In this piece
ChatGPT Search Console: what the reported Business Hub means for your business
A customer can shortlist your company in ChatGPT without visiting your website. Website analytics miss that part of the decision. The reported Business Hub could make some of that exposure measurable.
ChatGPT Search Console is an informal name for business visibility tools reportedly being developed inside OpenAI’s Business Hub. Public availability remains unverified. Prepare a measurement baseline now, then assess what the eventual dashboard counts.
Research status: 6 November 2026. We reviewed the original observation, screenshot reporting and current OpenAI and Google documentation. We have not tested Business Hub. The plan and numerical examples below are proposals, not customer results.
Is ChatGPT Search Console available yet?
A general launch is not established by the evidence reviewed on 6 November 2026. Juozas Kaziukėnas’s firsthand post describes daily impressions and website clicks, displayed in a chart, inside a developing Business Hub.
Check which product you are looking at. A Google Search Console plugin analyzes Google’s data. ChatGPT Business workspace administration concerns your team’s account. Neither establishes access to measurements of how ChatGPT presents your company to customers.
What would the Business Hub let a business do?
The reported banner combines business visibility, maintained business information and a business agent. Barry Schwartz’s screenshot report reproduces these propositions. The banner leaves data coverage and control permissions unspecified.
| Proposition | Practical question to resolve |
|---|---|
| Track appearances in ChatGPT conversations | Which answer types and counting rules does the number cover? |
| Keep products and business information current | Which source owns each field, and how quickly do corrections appear? |
| Let a business agent answer using supplied context | Where is that agent shown, how is it identified, and what may it do? |
The shared visual is a banner; the daily chart is described in the observer’s text. A reconstructed dashboard should not be presented as a released interface.
Search Engine Land’s follow-up describes verified-business access, with Shopify connection as an example, but says the feature is not live yet. It also relays the absence of prompts, user data, competitor benchmarks and rankings. Both publications rely on the same observer.
First-party totals could show whether exposure changed. Our assessment is that explaining the cause still requires context: audience growth or a different mix of questions could move the chart without any content changes.
Who can claim a business, and what should you prepare?
Verified business owners are the group described in the report; a public eligibility list is still unverified. Prepare according to your business model while waiting for access requirements. Search eligibility, product-data integration and ownership verification are separate tasks. Completing one does not establish access to the others.
For service businesses and B2B companies
Start with a consistent company name, service descriptions, delivery markets and evidence on your public website. OpenAI’s crawler documentation distinguishes OAI-SearchBot for search from GPTBot for model training. Allowing a search crawler is an access decision, not a measurement of mentions or a guarantee of recommendation.
Our guide to agent-readable websites and crawler controls covers the implementation work. Measure answer quality separately from crawler access.
For ecommerce merchants
Check the integration you already have before building another one. OpenAI’s product-feed onboarding guide currently limits direct onboarding to approved partners. Its shopping documentation separately says Shopify Catalog already supplies merchant product data. Neither fact establishes Business Hub analytics access.
Assign an owner for catalog accuracy. Check products with different variants, currencies and delivery destinations against what customers can see. Document disagreements before changing feeds. Avoid a second integration that leaves two systems claiming to own the same price.
What should a ChatGPT visibility report measure?
A useful visibility report separates exposure, answer quality and business outcomes. This is our proposed reporting dictionary, not OpenAI’s metric specification. Put the population, period and source next to each number. Platform totals alone cannot establish qualified demand.
| Signal | Question it can answer | What it cannot establish alone |
|---|---|---|
| Platform-reported impression | How much exposure did the platform count within its stated scope? | Unique people, positive sentiment or sales intent without further definitions |
| Mention in a sampled answer | Did this answer name the correct business? | The total number of real customer conversations mentioning it |
| Linked citation in a sampled answer | Did the answer link to a relevant page? | A customer click, endorsement or completed visit |
| Platform-reported website click | Did the platform count an outbound interaction? | A one-to-one match with website sessions |
| Attributed website session | Did your analytics record a visit from an identifiable source? | Exposure without a click or every later return visit |
| Qualified lead or accepted order | Was a qualifying outcome recorded under your attribution rules? | That one content change caused the outcome |
Review the wording around a mention. A recommendation and a warning can both contain your company name. For services, check whether the suggested offer, location and customer fit are correct. For merchants, inspect the product and purchase conditions.
Keep bot requests in a diagnostic report. A crawler fetching a page proves a request occurred, without establishing human exposure, citations or sales.
How does this compare with Google Search Console and GA4?
The reported Business Hub concerns exposure inside ChatGPT. Google Search Console measures Google’s search features, and GA4 measures visits to your website. Google now documents a generative-AI performance report for impressions in its own search features. It does not measure ChatGPT exposure.
GA4 also documents an AI Assistant default channel, including traffic from ChatGPT. Visits from Google’s AI Overviews and AI Mode fall under Organic Search. These are website acquisition categories; they do not reveal the conversations that produced a visit.
