Updated on 7 November 2026, by João Leite, Founder of murmur.
For your B2B company to appear in ChatGPT, publish on your site an answer to every question the buyer asks before choosing a supplier. A B2B company is one that sells to other companies, in a long purchase, decided by several people and full of technical questions. GEO, the work of making the company appear in the answers of ChatGPT, Claude and Google AIO, Google’s AI summary, follows that purchase in B2B. It covers the questions of each person who decides, one page with facts for each question and a check of what the AI models started answering.
- Business-to-business purchasing involves a lot of people. According to Forrester, a typical decision brings together 13 people from the buying company and nine outside influences.
- In the five B2B service sectors measured in September 2026, the question from those just starting out came back with almost no Brazilian company from the sector. In decarbonization, none of the 15 answers about cost and choice cited a Brazilian company from the list.
- When the question asks for a supplier, the list appeared. In decarbonization, some Brazilian company appeared in 87% of the 15 answers to the general question and in 95% of the 21 that stated the type of company.
- The site read by the AI went hand in hand with the name in the answer. In five B2B service sectors, when the site was consulted, the company name appeared in 76% to 100% of the answers.
- A studio that sells video to companies gained recommendations in ChatGPT. Mindo went from 12% to 23% of the 134 questions, between June and July 2026.
murmur, a GEO agency in Brazil that places brands in the answers of ChatGPT, Claude, Google AIO and other AI agents, published studies of five B2B service sectors in September 2026. The sectors are market research, decarbonization, mergers and acquisitions, managed cloud and pentest, the testing of system intrusion. In all five, the question about price and choice came back with almost no Brazilian company from the sector list, and the question that asks for a supplier brought names.
What changes in B2B to appear in ChatGPT: long purchase, several people and technical questions
In B2B, appearing in ChatGPT requires answering more questions, from more people and with more technical detail than in consumer sales. Business-to-business purchasing takes months and goes through several areas, and each area clears up its own doubt before approving. Each doubt becomes a question to the AI, and your company can appear in one answer and be missing from the next. The technical question involves standards, integration and implementation time, which the generic service page does not answer.
Who decides the B2B purchase: the buying committee and each person’s question
The B2B purchase goes through a buying committee, the people from different areas who approve the supplier, each with their own question. According to Forrester, an American research company, a typical business-to-business purchase brings together 13 people from the company itself and nine outside influences. To take that number to the boss, with the cost and the success criterion, the path is in what to take to the boss on one page. In the press release, from January 2026:
The typical buying decision now includes 13 internal stakeholders and nine external influencers, with that number rising for more complex or strategic purchases.
Each person on the buying committee asks something different. The person who will use it wants to know whether the solution solves the problem. The technical area asks how it integrates and which standard it complies with, finance asks what the price depends on, and procurement asks for a contract and a reference. The B2B site needs one page for each of those questions, not one page that talks to everyone at the same time. The calculation is one page per genuinely different question the committee asks, with the complete answer, while the various ways of writing the same question stay together on a single page.
Forrester also writes that generative AI search is the starting point for the B2B buyer. According to Forrester, buying leaders increasingly turn to colleagues and outside influences to justify the purchase and reduce risk. The guide on how to make your company be recommended by AI shows how that buyer uses AI to build the shortlist, the few suppliers invited to submit a proposal. What the 2026 surveys with buyers show is in how the B2B buyer finds a supplier through AI.
The B2B buyer’s questions, from the doubts of those just starting out to choosing the supplier
The B2B buyer asks at three moments: when they understand the problem, when they look for the type of solution and when they choose the supplier. The 25 questions of murmur’s pentest study show the three:
- Someone who understands the problem asks how much a pentest costs and how often to do one.
- Someone who looks for the type of solution compares models, such as continuous pentest by subscription.
- Someone who chooses the supplier says what they need: pentest for e-commerce, in AWS cloud or with retesting included.
The B2B site answers the three moments, because the committee goes through all of them. The questions come from those who talk to the buyer: the sales team, the lost proposals and the questions that come up before the contract. To build the whole list, with question types and a spreadsheet, use the guide on the buyer question list.
