MayaBook a demo
Back to Blog
AI VisibilityEnterpriseNovember 19, 2026

AI Visibility for Multi-Brand Companies: How to Track a Brand Portfolio

How a company with several brands should track AI visibility: one project per brand, sister brands on each other's competitor list, clear rules for owned domains and a two-layer portfolio report.

Prefer Maya AI in Google

Highlight our stories in Search, AI Mode & AI Overviews.

A company that owns several brands should track AI visibility one brand at a time and then read the brands side by side. Each brand gets its own project, competitors and questions, and sister brands that sell into the same category go on each other's competitor list. Maya supports one project per brand, with unlimited projects on the Enterprise plan. The setup works in any global market, but a branded house that uses one name everywhere can start with a single project.

Below you will find three ways to set up tracking for a portfolio. The guide then shows how to see where your own brands compete, how to keep a parent brand and its sub-brands apart, and what to do when the wrong brand wins an answer. If you manage client brands as an agency, the guide to AI visibility tools for agencies is the closer fit. If you are choosing a vendor, start with the enterprise AI visibility platform buying guide.

One answer can name several of your brands

When a buyer asks an AI assistant what to buy in a category, the answer is usually a short list. For a single-brand company the question is simple: is the brand on the list or not. For a group that owns several brands, the same answer can be good news and bad news at once.

Three portfolio problems show up in AI answers that a single-brand dashboard does not separate. The first is choice: which of your brands the assistant recommends for a given need, and whether that is the brand you intended for that need. The second is competition inside the group: two of your brands appearing in the same answer, or one taking the place of another. The third is identity: the assistant attributes a sub-brand's product to the parent, mixes up two of your brands, or describes a brand you sold as still part of the group.

None of these shows up in a single total for the group. A group-level mention rate can rise while the premium brand loses its place to the value brand on the questions that matter most for margin. The setup decides which of these problems you can see at all.

Set up a multi-brand AI visibility platform one brand at a time

The basic unit in Maya is a project, and each project holds one brand. The General tab of project settings stores the brand name, an alternative name, description, country and website. According to the help center, the brand name and its variants are what mention detection looks for. A portfolio starts with a decision about how brands map to projects.

The table compares three setups built from those settings. None of them is a separate portfolio product; each is a way of using project settings that Maya documents.

SetupWhat it showsWhat it hidesWhat it needs
One project for the parent brand, sub-brands added as alternative namesOne mention rate for the whole groupWhich brand was mentioned, and any answer where one of your brands displaces another1 project (available on Starter)
One project per brandEach brand's own mention rate, share of voice and citations against its own competitorsAnswers where a sister brand is recommended instead, unless the sister brand is on the competitor listAs many projects as brands: 2 on Premium, unlimited on Enterprise
One project per brand, with sister brands on each other's competitor listWhich of your own brands an answer recommends on the same question, next to outside competitorsA sister brand's gain lowers this brand's share of voice even when the group gainsAs many projects as brands: 2 on Premium, unlimited on Enterprise

Project limits come from the plans and quotas page. The first setup suits a group whose products all carry one name. The second suits a house of brands that rarely meet in the same answer. The third is the only one of the three in which competition between your own brands becomes visible.

Two limits shape the first setup. Alternative names are meant for spellings of the same brand, such as abbreviations, old names and concatenations, and the help center caps them at 5 per project. Maya's team edits them through support, because a wrong tag can inflate or deflate a mention rate overnight, according to the project settings article. A sub-brand name entered as a tag is no longer a spelling of the parent; it merges two brands into one number.

The tracked-question budget is a portfolio decision too. The plans and quotas page states that the tracked prompt quota is shared across all projects in an account. The group has to decide how many questions each brand gets. How to choose the questions themselves is covered in the guide to building a prompt set.

Decide which domains count as which brand

Many groups run a corporate site, separate brand sites and sometimes shared shops. Maya's project settings let you add up to 5 additional owned domains per project, according to the help center. Citations of any listed domain count as that brand's citations. Without the list, the brand's own pages appear as "Other" in Citation Share.

Write down one rule before anyone fills in the field. A sister brand's domain does not belong in another brand's project, or its citations are credited to the wrong brand. The corporate site needs a deliberate home: list it in the project of the brand it mostly describes, or in none, and note the choice next to the report. Owned citations are worth separating carefully, since Maya's research on ChatGPT found that brand-owned sites lost part of their citation share between September and mid-November 2026 (Maya research).

Give marketing teams unlimited seats on the platform, but limit who joins each brand

Seats are unlimited on every Maya plan, according to the plans and quotas page, so a group does not have to ration access across brand teams. Access is the harder question. Project settings include Domain Access, an optional auto-join through which anyone signing up with your company email domain joins the project automatically.

