Get cited by ChatGPT and other AIs without buying links
On commercial queries, what models cite most are comparison lists, and in professional services those lists are written by third parties 80.9% of the time. Publishing your own ranking where you come first therefore carries little weight, and buying a link carries even less. What moves the needle is being named in the documents the model considers independent.
- On commercial queries, 40.86% of cited sources are comparison lists, measured across more than a million citations.
- In professional services, 80.9% of the most cited lists are written by third parties rather than by the companies that appear in them.
- Brand mentions across the web correlate with citations three times more strongly than inbound links.
- Models also rely on non-editorial sources that nobody maintains: business profiles, professional directories, associations, public data.
- The work is done through outreach and updates, not through purchases.
On this page
Third-party mention
A third-party mention is the appearance of your company's name in a document you neither wrote nor paid for: a comparison published by a media outlet or an industry blog, a professional directory, an association's member list, a press article, a public database. It differs from an inbound link in that it does not need to be clickable to count: what a model picks up is the name tied to a context.
What the data shows, and its scope
The most solid analysis available on the subject covers 1,056,727 citations recorded in 75,000 answers from ChatGPT, Google AI Mode and Perplexity. It was published in March 2026, and it sorts the cited sources by content type and by search intent.
Two results matter for a service company. The first: on commercial queries, the ones where a person compares options before deciding, 40.86% of cited sources are comparison lists. The second: in professional services, among the thousand most cited URLs, 80.9% of those lists come from third-party sites that rank brands, against 19.1% written by the companies themselves.
The scope is worth naming. The analysis covers three models and a given period, it does not look at Quebec specifically, and the breakdown varies by industry. What it establishes solidly is the order of magnitude: on this type of query, the independent list largely dominates brand content.
A second source points the same way by another route. The review of 23 citation factors published on May 7, 2026 measures a correlation of 0.664 between brand mentions on the web and citations obtained, against 0.218 for inbound links. The ratio is about three to one, in favour of being named.
A correlation is not a cause. Companies that are mentioned often are also the ones that are well known, and brand awareness explains part of the gap. What remains usable is the direction: the actions that produce mentions are worth more than the actions that produce links.
Finding the comparisons that already exist in your category
The work starts with a review, and it is done by hand in half a day. The goal is to end up with a list of named documents, not a general impression.
Ask the questions your buyers ask
In two different models, ask for the best vendors in your category in your region, then for a comparison between two or three well-known names. Note the sources cited under each answer.
Repeat the review on three separate occasions
Answers vary from one run to the next. A source that comes back three times counts. A source seen only once proves nothing.
Cross-check with classic search
Search for wordings such as best agencies, comparison and top list in your industry and your region. The documents that rank well in Google often turn up among the cited sources.
Sort the list by accessibility
Three columns: the ones where you already appear, the ones that accept applications or updates, and the ones that are closed. The middle column is your work plan.
This review generally produces between ten and thirty documents for a Quebec industry. That is a manageable number, and it is what makes this project different from a link-building campaign with no end.
Understanding the inclusion criteria, then doing the outreach
Every comparison has an entry rule, and it is rarely secret. Three families come up, each with a different approach.
| Type of comparison | What decides inclusion | The approach |
|---|---|---|
| Directory or review platform | A complete listing and verified reviews | Create the listing, fill it in completely, ask your customers for reviews |
| Editorial list from a media outlet or a blog | The judgment of the person who writes it | Contact that person with a verifiable element they do not have: a case with numbers, a piece of data, a specialty |
| List from an association or an industry body | Membership and an up-to-date profile | Join, then check that the profile shows your current services |
The middle approach is the one that puts people off, and it pays off the most. A person who maintains a comparison wants the document to stay accurate: pointing out an omission with a verifiable element is a service you render, not a favour you ask.
What makes the approach acceptable is what you have to show. A measured result on a real mandate, a specialty nobody else covers in the region, a piece of data that you publish and nobody else has. It is the same material that makes your own pages citable, and that is why the two projects are run together.
Paying to appear in a list that does not disclose it brings back the problem of the bought link, with the added risk of a ranking your buyers will judge. A paid directory that clearly displays its commercial nature is advertising and is judged as such: how many people in your market consult it.
The non-editorial sources that models use
Comparisons are not the only material. Models rely on a set of documents nobody thinks of maintaining, because they never served classic SEO.
- Your Google Business Profile: it has become a cited source in its own right, to the point that google.com has moved to second place among the domains cited in AI Mode. Covered in our note on the Google Business Profile as a cited source.
- Professional directories and professional orders: they give the model outside confirmation of what you do and of your qualifications.
- Member lists of industry associations: often well indexed and rarely kept up to date by the members.
- Public data: the business registry, public calls for tenders, grants. They describe your company in terms you did not choose.
- Your job postings: they tell the model which services you are developing and what size you are, sometimes more clearly than your about page.