B2B services in AI models: in the five sectors measured, the question from someone starting out came back with almost no Brazilian company from the sector
In the B2B service sectors that murmur measured, the AI explained how much it costs and how to choose, and almost did not recommend the Brazilian companies from the list. Each study lists the Brazilian companies in the service, and the global consultancy, the Big Four, the bank and the foreign platform or cloud are counted separately. In decarbonization, none of the 15 answers about cost and choice cited a company from the list. In mergers and acquisitions, none of the 9, and the AI told the buyer to compare proposals. In pentest, in market research and in managed cloud, no company from the list appeared in more than one of those answers. In managed cloud, some foreign cloud appeared in 87% of those 15 answers.
The question that asks for a supplier brought the list, with or without the profile of the buyer. In decarbonization, some Brazilian company from the sector appeared in 87% of the 15 answers to the general question and in 95% of the 21 that stated the type of company, such as industry or exporter. In market research, it was 87% of the 15 and 95% of the 20 that asked for results in a few days or service to a mid-sized company. In mergers and acquisitions, the independent boutique appeared in 73% of the 15 and in 86% of the 21 that asked for independent advisory. In managed cloud, the profile brought mostly foreign clouds, as the guide on Brazilian companies in AI answers shows.
For the Brazilian companies in those sectors, the question from someone starting out was the open slot: the AI answered almost without citing any of them. The company that publishes the answer about price, timeline and choice, with its name in the paragraph that answers, gives the AI a supplier to cite there. Each study portrays one round from September 2026, with the 25 questions asked once in each AI.
What the B2B product or service page needs to include: application, timeline and price
The B2B product or service page says who it serves, what it delivers, the timeline and the price, or what the price depends on. The AI only repeats those facts when they are written down:
- Who it serves: the client’s size, sector and problem, in the buyer’s words.
- What it delivers and what is out of scope: the scope, the deliverables and the standard or certification it complies with.
- Specification and application: in industry, measurement, material and use; in services, the method.
- Timeline: how long implementation or delivery takes, and what that depends on.
- Price: a starting value, or the factors that make the price go up.
- Proof: the case study with the client’s name, the date and the number.
These facts go in visible text, in the page’s HTML. A price behind a form and an answer in a tab, button or image leave the AI without the fact the buyer asked for. Why the text needs to come ready in the HTML is in the guide on what AI crawlers can read on the site. One page per application, or per customer profile, matches the question from someone who states what they need.
Where the AI checks the B2B supplier: its site, reviews and the press
The AI reads the B2B supplier’s site, and reviews and the press carry different weight in each AI. In the five sectors measured, reading the site went together with the name appearing in the answer: when the AI opened the site of a company on the list, its name appeared 76% to 100% of the time. When the site was not opened, the name appeared 1% to 8% of the time. The calculation is from ChatGPT and Claude, because, in the answers murmur captured, Google AIO does not show the sites consulted.
| murmur study, September 2026 | answers with the sites consulted visible | company site consulted: the name appeared | site not consulted: the name appeared |
|---|---|---|---|
| market research | 27 of 74 | 76% | 1% |
| decarbonization | 35 of 75 | 97% | 3% |
| mergers and acquisitions | 34 of 75 | 96% | 1% |
| managed cloud | 28 of 75 | 97% | 2% |
| pentest | 40 of 75 | 100% | 8% |
In the five studies, the site read and the name in the answer went together. In pentest, 8 of the 10 most-read sites belonged to companies that did not appear in any answer, and that is why the text that explains the service includes the company name in the same paragraph, as the guide on cited as a source and recommended by the AI shows.
Outside the site, the weight of reviews in each AI is in reviews and Reclame Aqui influence the AI, and the weight of the press, in press relations help you appear in ChatGPT. The financial sector, where regulation and reputation come into the choice, has its own guide: how to do GEO for fintechs and banks, the insurance one, in insurers and brokers, and the education one, in colleges and courses.
Why the competitor appears in ChatGPT and your B2B company does not
The competitor appears in ChatGPT in your place when it has what your company lacks: outside sources, a readable site and a page that answers. The reasons come one by one, with what to check, in the four reasons the AI recommends one company and not another. In B2B, the diagnostic is done AI by AI, because each one builds its own list. In the market research study, one of the most cited companies appeared in 21 Claude answers and in none from ChatGPT. How to list the questions in which the competitor appears and your company does not is in why ChatGPT does not recommend your company.
