Updated on 7 November 2026, by João Leite, Founder of murmur.
To be recommended by AI, a company needs other sites speaking well of it: customer reviews, sector lists and comparisons. The page that explains a subject well becomes a source for the answer, but AI only recommends the company when it finds reasons to trust it. Those reasons are the rating given by customers, the place in sector lists and clear facts on the site, such as the audience and the price. ChatGPT, Claude and Google AIO choose these sources in different ways, and the buyer checks the recommendation before deciding.
- Cited as a source is not the same as recommended. murmur counts the two things separately: in July 2026, the Mindo guide was a source for Google AIO in 36% of answers, and ChatGPT recommended Mindo in 23% of the questions in its market.
- Reviews and lists enter the answers. In an Orbit Media study of 13,184 citations, Claude preferred review sites and lists for purchase questions.
- The buyer checks the recommendation. In a TrustRadius survey, 94% of technology buyers who used AI check the answers at least sometimes.
- AI gets the fact and the label wrong. The correct information, the same on the site and on the company’s profiles, is what it has available to repeat.
- Being cited in Google AIO can bring more clicks. In informational searches with Google’s AI summary, the cited brand received 120% more organic clicks per impression than without a citation, according to Seer Interactive.
SEO helps the page become a source, and it is not enough to be recommended by AI
SEO makes the company’s page findable and helps it become a source for the answer, but it is not enough for the company to be recommended by AI. In many questions, AI looks for pages on the web before answering, and the well-ranked page is part of what it reads. Recommending is another decision. AI recommends the company when it finds, on the page and outside it, a reason to do so.
murmur, a GEO agency in Brazil that puts brands into the answers of ChatGPT, Claude, Google AIO and other AI agents, measures the two things separately, as in the case of Mindo, a motion design studio in São Paulo. In July 2026, Google AIO used the Mindo guide as a source in 36% of the 538 answers measured, and ChatGPT recommended Mindo in 23% of the 134 questions in its market. In murmur’s case studies, only an answer in which the AI names the brand in the text counts as a recommendation. The page used as a source does not count.
The page becomes a recommendation when it gives AI a fact it can repeat as a reason. The reason can be the company’s audience, the price, a case it solved or the customer rating. The guide on why AI recommends some companies and not others shows the four places where a company can appear in an answer. The technical part, such as letting ChatGPT’s bot read the site, is in the guide on how to get your company to appear in ChatGPT.
Reviews and Reclame Aqui enter AI answers about the company
Customer reviews, on Reclame Aqui, on Google and on sector sites, are a source in AI answers about the company. None of the studies murmur has read separates their weight in the recommendation. What each AI shows about reviews, and what nobody has measured, is in the guide on reviews and Reclame Aqui in AI answers. In choice questions, such as “qual a melhor” (which is the best) or “quem contratar” (who to hire), AI looks for what customers and outside sites say about each company. The recommendation comes with the reason it found, such as well-rated service or a price that suits the person asking. A company with no published reviews gives AI less to say about it.
An Orbit Media study shows where each AI looks for that reason. Bill Widmer tracked 13,184 citations in 1,765 answers from ChatGPT, Claude, Gemini and Perplexity. The answers came from 72 purchase questions linked to three brands that sell to businesses, over about three months up to August 2026. Claude preferred review sites and lists. The outside site it cited most was Clutch, which gathers reviews of service providers. ChatGPT preferred primary sources, such as technical documentation, scientific articles and the pricing and help pages of the providers themselves.
A company that wants to be recommended by both AI models needs two fronts. One is having reviews on the sites AI reads. The other is having its own pages with the price, the audience and the case studies written in the text. That is the kind of page ChatGPT looked for on the providers.
What to do with your company’s reviews
- Ask the people who bought for a review, on the site where your buyer checks. It can be the company’s Google profile, Reclame Aqui or a sector review site, such as Clutch for services and G2 for software.
