How to increase your brand's visibility in ChatGPT and Google AI Overview

You increase a brand's visibility in ChatGPT and Google AI Overview on two tracks: through your own website and through a presence in the sources AI cites. AI systems have no opinion of their own about businesses. They summarise what they find on the web. According to a study of brand image in AI search engines, a brand's own site appears in 51–73% of answers about it, but makes up only 11–24% of cited sources. The rest are review sites, industry portals, forums and competitors' sites. That is why an effective strategy combines SEO for your own site with Digital PR in the places AI uses most often.

Where does artificial intelligence get information about businesses?

AI systems build an answer about a brand mainly from web pages they find while searching. Knowledge from model training only supplements it.

In simplified terms the process has three steps. AI breaks the question into several searches, picks sources and passages from the results, and finally writes the answer. There is an element of randomness at every stage. We cover the details in our article on how AI search works.

One Polish study from 2026 covered 15 brand analyses and more than 73,000 AI answers from six platforms. The results show how heavily answers rely on external sources:

• 9 out of 10 answers about a brand cite at least two sources,
• a typical answer cites 4 different domains,
• only 3.8% of answers give no source at all.

A small group of domains accounts for a large share of citations. In the eight most recent analyses from this study, 35 of 10,471 domains produced 42% of all citations. At the top of the list were pl.trustpilot.com, opineo.pl, ceneo.pl, gowork.pl, reddit.com and facebook.com.

AI also reaches for old material. In the study, Perplexity cited among other things a blog post from 2013, and ChatGPT a review from 2018. Outdated reviews can therefore shape today's answers.

Why does AI recommend your competitors and skip your brand?

AI most often recommends competitors because it finds more about them. A brand's recognition and scale weigh more than good reviews alone.

In questions without a brand name, AI mainly names the biggest players as the first choice. This was the share of the giants in first recommendations in the individual analyses:

A good rating does not guarantee a presence in answers. In one analysis a craft manufacturer had a rating of +2.83 on a scale from –5 to +5, and its competitor +2.87. Even so, the manufacturer appeared in 6% of answers and the competitor in 46%.

The relationship also works the other way. A service company rated +0.78 appeared in 39% of answers. Its best-rated competitor, rated +1.20, appeared in only 15%.

Rankings on external portals matter a lot. In one analysis AI cited 14 rankings, and in 11 of them a competitor was placed above the brand studied. In every answer based on those 11 rankings, AI put the competitor in first place.

The data shows that lack of presence is a more common problem than a weak rating. Across 90 analysed purchase decision areas, the “they don't know us” problem affected 28 areas. A rating worse than competitors' affected 15.

How does AI pick brands to recommend?

AI picks brands to recommend based on what it finds in the sources it cites. It does not rely on its own judgement. Several factors decide the result at once, and every answer contains an element of randomness.

Recommendations are unstable, so only large-scale measurement, described in the section on research methods, tells you where a brand stands.

Current and consistent information on frequently cited sites reduces the risk that AI bases a recommendation on old material.

How do you prepare a website for AI search?

You prepare a site for AI search the same way as for classic SEO. In its guide on generative search features, Google states plainly that good SEO practices still apply.

Your own site is the most important starting point. AI cites it in most answers about a brand, as the figures in the introduction show. Even so, only 4–11% of answers relied solely on company websites.

Information architecture and topic clusters

AI breaks the user's question into several searches. A site has a better chance of being cited when each important topic has its own subpage linked to the rest of the site.

In practice this means grouping content around the topics customers care about. For a store these will be, for example, pages on delivery, returns, warranty and product comparisons. Google also recommends limiting duplicate content.

Content that answers users' questions

The study's authors asked AI about purchase decision areas, including price, promotions, delivery, customer service and complaints. This is a good list of topics a brand's site should answer clearly and specifically. Google Trends helps assess seasonal interest in these topics.

An example from the report: AI rated a brand as expensive, but rated the quality and durability of its products highly. The recommendation was to justify the price with specifics: durability, warranty terms and the cost over the whole period of use.

Google stresses that unique and useful content is what counts. You do not need to write in a special way for AI or split texts into small chunks. Google has emphasised content quality at least since the Panda algorithm.

Structured data and technical accessibility

The site must be accessible to the crawlers that collect data for AI search engines. Without that it will not appear in answers, regardless of content quality.

OpenAI notes that blocking OAI-SearchBot removes a site from ChatGPT Search answers. The GPTBot crawler collects training data and does not decide visibility in search. OpenAI also publishes the IP ranges of its crawlers, which are worth allowing in the server firewall. Bing's index, which Copilot uses, is also one of the sources of DuckDuckGo results.

