7 Ways to Get Recommended by ChatGPT: Full 2026 Guide
Seven ways a service firm can get recommended by ChatGPT: what decides who gets named, which parts you control, how to test the answers every month and which shortcuts to refuse.

Published 1 May 2025 · Last updated and fact-checked November 2026
Short answer: To get recommended by ChatGPT, put five facts on your main service page, match your markup to it, earn mentions and reviews that name your service and city, and test monthly. Nobody can promise a place. In an audit of four generative search engines, only 51.5% of generated sentences were fully supported by their citations (Liu, Zhang and Liang, 2023).
Picture a finance director in Seattle with a contract dispute and a board meeting on Thursday. She asks ChatGPT which commercial litigation firms work with software companies her size. Three names come back. Yours is not one of them. Being recommended by ChatGPT is not a ranking you buy or a setting you switch on. It comes from what the model learned, what it finds when it searches, and what other people have written about you.
A buyer in that position may open only the named firms and book a call. The firm that was not named never learns the decision happened.
This guide covers what decides who gets named, which parts you control, and how to check where you stand when you want your professional service firm recommended by AI. I do this work for B2B and professional-service firms at Jackai Agency in Vancouver, and the honest version starts with a limit: nobody can promise you a place in the answer.
Key takeaways
- Models recall well-known names far better than less-known ones, and retrieval helps most with exactly those less-known names (Mallen et al., 2023).
- Five facts on one service page (what you do, who for, where, why you are safe, who to call) are the first fix, and in my projects they take about an afternoon of writing.
- In a benchmark of queries run on generative engines, changing the content itself lifted visibility by up to 40%, with gains that varied by subject (Aggarwal et al., 2024).
- In a Yelp study, about 16% of restaurant reviews were filtered as suspicious, so fake reviews are a poor shortcut (Luca and Zervas, 2016).
- The free AI Visibility Report from Jackai Agency shows what ChatGPT, Claude and Google AI Overviews say about your firm today, in 24 hours.
| # | Way | What you do | Best for |
|---|---|---|---|
| 1 | Know the two paths | Separate memory from live search | Choosing work you can schedule |
| 2 | Put five facts on the service page | State what, who, where, why and who to call | Giving search a quotable answer |
| 3 | Match structured data to the page | Mark up only what readers can see | Making true claims easy to read |
| 4 | Earn brand mentions | Get named beside your service and city | Memory and search together |
| 5 | Collect reviews with real text | Ask at handover, never write them | Saying what your site only claims |
| 6 | Test the answers every month | Same questions, three runs, dated record | Honest measurement |
| 7 | Turn down the shortcuts | Skip fake reviews and hidden prompts | Protecting trust |
In this guide
1. Know the two paths: best for choosing work you can schedule
ChatGPT builds a recommendation from memory, from a live web search, or from both. Memory favours well-known names, so a smaller firm may sit in the weak zone. Search reads pages that exist today. Search is the path you can work on this quarter, so most of the work below targets it.

- Memory: facts stored from training text and recalled on request
- Search: a live web search, then an answer written from the pages it reads
- Weak spot: people and firms that few pages discuss
- Your lever: the pages that exist today
The first path is memory: a language model stores facts from the text it was trained on and can recall them when asked (Petroni et al., 2019). The catch is popularity.
“LMs struggle with less popular factual knowledge, and that retrieval augmentation helps significantly in these cases,” stated Alex Mallen and colleagues (ACL, 2023).
A test of 16 public models found the same weak spot for “tail” entities that few people write about, which is where many smaller firms are likely to sit (Sun et al., 2024). The second path is search. When ChatGPT runs a live web search, it reads pages and writes an answer with links to its sources. This is the path to work on, because it depends on pages that exist today.
Search answers have flaws too. In an audit of four generative search engines, only 74.5% of citations backed the sentence they were attached to (Liu, Zhang and Liang, 2023), so being recommended with the wrong service attached is a real risk. I cover it in the questions below.
“A mere 51.5% of generated sentences are fully supported by citations,” stated Nelson Liu, Tianyi Zhang and Percy Liang (Findings of EMNLP, 2023).
The practical reading is simple. Treat memory as the vendor’s business and treat search as yours. A firm that is named from memory today got there through years of other people writing about it, and a firm that wants to be named next quarter needs pages a search can find, read and match to a buyer’s question.
2. Put five facts on your main service page: best for giving search a quotable answer
Put five facts in the first screen of your main service page: what you do, who it is for, where you work, why you are safe to hire, and who to call. An AI assistant arrives with one question and leaves with whatever it can lift, so a stranger should be able to copy all five without scrolling.