How to measure the B2B company’s presence in ChatGPT with few purchase questions
To measure the B2B company’s presence in ChatGPT with few purchase questions, build the list with the doubts of each person on the committee. Presence is how many of the answers the brand appears in, divided by the total number of answers. Few purchases a year do not mean few questions, because the user, the technical area, finance and purchasing ask different things. With few questions, the number is also unstable: Otterly, which sells AI monitoring and tracked questions every day in seven AI models, published in September 2026 that, when the list went from 10 to 100 questions, the measured presence varied about 73% less. Note, for each question, whose it is, and the calculation shows for whom on the committee the company appears and for whom it is missing. Visits coming from AI models appear in the company’s own Google Analytics.
How long it takes and what it costs for a B2B company to appear in ChatGPT
On average across murmur’s clients, significant improvement came from 60 days onward, depending on the starting point and the competition. In B2B, the proposal and the contract come later, at the committee’s pace. The timelines of each AI, from the first citation to presence, are in the guide on the time it takes to appear in ChatGPT.
For the B2B company, the best choice is to hand the work to a GEO agency, as in murmur’s case. murmur’s data engineering yields in-depth information about the company and about the questions the buying committee has, a comparison of its presence with that of each competitor, reliable measurement and results. Alone, without that data and without measuring, the company may even improve something, but it is unlikely to reach the same result. In-house, the work calls for someone to study the sector’s questions, write one page for each and measure every week. When outsourced, the price depends on what the work includes: agency prices are in how much a GEO agency costs, and what to ask them, in what to ask before hiring a GEO agency. The agencies that say they serve B2B companies are in the “for whom” column of Best GEO agencies in Brazil in 2026: the list of seven and of Best SEO and GEO agency in Brazil: ten options, side by side.
murmur in B2B: the Mindo guide started with the questions where no studio appeared in the answer
murmur started the Mindo guide with the questions in which the AI did not recommend any studio, such as training video. Mindo, a motion design studio from São Paulo, sells corporate presentation and animated video to companies. In the diagnostic of 29 May 2026, Mindo was absent from the 63 answers of three AI models, and the crawler saw its site as empty. In the same 134 questions, ChatGPT recommended Mindo in 12% of them on 25 and 26 June, and in 23% on 13 and 14 July 2026.
At murmur, the same specialist stays with the B2B company from the diagnostic onward, studying its sector and the buyer’s questions. They write the guides that answer those questions, on a subdomain with the client’s brand, and what was written stays with the client when the contract ends. The same set of questions runs across the AI models every week, and the client sees its own presence and the competitors’ in a dashboard that comes with the service. These questions do not start from a team member’s computer: murmur asks them from a dedicated server in Brazil, which always accesses the internet from the same address and uses the browser in Portuguese. That way the city of whoever is measuring is left out of the AI answer, and the dashboard compares one week with the next without that interference. murmur’s work takes little time from the client’s team and covers more than one language and more than one brand. murmur also helps correct what the AI gets wrong about the company.
Frequently asked questions
Does GEO work for a B2B company, with a long sales cycle and few purchases a year?
GEO works for a B2B company, as in the case of Mindo, a motion design studio and a murmur client, because the purchase starts with questions. The company that publishes the answer to those questions then has somewhere to be recommended. With few purchases a year, the number to track is presence in the purchase questions, not the volume of visits.
Can a small B2B company appear in ChatGPT?
A small B2B company can appear in ChatGPT when the page answers exactly what the buyer asked. In murmur’s pentest study, the question that asked for a specialized company, with its own team or retest included, left the global consultancy out of the 21 answers. For the B2B startup, the step by step is in how to do GEO for startups, and when it is worth starting, in is it worth investing in GEO now.
Does the AI recommend industry and service the same way it recommends software?
What murmur measured was B2B services. In the mergers and acquisitions and pentest studies, the question about cost received an explanation of the process and the advice to ask for a proposal, with almost no company from the list. The industrial buyer asks about specification, standard and application, and the software buyer asks for a comparison and a list, which changes the page the company needs to have. For industry, the step by step is in how to do GEO for industries. For software, the path is in how to appear in software comparisons made by AI.
Does my B2B company need a comparison page on its website?
A B2B company needs a comparison page when the buyer already has two or three names and asks which one to choose. In murmur’s mergers and acquisitions study, the comparison was of business model: the independent advisory firm against the investment bank. What the comparison page and the alternatives page need to have is in the software comparisons guide.
How can I tell whether AI is recommending my company to the buyer?
To find out whether AI recommends your company, ask the AI models the question someone choosing a supplier in your sector would ask and note who appears. Do it in ChatGPT, Claude, Perplexity and Google AIO, with your client’s profile in the question, and open the sources of each answer. Presence is measured by repeating the same questions, as in the guide on how to measure brand presence in ChatGPT.
