How to write product pages that answer your customers

Knowing how to write product pages starts before the first sentence: it starts with the questions customers ask that the page doesn't answer. The ShopGuru agent, which also prepares product pages, uses those questions to propose better ones. Here is the method, what a useful product page has to say and where the store's team decides.

Article summary

• A well-written product page answers the questions customers ask before buying: measurements in use, materials, care and compatibility.
• The best clues for writing product pages are in customer service conversations, where customers say what they couldn't find on the page.
• At ShopGuru, the agent logs every gap, a question the store knowledge couldn't answer, and uses that log to propose an improved product page.
• By default, every product page the agent prepares reaches the app as a proposal and is only published with the approval of someone on the store's team, recorded with a name and a time.
• When no source has the information, the agent leaves the field blank and sends the question to your team, instead of making up an answer.
• The method pays off most in stores with recurring questions about sizes, materials or compatibility.
• To find out which of your product pages generate the most questions, the first step is a 30-minute call with a ShopGuru engineer.

Why product pages fail where the customer decides

Most product pages start life as the supplier's spec sheet: reference, composition, box dimensions. It answers what the warehouse needs to know. The customer asks something else: whether the fit is narrow, whether the fabric shrinks in the machine, whether the table fits on a balcony 80 centimetres deep.

When the page doesn't answer, one of two things happens. The customer asks, and someone on your team types the same answer for the umpteenth time. Or they don't ask, buy from another store and leave no trace at all.

The Baymard Institute, which studies the usability of online stores, publishes more than 110 guidelines for the product page alone, including descriptions and specifications. Customer service conversations give each store the concrete version: its own customers' doubts about its own products.

"The supplier's sheet says what the product is. The store's product page has to say whether it suits the person reading it."

ShopGuru team

How to write product pages from your customers' questions

Our position is simple: a good product page starts from the log of questions, before any keyword list or description template. The method has five steps and can be done by hand.

1. Gather the questions from every channel: website widget, WhatsApp, email, reviews and return reasons.
2. Group the questions by product and by type of doubt: size, material, use, compatibility, delivery.
3. Answer on the product page with a verifiable fact, where the doubt comes up, ideally close to the buy button.
4. Write in the customer's words: if they ask "is it OK for wide feet?", the page says "wide fit" and skips the supplier's jargon.
5. Review the page when the questions change, for example after a change of supplier or a new collection.

The hard step is the first one: the questions are scattered across channels and nobody has time to count them at the end of the day.

The cycle: from conversation to product page, with your team approving

The agent looks after the store's customers before and after the purchase. When it can't find the answer in the store knowledge, it says "I don't know. I'll ask." and passes the question to your team. A gap is logged: your team answers once, and that answer becomes part of the knowledge.

Then it looks at those questions with another goal. If the same doubt comes up in several conversations about the same product, it prepares a product page proposal with the answer included and shows where it came from: the conversations and the customers' sentences that prompted it. Your team approves, edits or declines. Declining is a normal decision, and the agent only proposes the same thing again if the data changes.

"Today's question is tomorrow's product page."

ShopGuru team

An example of an agent proposal in a shoe store

Fictional example. In a shoe store, the Foz Sandal got several questions in the same week, and the agent prepared this proposal for Sónia, on the store's team, to review.

Sónia approved the first two rows, changed "we recommend" to "we suggest" and answered the third question in the app. In the next proposal, the care field already comes filled in with her answer.

What a useful product page has to answer

Every store has its own questions, but some types of doubt come up in very different catalogues:

• Measurements in use, not just on the box: seat height, width of the fit, strap length.
• Materials and care: composition, washing, what happens with wear.
• Who it's for, and also who it isn't for.
• Compatibility with other products, parts or sizes from the same brand.
• What's in the box and what is sold separately.
• Delivery or exchange exceptions specific to that product, such as made-to-measure items.

The third point is the most controversial. Many stores avoid saying who a product doesn't suit, for fear of losing sales. Our position is the opposite: a product page that warns off the wrong customer avoids a return and an after-sale conversation, and earns the trust of those who stay.

"A product page that says who the product doesn't suit is one the customer believes."

ShopGuru team

Where platform AI helps with product pages and where it falls short

The description generators built into e-commerce platforms help with a first draft, from the title, the attributes and an instruction from your team. They don't hear your customers, because they aren't in the conversations.

The two can work together, and we explain the differences in detail in the article on platform AI or a custom agent.

When this method doesn't pay off

A catalogue with few conversations per product generates few questions, and proposals take a while to appear. Products that leave the catalogue within weeks rarely justify reviewing the page.

It also helps to have conversations on record. A question that arrives after hours and goes unanswered gets lost; it's one of the topics of the article on after-hours customer service.

In the Diagnostic, we analyse the store and the conversations it already has, and that's usually where the product pages generating the most doubts show up. Before that, all it takes is to book a call of 30 minutes with an engineer.

Frequently asked questions

How many questions does the agent need before proposing a new product page?

There's no fixed number. The rule is agreed with your team during Implementation: information that is missing entirely can justify a proposal from the very first question, and other doubts only count when they repeat.

Does the agent publish product pages on its own?

No, unless the store owner decides so. Publishing a product page is an action, and for each type of action the store owner chooses the autonomy level: does it on its own, asks for approval, or never does it. On day one, every action with an effect asks for approval.

Where does the information in a product page proposed by the agent come from?

From the supplier's spec sheet, your team's answers and the conversations with customers, with the source of each piece of information visible in the proposal.

Can you write product pages this way without ShopGuru?

Yes. Export the conversations, count the questions per product and review the pages with the most doubts. The hard part is keeping that work going every month, as the questions change.

Sources and editorial review

This article was written by the ShopGuru team based on how the ShopGuru agent is designed. The store, the product, the names and the dates in the example are fictional.

External source: Baymard Institute, product page research.

Published on 09/09/2026 and reviewed on 16/09/2026.