Who Tells Your Product’s Story Outside Your Store?

If a page outside your store describes your product, how would you check its accuracy? A review site, a comparison article, a forum thread. You didn't write it, you can't edit it, and it may be the page an assistant links to when it mentions you. So today's question is this: what do other sources say about your product, and what can you honestly do when they get it wrong?

This is an educational series from Verity Score for Shopify merchants. Last time, we audited five pre-purchase answers and gave each one a source owner. All of those sources were inside the business. Today, we step outside, to the descriptions you don't control.

We're still following the fictional luggage team. The review site, the article, the editor and every description in this story are invented for the example. No real publisher, product or conversation is involved.

You may remember that in episode two, an answer about commuting bags mentioned the team's convertible backpack, and linked to an external review site rather than to the store. The team put that on a list to check later. Now they check it.

The page is a comparison article: eight backpacks for commuters, with a table at the top and a short paragraph about each bag. The team's backpack is in the table. Being included in someone else's comparison feels like good news. But it's worth slowing down, because a mention isn't an endorsement. The article doesn't say how the bags were chosen, whether they were tested, or whether anyone was paid. The team simply doesn't know what this inclusion means.

Then they read what the article says. In the table, the material for their backpack is listed as nylon. In the paragraph below, the same bag is described as waxed canvas. Two different materials, in the same article, for the same product. Neither matches the supplier's sheet, which describes a coated polyester.

How did that happen? The team can't know for sure. But when they reread their own product page, they notice something uncomfortable. Their page only says "coated fabric." It never names the base material. Their product truth card from episode three has the supplier's description, but the page hasn't been updated yet. So when the article's author went looking for the material, there was nothing clear on the official page to find. Part of the confusion outside may simply mirror vagueness inside.

They search a little further. A forum thread repeats "waxed canvas," apparently copied from the article. And a reseller listing describes the straps as leather, which they aren't. One unclear fact has started to travel in several directions.

Let's start with why these pages matter at all. OpenAI's help article on shopping in ChatGPT, which we reviewed on October twenty-second, twenty twenty-six, says that product selection considers the question, the context, and metadata from merchants and from third parties. The same help article also mentions other third-party content. This supports checking external descriptions, without establishing that this particular article was used in an answer. That doesn't tell us how much weight any page carries, or whether this specific article was used for any specific answer. But it does tell us that what others say about your product isn't irrelevant.

The same help article says that reviews and ratings shown in results are not verified by OpenAI. So a star rating next to your product, or next to a competitor's, hasn't been checked by the service showing it. Keep that in mind before you celebrate or panic.

For Google's AI features in Search, Google's documentation says that the fundamentals of search still apply, and that you don't need a special file or special markup for those features. It also says that a page needs to be indexed and eligible to be shown with a snippet to appear as a supporting link, and that meeting those requirements doesn't guarantee it. So there's no trick for getting your own page cited instead of the comparison article. There's only the ordinary work of having a page that's accessible, clear and accurate.

Now, what do you control? Let's be precise, because different relationships need different evidence.

You control your own content. Your product page, your policies, and any guide you publish on your own site. That content should be clearly yours. It's not independent coverage, and it shouldn't pretend to be.

You don't control independent coverage. A review site, a journalist, a forum, a reseller. You can ask for a correction when something is factually wrong, and you can provide evidence. You cannot directly edit independent coverage. Ask for a documented factual correction; distinguish that request from any commercial relationship or placement. Paying for placement without disclosure, writing reviews under other names, or asking friends to post praise is not a shortcut. That puts your reputation at risk. It adds exactly the kind of unverified material that we've been trying to remove.

So the team separates its work into two parts.

The first part is at home. They update the product page to name the material as the supplier describes it, attributed to the supplier, and they correct nothing they can't confirm. They also decide to publish a short buying guide on their own site: how to choose a commuter backpack. Not a list of reasons to buy theirs, but criteria a buyer can actually check. Measure the interior of the laptop sleeve, and compare it with your own device. Check how the bag carries, on the back or by hand. Look for a stated material and care method, and notice when a page doesn't give one. Check where the brand ships and how returns work from your country. Every criterion is something a buyer can verify, and the guide is clearly labeled as written by the brand.

The second part is outside. And here, the team is careful not to turn a correction into a campaign.

The team doesn't email every site that mentions them. They pick one page, the comparison article, because it's the one that appeared in an observed answer, and because it contains a factual error they can document. That's your action this week: prepare one evidence-based correction request for one external page. If you find no documented factual error, record the pages checked and the date; that is a valid result. If no relevant external page is found, record that too. There is no need to manufacture a correction request.

Here's how the team builds it. First, they capture the page as it stands: the URL, the date, and an uncropped screenshot of both the table and the paragraph. Then they write down the exact statements that are wrong, word for word. Then they gather their evidence: the supplier's material description, the updated product page, and a photo of the label inside the bag. They check that their own page now agrees with that evidence, before they ask anyone else to change anything.

Then they write a short, polite message to the article's editor, through the contact route the site provides. It says who they are. It quotes the two conflicting descriptions. It gives the correct material, with links to the evidence. It doesn't ask for a better ranking in the comparison, a different opinion of the bag, a link, or a review. It asks only that the factual description of the material be corrected. And it says, honestly, what they don't know, such as whether the bag is waterproof, so the editor doesn't replace one error with another.

Before sending, someone else on the team reads it. The message goes out only once that person and the founder agree. Then the team records the date it was sent and sets a date to check the page again. If nothing changes, the team records that outcome and decides whether a proportionate follow-up is warranted. It does not post under a false identity.

They leave the forum and the reseller for later. Each would need its own evidence and its own decision. Copying the same message to everyone would turn a correction into outreach. And the reseller is a different kind of relationship: if the team supplies that reseller, the fix may belong in the product information the team sends them, which is a conversation between partners, not a correction request. The full template is in the show notes.

So: a mention isn't an endorsement. Your own content is yours, and should say so. Independent coverage is different from paid promotion. Payment cannot turn an unverified claim into evidence. When it's wrong, fix your own source first, then ask for one correction with evidence. Next time, we'll look at agentic commerce: what an assistant might actually be able to do with your products, and why you should check your eligibility, market by market, before you build anything.

Sources and review dates are in the show notes. This episode is read by a synthetic voice.