- What you do: in the words a buyer would use
- Who it is for: by industry, company size or situation
- Where you work: including cities you serve remotely
- Why you are safe: the one credential or result that carries the most weight
- How to reach someone: a person’s name, not only a form
W Communications, a Seattle strategic communications and speaking practice on WordPress, is the clearest example from my own work. Four audiences arrived at one homepage: media, clients, event bookers and readers. All four got the same opening paragraph, and the proof sat below the fold. The case study names the problem in one line. Nothing on the site was structured so that a producer, or a machine answering a question about the practice, could lift a fact out of it and use it without asking a person first.
The strongest credential, former Starbucks senior vice president, sat under several paragraphs of narrative, and the topics of expertise were implied in prose, never listed. The rebuild ran to 8 pages across those 4 audiences, each with its own path from the homepage. The biography page now leads with the credential, and the expertise page has one topic per section, each quotable on its own.


No analytics survived the rebuild, so the W Communications case study carries no number, and I will not claim the practice is now recommended by ChatGPT. What changed is narrower and easier to check: the facts exist in a form a producer or a machine can use. Open your main service page on a phone and look only at the first screen. If any of the five is missing, that is the first fix, and it costs an afternoon of writing.
“If any of the five is missing, that is the first fix, and it costs an afternoon of writing,” stated Joshua Jackai, Vancouver web designer.
3. Match structured data to the page: best for making true claims easy to read
Structured data states the facts directly instead of leaving a machine to guess them from layout. Add the markup after the visible text is right, and only for content the page shows. Schema.org markup does not make a claim true. It makes a true claim easier to read, and it must say what the page says.

- Home or about page: Organization and Person
- Service page: Service and Offer, with price and area
- Guide or post: Article, with a Person as author
- FAQ block: FAQPage, matching the visible questions
Bing, Google and Yahoo created Schema.org in 2011, later joined by Yandex, as one shared vocabulary for describing people, places, products and offers in a form machines read (Guha, Brickley and Macbeth, 2016). The researchers who named generative engine optimization put the larger problem bluntly.
“Content creators have little to no control over when and how their content is displayed,” stated Pranjal Aggarwal and colleagues (ACM SIGKDD, 2024).
Their tests still found visibility gains of up to 40% from changing the content itself, and the gains varied by subject. The benchmark did not test ChatGPT or professional-service firms. The levers are small and the effect is real, but nobody gets a guarantee. One caution on FAQ markup: since August 2023, Google shows FAQ rich results only for well-known government and health websites (Google Search Central, 8 August 2023). I still add FAQPage markup because it takes minutes and labels each question and answer, but I do not sell it as a ranking lever. Nobody outside the engines knows how much weight ChatGPT or Perplexity give to markup, so validate it, keep it matching the visible text, and expect modest returns.
4. Earn brand mentions that name your service and city: best for memory and search together
Brand mentions are pages you do not own that name your firm. They matter on both paths. Memory comes from training text, and names few pages discuss are recalled worst. In search mode, the engine pulls pages that match the question, so a trade article naming your service and city can answer the question better than your homepage.

- Association directories: member listings filled in completely
- Client partner pages: lists of outside firms the client works with
- Podcast show notes: guest name, firm and role
- Conference speaker pages: your expertise in someone else’s words
Not all mentions help equally. “Thanks to our friends at your firm” tells an engine almost nothing. “Your firm, a Vancouver employment law practice for tech companies, led the session” gives it a name, a service, a buyer and a city. When you have a say in how you are described, ask for the second version.
When language models choose between web pages that disagree, they lean heavily on how relevant a page is to the question and largely ignore what people find persuasive, such as references or a neutral tone (Wan, Wallace and Klein, 2024). A plain page about litigation for software companies in Seattle matches the finance director’s question, and a polished About page does not.
Use one form of the firm name everywhere. “Smith and Partners”, “Smith & Partners LLP” and “Smith Partners” look like one firm to you, but to a system matching text they can look like three. That rule comes from my own projects, not from a study.
“Use one form of the firm name everywhere,” stated Joshua Jackai, Vancouver web designer.
This is the slow lever. A mention published this month may take a long time to reach a model’s memory, and some never will. Keep earning mentions anyway, because the same pages send calls from people who read them.
5. Collect reviews with real text: best for saying what your site only claims
Third-party reviews move human buyers first, and they give engines something your own site cannot: your claims in words you did not write. A review that names the service, the kind of company and one thing that changed gives an engine a service, a buyer and an outcome to match.

