OpenAI Just Gave Brands a Free AI Visibility Report. Hordus Turns It Into an Operating System.

OpenAI's new Business Hub shows brands how often ChatGPT mentions them, how many clicks follow, and which topics they appear in. Here is what the data covers, where it stops, and how Hordus turns it into a working GEO operation.

Written by Nimrod Fisher, Hordus AIPublished:
OpenAI Just Gave Brands a Free AI Visibility Report. Hordus Turns It Into an Operating System.

OpenAI launched Businesses in ChatGPT in beta on November 10, 2026, giving verified companies a Business Hub with native ChatGPT visibility data.

The Business Hub Insights dashboard reports mentions, website clicks, and discovery topics as aggregated activity over 7-day and 28-day windows.

OpenAI states that claiming a business profile does not guarantee more visibility or give a company control over ChatGPT's independent recommendations.

The dashboard covers ChatGPT only and does not expose individual prompts, user data, competitor rankings, or conversion outcomes.

Hordus AI runs a managed GEO and AEO operation that treats this first-party data as one diagnostic input inside a detect, assess, engineer, and validate cycle.

OpenAI's Business Hub is the first time a major AI answer engine has told brands, in its own numbers, how often it surfaces them. That makes it a real milestone and a useful baseline for any marketing team. It is also a measurement layer for a single engine, reported in aggregate, with no explanation of why ChatGPT answers the way it does. B2B teams get the most from it when the dashboard feeds a process that diagnoses answer gaps, builds content to close them, and checks whether the next round of AI answers changed. That process is what Hordus operates.

Mentions, clicks, and discovery topics: the three numbers OpenAI now reports

Businesses in ChatGPT lets an eligible company claim a verified profile through the Business Hub, then update its business and product information. Verification requires proof that the claimant owns or is authorized to represent the business. Once claimed, the profile unlocks an Insights view with three core metrics.

Mentions count how often ChatGPT names the business in its answers. Website clicks count recorded clicks on links to the company's site from inside ChatGPT. Discovery topics are the broad themes that recur across conversations where the business is mentioned or its website is cited. PPC Land's review of the live dashboards found a trend chart for mentions and a top-five list of discovery topics, with 7-day and 28-day views.

Eligible U.S. e-commerce and online-service businesses can also build a Business Agent, a company-controlled assistant that users open from the business profile. OpenAI labels Business Agent conversations and keeps them separate from ChatGPT's own answers. The agent represents the brand, while ChatGPT's recommendations remain independent.

Where the Business Hub data stops

OpenAI is direct about the boundary. Claiming a business does not guarantee more visibility and does not give a company control over what ChatGPT independently recommends, and ChatGPT may still draw on other sources when it answers. The profile is one input among many.

The data itself has hard edges. Every metric is aggregated, so a team cannot see the individual prompts that triggered a mention or the user who clicked. Competitor rankings are absent. Conversion outcomes are absent too, and OpenAI's own tooltip on the click metric says "Clicks don't confirm website visits or conversions." Insights also appear only after a business crosses a conversation threshold that OpenAI has not published, so smaller brands may open the dashboard and find dashes.

Then there is coverage. The Business Hub reports on ChatGPT and nothing else. A B2B buyer who shortlists vendors in Perplexity, checks a Google AI Overview, and asks Gemini or Claude for a comparison leaves no trace in it.

A free baseline still leaves the why unanswered

The free report is valuable for one reason above the rest: it is first-party. Until now, every estimate of ChatGPT visibility came from outside tools running sample prompts and inferring patterns. OpenAI's numbers come from real conversations, and that gives a marketing team a reference point it can defend in a budget meeting.

What the report cannot do is explain itself. A mentions count that drops 30 percent over 28 days tells a CMO that something changed. It does not say whether a competitor published a stronger comparison page, whether a review site started outranking the company's own documentation as a cited source, or whether ChatGPT began framing the category with terminology the brand never uses. A discovery topic tells a content team which theme the brand is attached to. It does not say which themes the brand is missing from entirely, and for most B2B companies the missing themes hold the bigger opportunity.

Diagnosis requires prompt-level answers, citation sources, and competitor context. Change requires content and distribution work. Neither lives inside a dashboard.

The Hordus cycle: detect, assess, engineer, validate

Hordus runs GEO and AEO as a managed operation built on four connected stages, and each stage hands a concrete output to the next.

Detect tracks how AI engines answer the prompts a company's buyers actually ask, across ChatGPT, Gemini, Claude, Perplexity, and Google AI Overviews. Hordus monitors brand presence in those answers, competitor framing, and the sources each engine cites.

Assess turns that tracking into an answer-gap backlog: the prompts where the brand is absent, misdescribed, or outranked, ranked by business value. The Hordus GEO Site Audit feeds this stage by crawling the company's site and scoring each page from 0 to 100 on AI accessibility, structured data, content completeness, and internal link authority.

Engineer produces the fix. That can mean rebuilding a product page so its claims are extractable, adding structured FAQ content and schema, or writing new pages that answer the questions buyers ask before they ever name a vendor. The same stage places that content where AI engines look, on the company's own site and across the third-party publications and platforms that engines cite for the category.

Validate re-runs the same prompts and checks whether the answers moved, then ties the change to the traffic and pipeline data the company already tracks.