According to Google, structured data is not required for generative search features. It is still worth using as part of regular SEO, because it gives access to rich results. Google also ignores files such as llms.txt. It does recommend a good page experience, which you can check in tools such as PageSpeed Insights and Google Lighthouse.

Credibility and consistency of brand information

AI combines data from many sources, so inconsistent information weakens a brand's image. Prices, delivery terms and contact details should be the same on the website and on external sites.

It is also worth keeping complete listings in business directories, such as Oferteo, PHD and Aleo, with data consistent with other sources. Offers on marketplaces such as Allegro, Amazon or Booking should be complete too.

Outdated or incorrect information calls for a different response than a genuine problem. The report recommends contacting the publisher to update it and publishing current data on your own site. An SEO audit helps assess the state of the site.

How do you optimise a site for Google AI Overview?

To appear in Google AI Overview, a page must meet the requirements of a regular search result with a snippet. Google does not require separate optimisation for AI.

Google Search Central's guide sets out what you need to do:

• Take care of indexing. The page must be accessible to Googlebot, indexed and eligible to be shown with a snippet.
• Do not turn off AI features in Search Console. The site must be included in generative AI features in search.
• Publish unique and useful content. Google discourages duplicated content that adds nothing and recommends limiting duplicates.
• Look after the technical side. Follow SEO best practices for JavaScript and provide a good page experience.
• Provide business and product data. Stores can use Google Merchant Center, and local businesses Google Business Profile.
• Measure the results. The generative AI report in Search Console shows how the site performs in these features.

Special files and tags for AI are not needed, as the part on structured data and technical accessibility explains.

Google also notes that even meeting all the requirements does not guarantee being shown. AI Overview also draws on more than the brand's own site. In the study cited, only 14% of domains appeared in all five AI Overview answers to the same question. As many as 40% appeared in only one. That is why optimising your site is worth combining with a presence on the sites Google cites in your industry.

How do you add a business to ChatGPT through Business Hub?

You add a business to ChatGPT through Business Hub, a dashboard where you can claim or create a business profile. OpenAI launched the Businesses in ChatGPT feature in beta on 10 November 2026.

It is the counterpart of Google Business Profile, with one important difference. Business Hub also accepts businesses that operate only online, which Google does not accept in its profile.

The business description matters, because according to the form ChatGPT may use it in conversations. It is worth writing it in the language customers use to describe their needs.

A profile does not guarantee a higher position in answers. OpenAI notes that ChatGPT still creates recommendations and comparisons on its own. The profile may also contain information from other sources, so the data consistency described in the part on brand credibility matters.

You can give collaborators and an agency access to the profile. Stores can also upload a file with products and prices following OpenAI's specification.

The feature is new and has limitations. Claiming an existing store or restaurant listing works only in the US for now. New profiles for physical businesses may land on a waiting list, while businesses operating through a website go through the process straight away. What Business Hub offers in Poland is summarised in our ChatGPT Business Hub glossary entry.

How do reviews, publications and Digital PR affect a brand's presence in AI?

Reviews and external publications make up most of the material AI uses to build an answer about a brand. In the answers analysed, more than three quarters of cited sources were sites the brand does not control.

The importance of each type of site depends on the industry. For large stores, review sites made up 15–39% of sources. For the producer and the craft manufacturer, industry portals and blogs dominated, with a 28–30% share.

Digital PR is best started with a map of the domains AI cites in your industry. You can build the map in four steps:

• Collect the sources. Find out which sites AI cites in answers to your customers' purchase questions.
• Rank them. Order the list by the number of answers in which each domain appears.
• Categorise the domains. Split them into review sites, portals, forums, marketplaces and other types.
• Choose actions. The type of the most important domains determines what you can do there.

Being present on a domain already does not take it off the plan. In 38% of cases AI cited at least two different URLs from one domain. Every additional article or profile on a frequently cited domain can become a new source.

Negative information is concentrated on a few sites. For all eight brands studied, five domains produced between a third and more than half of negative mentions. For the fashion store, 38% of answers citing Opineo and 36% of those citing GoWork rated the brand negatively.

Employee reviews also affect a brand's image. In the apartment operator analysis, AI used them in answers to guests' questions. The recommendation was to involve the HR team in work on the brand's image in AI.

User-generated content (UGC), such as reviews and forum posts, also shapes a brand's image. Avoid artificial mentions. Google warns that inauthentic mentions are not as helpful as they may seem. Publishing content on other people's domains purely for rankings risks being treated as parasite SEO.

How do you check whether ChatGPT and Gemini recommend your business?

A single question asked in a chat does not show whether AI recommends a business. Answers change even for the same question, so you need a larger sample.

The researchers asked the same question about a fashionable suit twice, a minute apart, in incognito mode. Three brands appeared in both answers, while four or five others changed.