- When to ask: at handover, while the result is fresh
- How to ask: one direct link to the review form
- What to suggest: the service, their industry and one thing that changed
- Where: your Google Business Profile and the one directory your buyers use most
In the classic study of online book reviews, better reviews led to higher relative sales on that site, and review length showed that customers read the text, not only the stars (Chevalier and Mayzlin, 2006). Books are not consulting work, but the reading habit carries over. For AI assistants the evidence is thinner. When I record answers to “who should I hire” questions, in the answer sets I have recorded for clients, the cited sources sometimes include directory and profile listings, the same places reviews live, not only the firms’ own sites.
That is an observation from my work, not a published finding. It is why my six-week engagement includes preparing the listings the engines cite most in a client’s category. “Great to work with!” gives an engine nothing to match. Suggest what to mention, never write it for them, and reply to every review in plain words that restate the service you delivered. Reviews alone will not get a firm recommended by ChatGPT.
“Reviews alone will not get a firm recommended by ChatGPT,” stated Joshua Jackai, Vancouver web designer.
6. Test the answers every month: best for honest measurement
A repeatable test beats a single screenshot, yours or a vendor’s, because the same question can return a different answer on a different run. Write buyer questions from sales calls, run each three times in a clean session, record the answer word for word, and repeat the same set monthly.

- Questions: 10 to 15, from sales calls and enquiry emails, not a keyword list
- Session: a clean one, such as a temporary chat
- Runs: three per question, noting which names appear every time
- Record: who was named, in what order, which sources, what was said about you
- Citations: click each one and check that it says what the answer claims
- Repeat: monthly, with the same questions, in each engine your buyers use
A study of ChatGPT on 829 coding problems found the same prompt often returned different code, and setting the randomness control to zero reduced the variation without removing it (Ouyang et al., 2025). That test was code, not recommendations, but the lesson carries over. One screenshot proves very little, and a dated record of many runs proves a lot more.
Mix category plus city, problem-first questions and “alternatives to” a named competitor. Repeat the set in the other assistants your buyers use, which usually means Claude and Google’s AI Overviews. A firm named in one engine and missing from another can point you to a source one engine found and the other did not. Date the file and keep the questions word for word, because changing the wording changes the answer and ruins the comparison. The point of the record is a fair comparison at week six against week one, with the same questions and the same engines.
“One screenshot proves very little, and a dated record of many runs proves a lot more,” stated Joshua Jackai, Vancouver web designer.
7. Turn down the shortcuts: best for protecting trust
Turn down five shortcuts: fake reviews, hidden instructions aimed at AI, scraped best-of lists, a full site rewrite before any baseline, and copying the competitor who gets named. The first three try to make an engine believe something the rest of the web does not say. Work that lasts makes the web say true things about you.
“Every shortcut tries to make an engine believe something the rest of the web does not say,” stated Joshua Jackai, Vancouver web designer.

