---|
| 1 | Default or strongest fit, longest description, often repeated in the closing line | Googles it, reads reviews, often books or signs up |
| 2 to 3 | Credible alternatives with specific strengths | Compares against position 1, may visit |
| 4 to 5 | “Also worth considering,” one clause each | Skims, rarely acts |
| 6 or lower | Listed for completeness, sometimes with a caveat | Ignores |
| Not mentioned | Absent | Never learns you exist for this question |
Position across a cluster, not a single prompt
Position only makes sense measured across a prompt cluster, because the ordering shifts with phrasing. A brand can hold position 1 for “best X for enterprises” and position 5 for “affordable X for startups,” reflecting exactly how sources describe it. Reporting average position across 40 prompts hides that; reporting position per prompt shows which buyer language you own and which you are losing.
Illustration: suppose a cluster of 36 prompts shows you in the top three for 12 of them, mentioned but lower for 10, and absent for 14. The 10 “mentioned but lower” prompts are usually the fastest wins, because the model already associates you with the category and only needs stronger, more specific evidence for those phrasings. How to build and read the cluster is covered in the prompt cluster guide.
Ads do not buy position inside the answer
A reasonable question in 2026 is whether position can simply be bought. It cannot. OpenAI announced ads in January 2026 for logged-in US adults on the Free and Go tiers, and third-party tracking reports that they appear as a “Sponsored” card under the answer’s sources, with no self-serve ads manager as of July 2026 and paid tiers excluded entirely. The brands named inside the answer are organic.
That separation matters for how buyers read the answer. The list is understood as the model’s judgment, and a labeled card below it is understood as an ad. What changed with the arrival of ads, and what did not, is covered in ads in ChatGPT vs organic recommendations.
How to move from mentioned to top three
Moving up is not a trick; it is a matter of what the model can confirm about you for the specific question. In our audits the brands that climb share a pattern: their own site states the exact fit in plain words (who it is for, at what size, at what price, with which integrations), independent sources describe them with the same specifics, their reviews are concentrated where their category compares, and their coverage is recent enough to appear when the model searches.
The brands that stall are present in the category but described generically. “A popular CRM” gets you position five. “The CRM most often recommended for agencies under 20 people because of its reporting and pricing” gets you position one for that prompt, provided enough sources say it. For our done-for-you service we target top-three position and mentions in up to 90% of a cluster’s prompts, with prompt-level reporting you can reproduce in ChatGPT yourself; how we do it is explained on your call. To see where you stand today, find out what ChatGPT tells buyers about you.
What to do next
- Run your five most commercial prompts ten times each and record your position in every answer, not just whether you appear.
- Separate “mentioned but below position 3” prompts from “absent” prompts; the first group is your fastest opportunity.
- Set up position tracking across the full cluster using the approach in the ChatGPT visibility measurement guide.
Frequently asked questions
Does ChatGPT rank brands like Google ranks pages?
Not in the same way. There is no fixed index with a stable order. Each answer is generated fresh, and the order of brands reflects how strongly the model associates each one with the specific question, with the first-named brand usually being the one it is most confident about. The order can change between runs, which is why position is measured as an average across many answers, not read from one.
Why does position matter if the buyer reads the whole answer?
Because buyers rarely treat the list as equal. The first brand named often gets the most descriptive text, is framed as the default choice, and becomes the one the buyer Googles next. A 2026 Idea Grove survey found 45% of consumers immediately Google a brand recommended by AI, and they typically verify one or two names, not six.
How is position measured in a ChatGPT audit?
Each collected answer is read and the brands named are recorded in order. Your position is your rank in that list, or "not mentioned." Across around 100 answers per prompt you get a distribution: for example, position 1 in 20% of answers, 2 to 3 in 35%, 4 or lower in 15%, absent in 30%. Average position when mentioned and top-three rate are the headline figures.
Can I pay to be the top recommendation in ChatGPT?
No. The recommendation inside the answer is organic and not for sale. OpenAI's ads, which began in early 2026 for free-tier US users, appear as a labeled Sponsored card below the answer's sources, separate from the brands named in the answer itself. Position within the answer is earned through the signals the model weighs.
What usually separates position 1 from position 5?
Specificity and consensus. The top-named brand is typically the one most sources describe as the fit for that exact buyer and constraint, with recent coverage and consistent descriptions. Lower positions are brands the model knows belong to the category but has weaker or more generic evidence for. Moving up means being described specifically for the prompts that matter, not just being in the category.
Find out what ChatGPT says about your brand before you spend anything.
Apply in about two minutes. We check your keywords, then run your real buyer prompts live on a 30-minute call. If we cannot create the result for your keywords, we tell you on the call, not after an invoice.
See if my keywords qualify →