Cited sources vary too. Across five answers to the same question, 42% of domains appeared only once. Platforms differ in this respect:

Reliable measurement requires many questions of four kinds, asked repeatedly (the methods are described in the section on researching brand presence). This was their share in the study:

Questions without a brand name matter most. They show whether AI names the business on its own when a customer is looking for a solution. It is worth asking the same questions about competitors to have a point of reference.

Platforms rate brands differently. In the study ChatGPT was usually the strictest, and Gemini and AI Overview the most favourable. The result can also change without the business doing anything. After the ChatGPT model change in summer 2026, one store's rating shifted by more than a point. Its rating in Gemini stayed almost unchanged over the same period.

How do you measure brand visibility in AI?

Brand visibility in AI is measured with several metrics at once: presence, rating, sources and traffic. None of them gives the full picture on its own.

A rating has to be read against competitors. In one analysis the craft manufacturer rated +2.83 was third of four brands. A small store rated +0.86 was second, because its competitors scored worse.

It is also worth taking into account how often AI talks about a given area. In the large store analysis, price was one of the three most important purchase decision areas. Its rating was 2.5 points below the competitors' average, so it became a priority.

Traffic from AI can be partly measured in Google Analytics 4. Missing events, such as a form submission, can be added through Google Tag Manager. Whether the measurement itself works properly is what the free GA4 audit checks. In the example fashion store, sessions from AI platforms over 12 months were split like this:

ChatGPT produced 98% of revenue from AI, even though it rated the brand most strictly. All measurable traffic from AI amounted to 1.7% of organic traffic. Traffic from Google AI Mode and AI Overview cannot be isolated in GA4. How AI traffic turns into sales is shown by a CRO analysis.

Data on these Google features comes from the generative AI performance report in Google Search Console. Since February 2026 Bing Webmaster Tools has shown in its AI Performance report how often Copilot cites a site. Since November 2026 ChatGPT Business Hub has reported mentions and clicks from ChatGPT. For an online business, a mention is only a reference to its website, not the brand name alone.

SEO and GEO: how do you combine them into one strategy?

You combine SEO and GEO (generative engine optimization) by treating optimisation for AI as an extension of SEO. Google says that optimising for generative search is still SEO. What GEO is and what it changes for brands is covered in our article GEO: will your brand disappear from results.

The difference lies in the scope of work. SEO focuses mainly on your own site and links. GEO also covers every place AI draws its knowledge of the brand from.

Actions are worth matching to the brand's position in each purchase decision area:

• AI does not know the brand: increasing reach in the sources AI cites in the industry.
• Weak rating: first a diagnosis of whether it is a genuine problem or outdated or incorrect information.
• Strong point: defending the advantage and communicating what sets you apart.
• No differentiator: leaving less important areas aside and building an advantage in the key ones.

A genuine problem, such as slow delivery, requires improving the process or communicating other strengths more strongly. Incorrect information requires contacting the publisher and publishing current data.

A combined strategy can look like this:

• Audit your own site. Check indexing, access for AI crawlers and answers to customers' questions.
• Measure visibility in AI. Ask many questions on several platforms, including questions without a brand name.
• Map the sources. Identify the few dozen domains AI cites in your industry.
• Digital PR. Work on the most cited domains first.
• Measure regularly. Repeat the study, because results also change after AI model updates. Convert the effects into revenue and ROI.

You do not have to be everywhere. It is enough to know your few dozen most important domains and start with them. We handle site audits and optimisation for AI search as part of our SEO services.

What methods are used to research a brand's presence in ChatGPT and Google AI Overview?

A brand's presence in ChatGPT and Google AI Overview is researched by systematically asking questions and analysing the answers at scale. Data from search engine tools and site analytics complements it.

A question panel should include questions about the brand, comparative questions and questions without a brand name. In the study cited, each question was asked five times on six platforms. The queries were made in the chat window rather than through the API, so that the answers matched what customers see.

Measurement has to be repeated every few weeks or months, because results change after model updates. A fixed set of questions lets you compare successive measurements and assess the effects of your work.

What sources is this article based on?

Data on sources, ratings and brand presence in AI answers comes from a 2026 Polish report on brand image in AI. The report is based on 15 analyses, 73,000 AI answers and 954,000 source citations from six platforms.

Google's recommendations are set out in the guide “Optimizing your website for generative AI features on Google Search”, updated on 10 July 2026. The rules for OpenAI's crawlers are given on the page “Overview of OpenAI Crawlers”. The process of adding a business to ChatGPT is described by Vahan Petrosyan in “How to Submit Your Website to ChatGPT Business Hub”, published in Search Engine Journal on 12 November 2026.