In a group where every brand team uses the same corporate email domain, auto-join lets people from one brand into another brand's project. The help center advises turning the option off if you share a domain with teams who should not see the data. Decide per project, and invite brand teams by email where the data should stay inside one team. Project deletion is restricted to the project owner, so name an owner for each brand project as well.

See where your own brands compete with each other

Competition between your own brands is not always a loss. If an answer recommends your premium brand for one need and your value brand for another, the portfolio is doing its job. The problem is an answer that recommends the brand you did not intend for that need, or two of your brands where one clear recommendation would serve the buyer better.

To tell the two apart, give every category question one owner: the brand the group wants an AI assistant to recommend for that need. Questions that no brand owns are worth listing too, since they show where the portfolio has a gap rather than an overlap. The owner map is a short list kept by group marketing, not a setting in any tool.

Then track each owned question in the owner's project, with the sister brands on that project's competitor list. Maya's Overview dashboard ranks you and your listed competitors by mention rate for the current filter, and the competitor list also defines the Share of Voice calculation. With sister brands on the list, the owner's project shows on the same questions whether the answer named the owner, a sister brand or an outside rival. Competitive Benchmarking lets you open the answers where presence differs and read them.

Read two numbers separately. The owner brand's share of voice tells you whether the right brand wins the question. A portfolio share, the sum of your brands' mentions on the same answers, tells you whether the group as a whole is present. A rising portfolio share with a falling owner share is the signature of internal displacement.

Keep the parent brand and its sub-brands apart

AI assistants build their picture of a company from many pages, and a group structure gives them several ways to get it wrong. Four patterns are common. A sub-brand's product is described as a parent-brand product. The parent's positioning or price is applied to a sub-brand. A brand that was sold is still listed as part of the group. An unrelated company with a similar name is merged into one of your brands.

Find these in the answers to brand questions, such as "who makes" or "is this brand part of" questions, tracked in the project of the brand being described. The finding then has a clear owner. The guide on what to do when AI misrepresents your brand covers how to classify a wrong statement and correct its source, so this guide does not repeat it.

Measurement can create the same confusion. If the parent's name is part of a sub-brand's name, decide before tags are set which project should count which name, and ask Maya's team to set the tags to match. Otherwise the parent's mention rate may include answers that were really about the sub-brand.

Track AI visibility at both brand and product level

Within a brand, Maya's Prompt Lab lets you label prompts, for example by product line or campaign. You can then filter the prompt list by label, according to the managing prompts article. The Overview dashboard can also be filtered by label. A brand project with product-line labels gives you a brand-level reading and a product-line reading from the same answers.

For categories where ChatGPT answers with product cards, Maya's Product Radar collects the products that surfaced in answers to your tracked prompts. Each entry shows the brand, the price as surfaced and the prompt that triggered it. Product Radar reads ChatGPT product cards, and not every category triggers them. Matching products and variants in AI answers is covered in the guide to AI visibility in e-commerce.

Track each brand across several countries and languages

Every tracked prompt in Maya has its own country and language, set when the prompt is added or imported, and the Overview dashboard filters by country. Assistants give different answers per market, so a brand sold in several countries needs prompts written for each market, not one translated list. The article on testing AI visibility by language and country explains why language and market are recorded as two separate fields. The Enterprise plan lists unlimited countries on the pricing page.

Situations a portfolio team runs into

Most portfolio questions about AI visibility arrive in a few recurring forms. The list pairs each with what to check first and what to change.

SituationWhat to check firstWhat to change
Group marketing wants one AI visibility dashboard for every brand in the holdingWhether each brand has its own project and the same reporting periodKeep brand projects separate and combine exports into one portfolio sheet
A buyer asks which multi-brand company makes a product and the answer names the parent instead of the sub-brandThe brand questions in the sub-brand's project and the pages they citeState the relationship on the corporate site and the sub-brand site
An answer recommends your value brand on a question the premium brand should ownThe owner map, and the premium brand's project with the value brand on its competitor listMake the owner brand's page answer that need directly
A country team wants to track its brand across several countries and languagesWhether prompts carry their own country and language, not one translated listWrite market-specific prompts for each country
Product managers want visibility tracked at product level as well as at brand levelWhether prompts carry product-line labelsAdd labels and read the brand project by label
Brand managers across marketing teams all need seats on the visibility platformWhether Domain Access lets other brand teams join automaticallyInvite by email and turn auto-join off where data should stay inside one team

The rows that need a decision from group marketing are the ones where a fix touches more than one brand. Put those rows first in the portfolio report.

What to do when the wrong brand wins the answer

The changes below start from a hypothesis, not a proven cause. Each one is made, dated and read again on the same questions in the next period. If the answer does not move, the hypothesis was wrong or incomplete, and the next test starts from the cited sources.

Make the owner page the clearest answer for its need

One possible explanation for a sister brand winning a question is that its page states the need more plainly than the owner's page does. Test it by making the owner brand's page say, in a quotable sentence, who the product is for and what sets it apart from the group's other brands. A page on the group site that explains which brand is for whom gives assistants one source that separates the brands. Check the next period whether the owner brand is named on that question and whether the new pages appear among the cited sources.

