When ChatGPT recommends competitors but leaves out your brand, save the question and the full answer before changing the website. The useful starting point is the buying requirement behind the recommendation: does your product meet it, can a reader verify that on your site, and what sources support the competing options? These checks can identify a specific correction. They cannot establish the model’s internal reason for leaving a brand out.
Use a question your target customer would actually ask
A question that names your brand tests how it is described after someone has heard of it. An open category question tests whether it enters the shortlist at all. Compare the two while keeping the buying conditions fixed. Take those conditions from enquiries or sales conversations: order size, destination, required capability and deadline are useful when they change which suppliers qualify.
The examples below use a hypothetical packaging supplier serving small consumer brands. The customer needs 300 custom skincare cartons delivered to Germany. This is a diagnostic scenario, not a real customer enquiry or a claim about any supplier’s capabilities.
- Category question: “Which suppliers can make 300 custom skincare cartons and deliver to Germany? Compare minimum quantities, printing limits and how to confirm the delivery date.”
- Brand question: “Can [brand] make 300 custom skincare cartons and deliver to Germany? Where can I verify the minimum quantity, printing limits and delivery date?”
Run each question in a fresh conversation and record the date, language, region, product or model shown, and search setting. Repeat the matched pair before treating an omission as a pattern. If the named answer is accurate but the open answer omits you, investigate discovery and supporting information. If both describe the product incorrectly, correct the underlying facts first. Neither observation identifies a single cause by itself.
Make the buying conditions verifiable on the page
Open the page a buyer would use to check the recommendation. Look for three things together: who the offer serves, exactly what is available, and the conditions that limit it. A homepage claim such as “flexible customization and fast delivery” does not tell this buyer whether a 300-unit order qualifies. A specification, product page or ordering guide is often the better place to resolve it.
Before publishing new wording, ask the product or sales owner to confirm each condition. In the illustrative rewrite below, assume the supplier really accepts 300 units for one specified carton and printing process. The sentence must change if the actual minimum is higher or the customer needs a different finish.
- Before: “Flexible custom packaging with fast international delivery.”
- Illustrative rewrite: “This carton supports custom dimensions and single-colour printing from 300 units. Other finishes require a separate minimum-order check. We can quote delivery to Germany after confirming the artwork, quantity and destination postcode.”
Put the verified terms beside the relevant product, not only in a general FAQ or a sales reply. Link a broad category page to those details. Ask a colleague unfamiliar with the order to identify the passage supporting each requirement. If the answer still depends on an unwritten exception, publish that qualification where the buyer will need it.
Check the recommendation against its cited evidence
For each relevant competitor, record the recommendation, the exact cited URL and the passage that supports it. A source about standard cartons does not establish the minimum quantity for custom printing. A production estimate does not establish the arrival date in Germany. Mark these mismatches rather than copying the recommendation into your own marketing.
Then compare the supported requirement with your own offer. A capability you have but do not explain is an information gap. An old offer still being quoted is a freshness problem. A requirement you cannot meet is a product-fit issue. These lead to different actions. If the answer provides no verifiable source, keep the cause unresolved; the absence of a link is not proof that a particular page was ignored.
Investigate access separately if a relevant public page may be blocked. OpenAI identifies OAI-SearchBot as its search crawler and publishes access guidance. Give the web team the affected URL and ask it to verify the applicable rules and the returned page content. Successful access does not guarantee a recommendation.
Choose the correction that follows from the finding
Start with an inaccurate buying condition or a confirmed access failure on an important page. Next, clarify relevant facts the product can support. Use one task per page or dependency so an owner can finish and verify it. Keep a copy of the previous wording and the publication date.
| What you found | What to change | How to verify the work |
|---|---|---|
| The minimum order exists only in sales replies. | Add the confirmed quantity and process limits to the product page. | A reader can determine whether 300 units qualify. |
| An old announcement quotes an expired offer. | Add dated status beside the offer and link to current terms. | Historical and current terms are clearly distinguishable. |
| The public product page returns a challenge screen. | Have the web owner review verified crawler access. | The permitted request receives the intended product text. |
| The product genuinely requires at least 500 units. | Keep the limit accurate; test a suitable customer requirement. | No page promises an unsupported 300-unit order. |
Retest the original requirement and record the outcome
After the revised page is publicly available, rerun the original questions under the recorded conditions. Keep all answers. Track shortlist inclusion, factual accuracy, cited destination and identifiable enquiry activity separately. An accurate minimum order is an improvement in information quality even if the brand is still absent; a new mention with an incorrect delivery promise is not a satisfactory result.
For a small review, write the result plainly: “The brand appeared in two of six answers; one of those two misstated the printing limit.” These are illustrative counts, not a success target or an estimate of customer reach. Repeat observations over time before committing more work. Record referrals and customer-reported discovery separately from mentions, since the two cannot be equated.
When reviewing AskScope results, use the original answer, competitor mentions and source URLs to assemble this record. The completed review should identify one supported page correction, or explain why the buying requirement does not fit. That is enough to make the next piece of work specific.