- Answer every review, including the negative one, with what the company did to resolve it. The answer stays public, for the buyer and for AI.
- Bring the rating to your site, with the number of reviews, the date and the link to the page of origin. That way the rating becomes a fact AI can repeat and the buyer can check.
- Use the same name, the same description and the same address on every profile, so AI recognizes that the reviews are about the same company.
- Get into the lists and comparisons that AI uses as a source. Ask ChatGPT the question buyers in your sector ask when choosing, open the sources in the answer and see which ones your company is missing from. The guide on when AI cites the competitor and not your company shows this path.
- Do not buy or invent reviews. The buyer checks, and a fake review that is discovered undermines trust in the real ones.
The buyer trusts the AI recommendation, and checks before deciding
The buyer trusts the AI recommendation to build the list of options, and checks each name before deciding. TrustRadius, a software review site, surveyed 1,862 technology buyers and published the numbers in July 2026. Among them, 63% used AI in the purchase, and 94% of those check the answers at least sometimes. In the same survey, 74% of buyers use reviews to decide the purchase.
G2, another software review site, asked twice in 2026 what shapes the shortlist of providers. In March, with 1,076 software purchase decision-makers, 54% cited AI chatbots and 43% cited review sites. In July, with more than a thousand buyers, review sites moved ahead for the first time, with 38% against 37% for chatbots. How this changes the page of companies that sell to businesses is in how to do GEO for a B2B company, and the case study of software compared side by side is in how to appear in ChatGPT in software comparisons. At what point in the purchase the buyer asks AI, and what they check afterwards, is in how the B2B buyer finds a provider through AI.
The AI recommendation puts the company’s name on the list, and what the buyer finds when they check decides whether they stay. That is why the rating, the comments and the site pages need to tell the same story that AI told.
AI recommendations are not always accurate, and the correction starts at the source
AI recommendations are not always accurate, and the correction starts with the sources it reads, such as the company’s site and profiles. AI can repeat an old price, cite a service the company does not offer or confuse it with another one with a similar name. Forrester, a market research firm, wrote about this in January 2026. According to the firm, AI search tools often deliver incomplete or unreliable information, which creates distrust. The same Forrester writes that the buyer compensates by seeking confirmation in trusted sources.
AI repeats what it finds, and that is why the error is corrected where it reads. The right information needs to be on the website, in the company’s Google profile and on industry sites, the same everywhere. The guide on how to correct wrong information that ChatGPT says about your company shows the step-by-step process.
murmur helps clients correct what AI gets wrong and position the brand the right way in the answers. In June 2026, in the opening diagnostic for FotoRegistro, a photo printing e-commerce, AI was already citing the brand. But it presented it as the cost-benefit option. The problem there was not appearing: it was the label the brand appeared with. The case study tells how FotoRegistro moved to first place in more ChatGPT questions.
Local business: AI finds the website, the Google profile and the reviews
For a local business to be recommended by AI, it is worth keeping the Google profile and the customer reviews from the city complete and recent. A neighborhood business usually has little published outside its website: the Google profile, the rating and the comments. AI answers with what it finds: the company website, the profile and the reviews.
Brazilians already use AI to look for reviews. According to the Startups article on Bain’s Consumer Pulse 2026, searching for product and service reviews was among the main uses of AI. In the same survey, 77% of Brazilians said they use AI, according to a Startups article from April 2026.
A local business needs address, phone, hours and category the same in the profile, on the website and in city guides. It also needs reviews requested from people who were served and replies to comments. The guide on how to appear in ChatGPT when someone looks for a service in your city shows the work of a local business in detail.
What changes in the business when AI recommends your company
When AI recommends your company, it enters the buyer’s list of options before any visit to the website. HubSpot, which sells a product to appear in AI models, surveyed more than 3,000 CRM purchase decision-makers, the customer management system. AI search was the strongest signal of purchase intent, ahead of demo, review site and sales call. And 42% of those buyers used AI search to evaluate vendors.