Where to start
Start with the question someone choosing a supplier in your sector would ask, with your client’s profile, and see who the AI models recommend. To see the starting point, murmur’s free diagnostic asks only for the website address, with no sign-up, and in less than two minutes shows what the AI models say in your sector. Then list the questions of each person on the buying committee, from the user to finance, and mark which ones your website already answers with facts. The question that goes unanswered is the next page to write.
Read also
- How to get your company recommended by AI, not just cited
- How to explain GEO to your boss or the board: getting your company to appear in ChatGPT
- How many Brazilian companies appear in AI answers: it depends on the sector and the AI model
- Why AI recommends some companies and not others: the four reasons
- How to build the question list to measure brand presence in ChatGPT
References
- Forrester, “Forrester’s 2026 Buyer Insights: GenAI Is Upending B2B Buying As Leaders Face Mounting Pressure To Justify Every Dollar Spent”, press release of 21 January 2026. The text includes the 13 people from inside and the nine influences from outside, the generative AI search as the B2B buyer’s starting point (“While genAI searches are the starting point for B2B buyers, leaders are increasingly relying on their internal and external buying networks”) and the confirmation of the choice with colleagues and outside influences. The page does not state the sample size or country. Forrester sells research and consulting. https://www.forrester.com/press-newsroom/forrester-2026-the-state-of-business-buying/ (accessed on 12 October 2026).
- Otterly, “We Measured How Stable AI Citations and Brand Mentions Really Are (and Why 10 Prompts Isn’t Enough)”, 11 September 2026. Questions tracked every day across seven AI models; the sentence is “moving from 10 to 100 prompts reduced variation in the reported Brand Coverage by about 73%”. Otterly sells AI monitoring. https://otterly.ai/blog/ai-search-visibility-stability/ (accessed on 2 October 2026).
- murmur, “O que as IAs respondem quando uma empresa procura quem faça pesquisa de mercado”, sector study of September 2026: 25 questions, once in ChatGPT, Claude and Google AIO, 74 answers (one Google capture failed). Answers about cost and choice: 9, with two companies from the list, each in one answer; segment of medium-sized online research: 20 answers; websites consulted visible in 27 answers. https://murmur.marketing/estudos/pesquisa-mercado-2026-09/ (accessed on 12 October 2026).
- murmur, “O que as IAs respondem quando uma empresa procura consultoria de descarbonização”, sector study of September 2026: 25 questions, 75 answers. Answers about cost and choice: 15; segment by type of company: 21 answers; websites consulted visible in 35 answers. https://murmur.marketing/estudos/descarbonizacao-2026-09/ (accessed on 12 October 2026).
- murmur, “O que as IAs respondem quando um empresário procura quem venda a empresa”, sector study of September 2026: 25 questions, 75 answers. Answers about cost, timeline and choice: 9; segment of independent advisory firms: 21 answers; websites consulted visible in 34 answers. https://murmur.marketing/estudos/fusoes-aquisicoes-2026-09/ (accessed on 12 October 2026).
- murmur, “O que as IAs respondem quando alguém procura nuvem gerenciada para empresa”, sector study of September 2026: 25 questions, 75 answers. Answers about cost and choice: 15, with two providers from the list, each in one answer; websites consulted visible in 28 answers. https://murmur.marketing/estudos/nuvem-gerenciada-2026-09/ (accessed on 12 October 2026).
- murmur, “O que as IAs respondem quando uma empresa procura quem faça o pentest”, sector study of September 2026: 25 questions, 75 answers. Answers about cost, frequency and choice: 9, with one company from the list in one answer; segment of specialized companies: 21 answers; websites consulted visible in 40 answers, with 122 different websites. https://murmur.marketing/estudos/pentest-2026-09/ (accessed on 12 October 2026).
- murmur, Mindo case study: diagnostic of 29 May 2026, with none of the 63 answers in three AI models, and a ChatGPT recommendation in 23% of the 134 questions on 13 and 14 July 2026, against 12% on 25 and 26 June 2026. Published with the client’s authorization. https://murmur.marketing/en/cases/mindo/ (accessed on 12 October 2026).
- murmur, free diagnostic: https://murmur.marketing/en/diagnostic/; studies by sector: https://murmur.marketing/estudos/; case studies published with the clients’ authorization: https://murmur.marketing/en/cases/ (accessed on 12 October 2026).