- Your profile and your partners' profiles on professional networks: these are strong pages, often cited, and they easily end up contradicting the site.
These sources have one thing in common: they get corrected without a media budget, and they stay outdated for years when nobody looks after them. The consistency between them is what decides how a model understands your company, a topic covered in our analysis of what an AI understands about your business.
What can be measured and what cannot
This project takes a long time to get going, which makes it important to frame the measurement from the start. Three elements can be measured honestly.
The first is the number of third-party documents where your name appears, recorded quarterly. It is a count, not a score, and it moves up in steps rather than steadily.
The second is how often you appear in the models' answers on your commercial questions, judging on repetition rather than on a single run. The method and the sources available are detailed in our comparison of AI visibility measurement sources.
The third is the traffic referred by assistants, which remains low everywhere and must not serve as the judge. What happens upstream of a request cannot be read in a traffic report: a person who saw your name in an answer often arrives through a brand search, several days later.
Where to start depending on your situation
The first move depends on what already exists around your name, not on your budget.
- If you appear in no comparison: start with directories and associations, which have objective criteria and an open door.
- If you appear in a few: focus the effort on the documents the models actually cite in your reviews, and leave the others.
- If you appear with outdated information: correct before adding. A false mention costs more than a missing mention.
A proposal that promises mentions in a number of publications set in advance is selling a volume of paid placements. One that starts with a named review of the comparisons in your category is working with what exists.
Drawing up this review and prioritizing the outreach is part of what we cover in a 90-minute consultation.
What stays in-house is the material: the list of associations you belong to, of customers who would agree to give a testimonial, of measured results you can publish and of people in your industry with whom you already have a relationship. No agency can produce it for you, and it decides the speed of the project. What can be delegated is reviewing the comparisons, reading their criteria, writing the outreach and tracking the mentions obtained each quarter. The order we apply is to correct what is false first, then maintain what exists, and aim for editorial lists last, because it is the only one of the three where the answer does not depend on you.
Checking your mentions
This project is the workable counterpart of the link budget, whose calculation is detailed in our analysis of buying backlinks. The way an assistant shortlists vendors is covered in AI-assisted vendor shortlisting. The authority criteria Google evaluates are in our analysis of E-E-A-T, and what brand awareness changes for everything else is in our analysis of branding.
Getting named where your buyers search is at the heart of the Strengthen your visibility in AI answers goal.
Already running a marketing team? See how we plug in as reinforcement on answer engine optimization.
Frequently asked questions about brand mentions and AI citations
How do I show up in ChatGPT's recommendations?
By being named in the documents the model consults to answer. On commercial queries, these are mostly comparison lists written by third parties, supplemented by directories, business profiles and association lists. Publishing your own ranking or buying a link does not get you into those documents.
How do I get included in a vendor comparison?
By contacting the person who maintains it with a verifiable element they do not have: a measured result, a specialty nobody else covers in the region, or a piece of data that you publish. An editorial comparison wants to stay accurate, so an omission pointed out with facts is acceptable.
Do brand mentions influence AI answers?
They are more strongly associated with them than links are. The review of citation factors published on May 7, 2026 measures a correlation of 0.664 for mentions against 0.218 for inbound links. Correlation is not a cause, but the direction is consistent from one study to the next.
Should you publish your own vendor comparison?
It is useful for the reader and not very effective for citation. In professional services, 80.9% of the most cited lists are written by third parties. A brand comparison has its place when it is honest about its limits, but it does not replace a presence in independent documents.
How long before you see an effect?
The project moves up in steps rather than steadily, and the first results are read quarterly. Directories and associations have an effect within a few weeks, editorial lists within a few months, and repetition in the models' answers follows with a further lag.
How do I measure mentions without a paid tool?
By counting each quarter the third-party documents where your name appears, and by asking your commercial questions to two models on three separate occasions to judge on repetition. The free citation measurement sources round out this review on Microsoft and Google surfaces.
- Tom Wells for Peec AI, The content types most cited by LLMs, 1,056,727 citations recorded in 75,000 answers from ChatGPT, Google AI Mode and Perplexity, March 2026.
- Zyppy, AI citation ranking factors, 23 factors scored, May 7, 2026.
- Google Search Central, AI features and your website, accessed September 2026.
Geneviève puts the strategy for your engagement into action. She leads all our web development projects: Shopify, WordPress and the new ways of building a site with AI. She manages our team of developers and translates your business needs into technical language. She runs your organic search (SEO), your visibility in AI answers (GEO) and your site's conversion rate optimization (CRO). Her work is at the heart of three goals: Attract customers with SEO and AI, Improve your site's conversion, and Strengthen your visibility in AI answers. With Gabriel, she also builds the landing pages for your advertising campaigns. She writes mainly about SEO, AI visibility and web design.
About Falia →
SEO
Paid advertising
Conversion
AI visibility
Web design