A company cited in Google AIO can also receive more visits. The AI Overview is the summary written by AI that Google shows at the top of many searches. In informational searches with that summary, the brand cited in it received 120% more organic clicks per impression than when it was not cited. The measurement is from Seer Interactive, an American agency that sells SEO, with 53 brands and 5.47 million searches, from January 2025 to February 2026.
Part of the effect of a recommendation does not show up as a visit coming from AI. Someone who read the recommendation may search for the company by name on Google, and that visit arrives as a search. That is why the figure that matters is how many answers recommend the company, question by question, and not only how many clicks came from AI.
murmur shows who AI trusts in your sector, and what your company is missing
murmur shows which companies AI recommends in the client’s sector, which sources it relies on and what the client’s brand is missing. A specialist dedicated to the client’s brand studies the company and the sector. They analyze what AI answers and what reviews, lists and the website say about the company, and point out what diverges or is missing. The specialist takes care of the pages that give AI the reasons to recommend the brand, with the audience, the price and the case studies written in the text. The pages go out on a subdomain of the client’s brand, and they closely monitor whether the recommendation changes.
In the murmur dashboard, included in the service, the client sees the brand alongside competitors in each question in the sector. The answers that feed the dashboard come from a server in Brazil that murmur uses only for measurement, with a fixed internet address and the browser set to Portuguese, and not from the machine of someone on the team. That way, the answer does not carry the city of whoever is measuring, and it is possible to put the brand’s presence in one week next to its presence in another. murmur charges from R$ 5,000 per month, with no minimum contract term. The amount varies with the number of business lines, languages and brands, and with the complexity of the market. The work has no cap on questions monitored or pages published.
Frequently asked questions
Why don’t Google AIO and ChatGPT recommend my company?
Google AIO and ChatGPT do not recommend a company without reasons to trust it: customer reviews, industry lists and clear facts on the website. The recommendation is also taken away by a bot that cannot read the website and by the name being confused with another company. When AI recommends a competitor, the guide on when AI cites the competitor and not your company shows what to do. Anyone who needs to explain the problem to the boss will find the answer in the guide why ChatGPT does not recommend the company, for the head of marketing.
How do I get my brand cited by AI models?
For the brand to be cited by AI models, publish one page per buyer question, with the answer in the first sentence. A new page is only justified when the question truly changes, and it carries the full answer, while the same question written another way stays on the page that already exists. The page needs to be read by each AI’s bot and include facts that can be verified, such as numbers, dates and names. Being cited is the start, and being recommended also calls for what other sites say about the company. The guide on how to get your company cited by AI models shows the differences between AI models.
Why does my brand appear in ChatGPT but not get cited?
The brand appears and is not cited when ChatGPT knows the name but does not find one of its pages that answers the question. In that case, it shows up in passing, in a list of names, and the source and the recommendation go to other companies. The fix is to have the page that answers, with the company name in the answer excerpt, and reviews that give a reason to recommend it. The guide on why AI recommends some companies and not others shows the difference between being mentioned, cited as a source and recommended.
How do I know if my brand appears in ChatGPT?
To know if the brand appears in ChatGPT, repeat the buyer’s questions on different days and note whether it is recommended or only cited. The answer changes from one conversation to another, and the count over several rounds is worth more than a single test. Also note which companies were recommended and which pages appeared as a source. The guide on how to measure the brand’s presence in ChatGPT shows the step-by-step process.
Where to start
Start with your buyer’s choice questions, the ones about which company to hire for their case. Ask each one to ChatGPT, Claude and Google AIO and note whether your company is recommended, only cited or left out. Then open the sources of the answer and see what reviews and lists say about you and about whoever was recommended. The free diagnostic from murmur takes your website URL and shows what AI models answer when asked the questions in your market. It does not ask for registration and takes less than two minutes. murmur’s sector studies show which companies AI models recommend in markets such as online therapy, exchange agency and European citizenship consultancy.