State the relationship between the parent and its sub-brands

Where assistants mix up the parent and a sub-brand, write the relationship on the pages they are likely to read. The likely pages are the corporate "our brands" page and the About page of each brand site. Schema.org defines parentOrganization as the larger organization that an organization is a subOrganization of, and subOrganization as the inverse relationship (schema.org). Google's documentation on organization structured data, which does not list these two properties, says such markup can help Google disambiguate your organization from other organizations in search results (Google Search Central). Whether the same markup changes AI answers has not been shown, so treat it as a test: add it, date it and read the brand questions again.

Correct third-party pages that mix your brands up

When the cited source of a wrong attribution is a directory, a retailer listing or an old news article, the fix sits with that publisher. Ask for the correction with the evidence from your own pages, and track the brand question until the cited source changes. The misrepresentation guide describes the correction process step by step.

Illustrative example: two sister brands on one question

The following example is illustrative and does not describe a real customer. A household-goods group owns a premium detergent brand and a value detergent brand. The owner map assigns "detergent for sensitive skin" to the premium brand.

Observation. In the premium brand's project, with the value brand on its competitor list, the competitor comparison shows the value brand named more often than the premium brand on the sensitive-skin questions. The cited sources are a retailer category page and a comparison blog; the premium brand's product page is not among them.

Possible cause. One explanation is that the premium brand's page does not mention sensitive skin in a sentence an assistant can quote. Another is that the comparison blog reviewed only the value brand.

What Maya shows. The premium brand's Overview, filtered by the sensitive-skin label, ranks both brands on the same answers, and Source Analysis shows which pages those answers cite.

Action and check. The premium brand adds a clear statement of who the product is for to its product page. The group site adds a page explaining which detergent is for which need. The change dates go into the report; the next period reads the same labelled questions and states whether the premium brand now leads on them.

Report the portfolio without hiding a weak brand

A portfolio report has two layers. The first is one page per brand, with the same lines in the same order, so that brands are read on equal terms. The second is a portfolio page that puts the brands side by side on the questions they share. The portfolio page also shows the owner map: which brand owns which need, and which owned questions a sister brand is winning.

Compare brands only where they answer the same questions, and with the Overview's Non-branded filter on. The help center calls that filter the honest view of discovery visibility, since branded prompts almost always mention the brand and inflate the averages. The guide to the seven AI visibility metrics explains the definitions and why a single run is a sample. The guide to AI visibility reporting for CMOs covers which lines leadership needs.

To build the portfolio page, take the data out of each brand project in the same format. Prompt Lab exports the current view as a spreadsheet. According to the pricing page, the Premium plan adds CSV and webhook export, and the Enterprise plan adds API access and a Looker Studio integration. On Enterprise, Maya Report also produces a narrated report per brand, with two reports per brand per month according to the Maya Report article.

What this guide does not cover

The guide does not compare AI visibility vendors or explain how to price a multi-brand contract. The enterprise buying guide covers evaluation and cost. Client portfolios managed by agencies are covered in the agency tools guide. The guide does not estimate revenue from AI mentions, and it does not claim that structured data alone changes how assistants describe a group.

Frequently asked questions

What is an AI visibility platform for multi-brand companies?

An AI visibility platform for multi-brand companies records how AI assistants mention, recommend and cite each brand in a group on a fixed set of buyer questions. The point is to read the brands separately and also against each other.

Can Maya AI support multiple brands or regions?

Yes. Each brand is a separate project, and the number of projects depends on the plan: one on Starter, two on Premium and unlimited on Enterprise. Each tracked prompt carries its own country and language, and the Enterprise plan lists unlimited countries.

Which AI visibility tool can track my brand across several countries and languages?

Look for a tool that stores the country and language on each tracked question and lets you filter results by market. Maya works this way, and its Enterprise plan lists unlimited countries.

Conclusion: Track each brand alone, then read them together

A multi-brand company needs two readings from the same answers. The first is how each brand does against its own competitors. The second is which of the group's brands an assistant picks when they meet. Separate projects, an owner for every category question, sister brands on each other's competitor list and a clear rule for owned domains make both readings possible.

Which of your brands should win your most valuable category question today, and does the answer agree? Maya tracks each brand as its own project on a fixed question set every day, and on the Enterprise plan the number of brand projects is unlimited. Starter, Premium and Enterprise include a 3-day free trial, and the plans are compared on the pricing page.

Sources

About the author

Efsa Ediz

GEO researcher at Maya. Studies how large language models retrieve, rank, and cite sources — and what brands can do to show up in AI answers.

Ready to improve your AI visibility?

See how your brand appears across ChatGPT, Claude, Gemini, and other AI assistants.