- Fake reviews: about 16% of Yelp restaurant reviews were filtered as suspicious
- Hidden instructions: white text aimed at AI is indirect prompt injection
- Scraped best-of lists: paying for lists that exist only to be scraped bets that no human reads them
- Rewrite first: record what the engines say today, or you will never know what worked
- Copying the named competitor: their page answers their buyer, yours has to answer yours
In a study of Yelp, businesses with weak reputations were more likely to commit review fraud, and the platform filtered about 16% of restaurant reviews as suspicious (Luca and Zervas, 2016). A fake review may never be shown at all. Hiding instructions is worse: white text that tells AI assistants to recommend your firm is indirect prompt injection, the method security researchers used to manipulate tools like Bing Chat (Greshake et al., 2023). I will not put a client’s name on an attack technique. Every shortcut tries to make an engine believe something the rest of the web does not say.
The other three are quieter mistakes, and I see them more often than the first two. They come from the same wish for a quick result. Paying for “best of” lists that exist only to be scraped is a bet that no human reads them. Rewriting the whole site first leaves you with no baseline. Copying the competitor who gets named copies an answer written for a different buyer.
Which parts of getting recommended by ChatGPT can you control?
You control your own site, structured data and listings, and you partly control reviews and mentions. You do not control what the model learned in training, whether it searches, or the wording of any one answer. Schedule the first group, influence the second, and measure the third.
Filter the matrix by how much control you have.
| Way | Your control | Most common failure |
|---|---|---|
| 1. Know the two paths | Partial | Memory is the vendor’s; search is yours to feed |
| 2. Five facts on the service page | Full | Facts buried below the fold |
| 3. Match structured data to the page | Full | Markup that says more than the page |
| 4. Earn brand mentions | Partial | Mentions that name no service or city |
| 5. Collect reviews with real text | Partial | Reviews that say only great to work with |
| 6. Test the answers every month | Full | One screenshot treated as proof |
| 7. Turn down the shortcuts | Full | Fake reviews and hidden instructions |
How many answers does one round of testing give you?
One round gives 90 to 135 answers if you test 10 to 15 questions, run each three times and check three assistants: ChatGPT, Claude and Google AI Overviews. A week-six comparison doubles that to 180 to 270. Enter your own numbers to size the afternoon, then keep the same set every month.
Questions about getting recommended by ChatGPT, answered
Why does ChatGPT name a competitor instead of you?
It is rarely the quality of the work, because the model never saw the work. It saw text about it. The firms I see named have more brand mentions, plainer first lines and profiles on the directories engines cite. That is my client pattern, not a study.
What if ChatGPT gets your firm wrong?
Assistants sometimes name an old address, a dropped service, or a similar firm’s facts. The 2023 Liu, Zhang and Liang audit found only 51.5% of sentences fully supported, so expect it now and then. Click the cited source, fix the stale page, and re-run it next month.
Can you see AI referrals in your analytics?
Without setup, an AI visit can land in analytics as direct traffic. My lead attribution method tags links, carries the tags into the CRM, records the converting page and separates AI referrals from direct. In my projects, setup takes two weeks and needed no design change.
What does this cost, and how long before anything moves?
You can do most of this in-house. The Jackai Agency AI Visibility Optimization engagement is CA$4,775 over six weeks, and Lead Source Reporting is CA$2,600 over two weeks. Site changes can be live within days, but when assistants pick them up is not up to me or you.
Find out what ChatGPT says about your firm today
Nothing about your firm has to change first. Same site, same services, same prices. Before you spend a dollar on getting recommended by ChatGPT, see the answer your buyer already sees. The free report runs your buyers’ questions through three assistants and returns a dated page within 24 hours.
The free AI Visibility Report runs your buyers’ questions through ChatGPT, Claude and Google AI Overviews. It names the competitors recommended in your place, lists what the engines could not find on your site as pages, and gives the three fixes I would make first. Send the website and a work email. There is no sales call unless you book one.
If you want the work done, the AI Visibility Optimization engagement runs on your existing site: a baseline of what each engine says, structured data, rewritten service pages, FAQ content, an llms.txt file, the directory listings, and the same questions re-run at week six. None of it buys a place in the answer; it buys the work and an honest measure of it. Every price is on the pricing page.
For the full order I run this in, from the baseline to the week-six re-test, see my nine AI visibility optimization steps. If you are comparing agencies that sell this as GEO or AEO, my seven checks for a GEO or AEO agency list what to ask.
Sources
Peer-reviewed research
- Aggarwal, P., et al. (2024). GEO: Generative Engine Optimization. Proceedings of ACM SIGKDD 2024. https://doi.org/10.1145/3637528.3671900
- Chevalier, J. and Mayzlin, D. (2006). The Effect of Word of Mouth on Sales: Online Book Reviews. Journal of Marketing Research. https://doi.org/10.1509/jmkr.43.3.345
- Greshake, K., et al. (2023). Not What You’ve Signed Up For: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection. Proceedings of ACM AISec 2023. https://doi.org/10.1145/3605764.3623985
- Guha, R., Brickley, D. and Macbeth, S. (2016). Schema.org. Communications of the ACM. https://doi.org/10.1145/2844544
- Liu, N., Zhang, T. and Liang, P. (2023). Evaluating Verifiability in Generative Search Engines. Findings of EMNLP 2023. https://doi.org/10.18653/v1/2023.findings-emnlp.467
- Luca, M. and Zervas, G. (2016). Fake It Till You Make It: Reputation, Competition, and Yelp Review Fraud. Management Science. https://doi.org/10.1287/mnsc.2015.2304
- Mallen, A., et al. (2023). When Not to Trust Language Models: Investigating Effectiveness of Parametric and Non-Parametric Memories. Proceedings of ACL 2023. https://doi.org/10.18653/v1/2023.acl-long.546
- Ouyang, S., et al. (2025). An Empirical Study of the Non-Determinism of ChatGPT in Code Generation. ACM Transactions on Software Engineering and Methodology. https://doi.org/10.1145/3697010
- Petroni, F., et al. (2019). Language Models as Knowledge Bases? Proceedings of EMNLP-IJCNLP 2019. https://doi.org/10.18653/v1/D19-1250
- Sun, K., et al. (2024). Head-to-Tail: How Knowledgeable are Large Language Models (LLMs)? A.K.A. Will LLMs Replace Knowledge Graphs? Proceedings of NAACL-HLT 2024. https://doi.org/10.18653/v1/2024.naacl-long.18
- Wan, A., Wallace, E. and Klein, D. (2024). What Evidence Do Language Models Find Convincing? Proceedings of ACL 2024. https://doi.org/10.18653/v1/2024.acl-long.403
Industry and vendor data (not peer reviewed)
- Google Search Central Blog (8 August 2023). Changes to HowTo and FAQ rich results.
- Jackai Agency (November 2026). Pricing and the W Communications case study: first-party, not peer reviewed.