Read also
- Why AI recommends some companies and not others: the four reasons
- How to get your company to appear in ChatGPT: the step by step, with no registration or ads
- When AI cites your competitor and not your company: what to do
- How to correct wrong information ChatGPT says about your company: the step by step
- How to measure brand presence in ChatGPT: the step by step
References
- Orbit Media, “I Tracked 13,184 AI Citations Across 4 LLMs: Here’s What I Learned”, by Bill Widmer, 2 September 2026. https://www.orbitmedia.com/blog/ai-citation-sources/ (accessed on 28 September 2026). 72 purchase questions tied to three brands that sell to businesses, over about three months up to 23 August 2026. The author states that one of the three brands, Semrush, is his client.
- G2, “The Answer Economy: G2’s 2026 AI Search Insight Report”, March 2026. https://learn.g2.com/g2-2026-ai-search-insight-report (accessed on 28 September 2026). 1,076 software buying decision-makers from North America, Europe, the Middle East, Africa and Asia-Pacific. G2 makes its living from software reviews.
- G2, “New G2 Research: AI Is Reshaping How B2B Software Deals Are Won and Lost”, 22 July 2026. https://company.g2.com/news/buyer-behavior-2026 (accessed on 28 September 2026). More than a thousand software buyers and buying decision-makers.
- HG Insights, “TrustRadius 2026 B2B Buying Disconnect Report Reveals AI Has Changed How Buyers Research, But Not What They Trust”, 15 July 2026. https://hginsights.com/news/trustradius-2026-b2b-buying-disconnect-report-reveals-ai-has-changed-how-buyers-research-but-not-what-they-trust/ (accessed on 28 September 2026). 1,862 buyers and 444 technology vendors. TrustRadius makes its living from software reviews.
- Forrester, “Forrester’s 2026 Buyer Insights: GenAI Is Upending B2B Buying As Leaders Face Mounting Pressure To Justify Every Dollar Spent”, 21 January 2026. https://www.forrester.com/press-newsroom/forrester-2026-the-state-of-business-buying/ (accessed on 28 September 2026). In the original: “they often deliver incomplete or unreliable information, creating mistrust”.
- HubSpot, “HubSpot AEO data: Buyers using AI search are more likely to purchase”, updated on 15 September 2026, with research from January 2026. https://www.hubspot.com/company-news/aeo-data-buyers-using-ai-search-more-likely-to-purchase (accessed on 28 September 2026). More than 3,000 CRM buying decision-makers worldwide. HubSpot sells an AEO product.
- Seer Interactive, “AIO Impact on Google CTR: 2026 Update”, by Tracy McDonald, Hannah Cooley and Marketa Williams, 24 April 2026. https://www.seerinteractive.com/insights/aio-impact-on-google-ctr-2026-update (accessed on 28 September 2026). 53 brands, 5.47 million searches and 2.43 billion organic impressions, from January 2025 to February 2026. Seer is an agency and sells SEO and GEO.
- Startups, “Brasil supera EUA no uso de IA e aponta para próxima ruptura no varejo digital” (“Brazil overtakes the US in AI use and points to the next disruption in digital retail”), by Gabriela Del Carmen, 3 April 2026, about Bain’s Consumer Pulse 2026. https://startups.com.br/negocios/inteligencia-artificial/brasil-supera-eua-no-uso-de-ia-e-aponta-para-proxima-ruptura-no-varejo-digital/ (accessed on 28 September 2026).
- murmur, case studies published with client authorization: Mindo, https://murmur.marketing/en/cases/mindo/; FotoRegistro, https://murmur.marketing/en/cases/fotoregistro/ (accessed on 28 September 2026).
- murmur, free diagnostic and sector studies: https://murmur.marketing/en/diagnostic/ and https://murmur.marketing/estudos/ (accessed on 28 September 2026).