The Question Your Store Still Cannot Answer

A shopper is one click away from buying. Then they stop, and they ask a question. Can I clean this? Will you ship it to me? If it doesn't work out, where do I send it back? Each of those questions needs an answer based on information your team can verify. The real question for today is this: when that answer is wrong, who in your team would know, and who would fix it?

This is an educational series from Verity Score for Shopify merchants. Last time, we reconciled one product across the admin, the page, the structured data and the feed, market by market. Today, we move from the product itself to the questions people ask before they buy. Those answers can reduce uncertainty, or they can quietly add to it.

We're still with the fictional luggage team. The shopper, the questions, the policies and the answers in this story are invented for the example. No real customer, conversation or store data is involved.

Imagine a shopper in the US looking at the team's convertible backpack. They've read the page. They've checked the sleeve measurement from the product truth card. Now they want three more things before they pay.

First: how do I clean it? The bag has a coated fabric and a leather pull. The shopper wants to know whether they can wipe it, wash it, or neither.

Second: do you ship to my address? The shopper lives in a US state the team doesn't ship to often, and they're not sure the store covers it.

Third: if I return it, where does it go? The team is based outside the US. The shopper wants to know whether a return means an international parcel, at their own expense, or something simpler.

Now let's look at where those answers live inside the team's business.

The care answer is on the product page, in one line: "Wipe clean." It was copied from the supplier's sheet. But the team's customer service inbox has an older reply template that says the bag can be hand-washed in cold water. Two answers, from two different people, written at two different times. Nobody remembers who wrote the template.

The shipping answer lives in the shipping settings, which the operations person manages. There's also a shipping page written by the founder last year. Nobody has checked recently whether they agree.

The return answer lives in the store's refund policy. It describes a return address in the country where the team is based. But the team recently started working with a returns partner for US orders. The policy hasn't been updated. So the written answer and the real route are different.

None of these are dramatic failures. They're ordinary drift. And each of them is exactly the kind of gap that makes a careful shopper hesitate, or makes a careless one buy and then regret it.

Shopify offers an app called Knowledge Base. Shopify's documentation, which we reviewed on October twenty-second, twenty twenty-six, describes it as a place to manage answers about your store. Two details in that documentation matter for this story.

The first: the FAQs in Knowledge Base are not displayed directly on your storefront. So adding an answer there doesn't put it on your product page, or on your shipping page. If a shopper only reads your site, they won't see it there. That's not a flaw. It just means Knowledge Base is a separate place where answers live, and you have to keep it consistent with the others.

The second: Shopify distinguishes between the accuracy of answers and how often they appear. According to its note, the app does not affect how often your store appears in AI platform results. So filling in Knowledge Base doesn't make your store show up more. What it can help with is accuracy, which is what this episode is about. And it doesn't mean you control what every assistant says about you. Other services form their answers in their own ways, from their own sources.

Shopify's documentation on managing FAQs adds more detail. It says that the generated FAQs draw on information such as your store's languages, shipping and returns. That's useful, and it's also a warning. If your refund policy is out of date, as it is for our team, a generated answer about returns could repeat the outdated route. The documentation advises checking the source before you override an answer. In other words, if a generated answer is wrong, the first question isn't "how do I rewrite this answer?" It's "where did it come from, and is that source wrong too?"

There's one more detail worth knowing. The documentation describes a testing tool, and says that the questions you send through it are marked as Test, separate from real questions. So when your team tries out a few questions to see what comes back, those tries don't pretend to be shoppers. Keep it that way in your own notes, too. A test question you wrote is not evidence of what shoppers ask.

So here's the picture. An answer to "where do I send it back?" can exist in at least four places: your written policy, your support templates, your Knowledge Base, and whatever a service has assembled from the sources it can reach. The first three are yours to fix. The last one is not directly under your control. Start by checking the sources you can verify; if the error is in another copy, address that copy too.

The team decides not to write more answers. Instead, they audit the ones they already have, and they give every answer a source owner. That's your action this week: audit five pre-purchase answers and assign a source owner to each.

Here's what that looks like. Choose five questions buyers actually ask before buying. If they come from real customer messages, use them only if you're allowed to, and remove anything that identifies a person. For our team, the five might be: how to clean the bag, whether the laptop fits, which addresses they ship to, how long delivery takes, and how returns work from the US.

For each question, write down every place an answer currently exists: product page, policy page, support templates, Knowledge Base if you use it. Copy the exact wording from each place, with the date. Then check whether they agree.

Next, name the source. Not the page where the answer appears, but the place where the truth is decided. For shipping, that's the shipping settings. For returns, it's the actual returns process, then the refund policy that describes it. For care, it's the supplier's documentation, checked against the actual material. And then name the owner of that source: the person who will know when it changes.

For our team, the results look like this. Care: two conflicting answers, and the source is the supplier's sheet, owned by the product lead. Until the supplier confirms the care method in writing, the team removes the hand-wash template and keeps "wipe clean," attributed to the supplier. Shipping: one source, the settings, owned by operations; the founder's page needs to be checked against it. Returns: the source is the new returns process, owned by operations; the refund policy needs updating first, and every other answer follows from it.

Notice the order. Fix the source, then the policy text, then the copies. If you rewrite the FAQ first and leave the policy wrong, you've just created a third version. And when the source changes later, say the returns partner changes its address, the owner is the person who triggers the update everywhere else. Without an owner, the next change makes the answers drift apart again.

Also notice what's not on this list. There's no answer written to sound good for an AI assistant. There's no promise you can't keep, like delivery dates you don't control. And there's no claim that fixing these answers will make anything appear in any assistant. You're doing this because a shopper deserves one consistent answer, and because your team deserves to know who keeps it true. The full audit template is in the show notes.

So: an answer has a source, and a source has an owner. Knowledge Base is a separate place where answers live; it's not a switch for visibility, and not a control panel for every assistant. Next time, we'll leave your own store and look at what other people say about your products. Some of it will be helpful. Some of it will be wrong. And none of it is an endorsement just because it mentions you.

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