From a dashboard number to a backlog item

OpenAI's data slots into the detect and validate stages as an extra layer of first-party evidence. The value shows up in how a single metric becomes an action.

Take a falling mentions trend. Hordus checks its own cross-engine tracking for the same period. If mentions dropped in ChatGPT but held in Perplexity and Gemini, the cause is likely specific to how ChatGPT retrieves sources for that category, and the team looks at which pages ChatGPT stopped citing. If mentions fell everywhere, a competitor or a new third-party source probably displaced the brand, and the backlog item becomes a content response to that source.

Clicks work differently. A brand with steady mentions and few clicks is being named without being cited as the place to learn more. That usually points to pages that ChatGPT finds hard to quote, and the site audit shows which ones. The engineering task is to make those pages the cleanest source for the claim.

Discovery topics are the most direct input for content planning. Suppose a cybersecurity vendor's top five topics are all about pricing and integrations, while its sales team closes deals on compliance automation. The dashboard has just shown that ChatGPT does not associate the brand with its strongest use case. Hordus maps the compliance prompts buyers ask, measures who currently owns those answers, and builds the pages and off-site coverage to put the brand into them. Four weeks later, the 28-day discovery topics view becomes one of the checks on whether that work landed.

What sits outside ChatGPT's view

The Business Hub and a managed GEO operation answer different questions. Here is how they compare.

  • Engine coverage: OpenAI Business Hub: ChatGPT only. Hordus: ChatGPT, Gemini, Claude, Perplexity, Google AI Overviews. Why it matters: B2B buyers research across several engines before they shortlist.
  • Prompt visibility: OpenAI Business Hub: Aggregated; no individual prompts. Hordus: Tracked prompt sets by topic, branded and unbranded. Why it matters: Answer gaps show up prompt by prompt.
  • Competitor context: OpenAI Business Hub: Not shown. Hordus: Competitor presence and framing per topic. Why it matters: Visibility only means something relative to who else gets named.
  • Diagnosis: OpenAI Business Hub: Discovery topics with no explanation of answer shape. Hordus: Citation source analysis plus a 0 to 100 GEO Score per page. Why it matters: A fix starts from a known cause.
  • Action: OpenAI Business Hub: Edit profile and product information. Hordus: Engineered pages, FAQs, schema, and off-site distribution. Why it matters: ChatGPT may still use sources beyond the claimed profile.
  • Outcome: OpenAI Business Hub: Recorded clicks; OpenAI notes clicks don't confirm visits or conversions. Hordus: Re-measured answers tied to traffic and pipeline data. Why it matters: Leadership funds programs that connect to revenue.
  • Cost: OpenAI Business Hub: Free with a verified claim. Hordus: Managed service. Why it matters: The free layer sets the baseline; the operation does the work.

Claim the profile, then build the operation

Every B2B company eligible for Businesses in ChatGPT should claim its profile now. The information is accurate only if the company keeps it accurate, and the Insights data costs nothing. Record the first 28-day baseline for mentions, clicks, and discovery topics before any other change goes out, so later movement has a reference point.

Then treat that baseline as the start of the work. The teams that gain the most from OpenAI's data will be the ones that pair it with cross-engine tracking, a ranked answer-gap backlog, and content built to be cited.

Already claimed your ChatGPT business profile? Talk to us about turning your first Business Hub report into an answer-gap backlog across every major AI engine.

Frequently Asked Questions

No. OpenAI states that claiming a business does not guarantee more visibility or give a company control over ChatGPT's independent recommendations, and ChatGPT may still use other sources. Hordus focuses on those other sources, including the company's own pages and the third-party content engines cite.
The dashboard shows aggregated mentions, website clicks, and discovery topics, with 7-day and 28-day views. It does not show individual prompts, user data, competitor rankings, or conversion outcomes. Data appears only after a business reaches a conversation threshold that OpenAI has not disclosed.
Hordus treats the Business Hub numbers as an extra first-party input in its detect and validate stages. A shift in mentions, clicks, or discovery topics is cross-checked against Hordus's own tracking across other engines, then turned into a specific item on the answer-gap backlog.
OpenAI's report measures one engine in aggregate. Hordus tracks prompt-level answers across ChatGPT, Gemini, Claude, Perplexity, and Google AI Overviews, adds competitor and citation context, engineers content on and off the company's site to close gaps, and validates the results against business outcomes.
No. A Business Agent is a labeled, company-controlled assistant that runs separately from ChatGPT's own answers, and it is currently limited to eligible U.S. e-commerce and online-service businesses. Buyers who never open the agent still see ChatGPT's independent answers and the answers from other engines. Hordus works on that layer.
Record the first 28-day baseline for mentions, clicks, and discovery topics, and compare the discovery topics with the use cases that actually win deals. Hordus can then map the prompts behind any missing topics and build the backlog to close them.

Methodology & Sourcing

Data Accuracy & AI Visibility Metrics:The statistics and AI visibility scores cited in this article are generated using Hordus AI's proprietary Answer Share of Voice (A-SOV) engine. Data is derived from consented, anonymized real user interactions across major LLM interfaces (ChatGPT, Claude, Gemini).

Editorial Integrity:All AI-assisted research undergoes mandatory human editorial review by our GEO strategy team prior to publication to ensure factual accuracy and alignment with Google's YMYL (Your Money or Your Life) search quality rater guidelines.