To inspect recorded visits, open GA4's Traffic acquisition report and examine Session source / medium. Filter the ChatGPT sources present in your property and compare outcomes. Keep paid campaigns identifiable.
Inspect the links and source values your setup receives; do not assume every ChatGPT link carries a particular UTM parameter. To investigate discrepancies, check redirects, consent state, session scope, reporting dates and campaign tags.
How do you prepare a baseline before access?
Prepare a repeatable panel of buyer questions alongside your acquisition and sales reports. Use it to diagnose selected answers about your business. Label it as a controlled sample; your chosen prompts do not represent the distribution of customer demand.
OpenAI’s current search guide explains that search can run automatically and use relevant context. Record the product, search state, language and account context so a colleague can repeat the sample.
- Choose one outcome. Define a qualified enquiry with sales, or an accepted order for a shop. Keep that commercial outcome distinct from page views.
- Build questions from real buying decisions. Group discovery questions separately from branded accuracy checks and comparisons. A proposed starter panel of 30 unbranded questions, repeated 3 times each, gives 90 observations. This operating example makes no claim of statistical sufficiency.
- Record the conditions. Save the date, product or model label shown, search behavior, language, stated market, account context and exact prompt. Use fresh conversations for a consistent baseline. Keep personalized checks in a separate series.
- Review the answer. Count each answer once for brand presence, record relevant citations separately, and flag incorrect facts. If 18 of the 90 answers name the correct business, report a 20% mention rate in this selected sample. Do not call it market share.
- Keep acquisition and sales alongside the sample. Record attributable sessions and qualified outcomes in their own columns. Ask buyers how they found you where appropriate, but retain self-reported discovery as a separate evidence type.
- Log changes and repeat consistently. Log content edits, feed corrections, campaigns and measurement changes. Compare equivalent periods and prompt sets. A before-and-after difference alone cannot establish causation.
For an Austrian company selling internationally, run German discovery questions and English export-market questions as separate cohorts. A rise in the English panel should not be reported as stronger visibility among Austrian buyers. Use separate cohorts wherever markets differ.
If you automate collection, use a documented, permitted interface and label its output accurately. Our data-pipeline acceptance checklist helps define source ownership, freshness and reconciliation before a dashboard becomes a recurring management report.
Worked example: more appearances, unchanged click ratio
The same click ratio can accompany more exposure and more leads. This hypothetical example assumes compatible impression and click definitions across two equal periods. The numbers illustrate the calculation; they are not observed Business Hub data or a forecast.
| Measure | Period A | Period B |
|---|---|---|
| Comparable platform impressions | 10,000 | 12,000 |
| Comparable platform clicks | 200 | 240 |
| Clicks divided by impressions | 2% | 2% |
| Website sessions attributed to ChatGPT | 150 | 180 |
| Qualified leads attributed under the same rule | 8 | 12 |
The arithmetic is 200 / 10,000 = 2% and 240 / 12,000 = 2%. Clicks increased by (240 - 200) / 200 = 20%. The platform ratio stayed constant. Qualified leads increased from 8 to 12, but these small counts deserve investigation before any budget decision.
Check that impressions and clicks cover the same surfaces and period before calling their ratio a click-through rate. Dividing 240 platform clicks by 90 sampled answers mixes populations. Reconcile website sessions with platform clicks rather than forcing equality.
To choose an action, inspect which offers, landing pages and customer types produced the attributed leads. An unchanged click ratio alone does not justify rewriting every page.
What must the dashboard explain before you rely on it?
Before making the dashboard a management KPI, verify its counting rules and coverage in the documentation available at launch. Our proposed acceptance checklist asks about capabilities whose final scope remains unknown.
- Counting unit: does an impression mean an answer, a visible business card, a link or another event? How are repeated appearances counted?
- Coverage: which chat, search, shopping or agent surfaces are included? Are paid and unpaid interactions distinguishable?
- Identity: how are brand aliases, domains, subsidiaries, locations and retailers matched? How can a mistaken match be corrected?
- Time: which timezone, reporting delay and historical window apply? Are preliminary values revised or missing values suppressed?
- Access: which businesses and markets qualify? What ownership evidence and agency roles are supported?
- Portability: is an export or API documented, and can your team preserve definitions when the product changes?
Record an unanswered question as unknown. Zero means an event was counted as zero; unavailable means you cannot observe it. Replacing missing data with zero can turn an access limitation into a false story about lost demand.
What would we do this week?
We would establish a baseline now. Assign owners to measurement definitions and public business information, then set a review date. If your account already shows access, document its scope before adding the numbers to client reports.
Prioritize accurate offers and qualified enquiries. Merchants also need catalog ownership and product-data checks. If a business agent becomes part of the customer path, test whether it can complete the intended interaction; our guide to whether an AI agent can use your product covers that different acceptance problem.
For help turning this into a repeatable process, Wavect’s AI visibility service connects content priorities with technical verification and outcome measurement. Bring your current acquisition report, target markets and one offer to a conversation about measuring ChatGPT visibility. The first decision is what evidence would change your next content or distribution investment.