What a product page has to answer before someone will buy

In a shop, a customer picks up the jumper, holds it against themselves, feels the fabric, checks the label and asks someone whether there’s a medium in the back. On a product page, every one of those moments has to be handled by photos, copy and a few lines of delivery information. When the page can’t answer a question, the shopper either guesses, which leads to returns, or goes to find out somewhere else, which often means a competitor.

That makes the product page less about persuasion than people expect. Most of the work is answering questions plainly, in the order people ask them.

What usually goes wrong

Baymard Institute has benchmarked product pages across hundreds of large ecommerce sites for years, and its current findings read like a list of unanswered questions. Its 2026 update reports that 37% of sites don’t provide images that show a product in scale, even though 42% of users in its testing tried to judge size from the photos alone. 23% don’t show products on a human model, which matters most for anything you wear. 44% don’t show or link to the returns policy from the main product content, while 60% of users looked for it on the product page.

Cost is the gap that does the most damage later. Baymard found 67% of sites don’t show an estimated total cost, delivery included, near the buy section, so shoppers only learn the real price in the basket. That surprise then gets counted as a checkout problem when it started here.

Variant selection is another quiet leak. Baymard reports that 57% of sites use drop-downs for size selection, which hides the options, so people only discover their size is out of stock after opening the menu. A shopper who picks a colour, opens the size list and finds nothing available has wasted effort and is likely to feel a bit misled.

Reviews are often treated as decoration. Baymard found 89% of sites don’t respond to negative reviews, which leaves the most worrying comment on the page unanswered. The Spiegel Research Center at Northwestern found that a product with five reviews was far more likely to be bought than one with none, with the benefit flattening sharply after the first few. It also found purchase likelihood tends to peak for ratings between 4.0 and 4.7 stars and falls as ratings approach a perfect 5.0, because shoppers suspect a perfect score is too good to be true. Hiding the bad reviews can make the good ones less believable.

Where to look for evidence

Customer service is the best source most teams ignore. Pull a month of pre-sale questions from chat and email and sort them by product. If people keep asking whether the bag fits a laptop, the page hasn’t said so. Returns reasons are the other half. “Smaller than expected” and “colour different from photo” are product page problems that show up weeks later in a different department’s report.

In analytics, look at how many product page visitors add to basket, broken down by device and by traffic source, and how often people open the size guide, the delivery tab or the reviews before buying. Heavy use of one element suggests it’s answering a question the main content doesn’t. Session recordings will show people zooming into photos, scrolling up and down hunting for delivery details, or opening and closing the size drop-down. Usability testing with five or six people asked to choose a product and decide whether to buy will surface most of the big gaps quickly.

What is worth testing

Test the answers to the questions you’ve found, starting with the ones that cost the most. Showing delivery cost and an expected delivery date next to the price is often the first candidate, because it moves the surprise forward to where people can decide calmly. Adding an in-scale image, or a photo on a model, for products where size or fit is the common return reason is another. Showing size options as buttons with out-of-stock sizes visible, rather than a drop-down, is cheap to build and easy to measure.

Measure more than add to basket. A change that makes people more confident should raise completed orders and, over time, lower returns. A change that only makes the button bigger might raise add to basket while orders stay flat, which tells you very little. Ask for returns or refund rate as a guardrail on any test about fit, size or appearance, and accept that you may need to read it weeks after the test ends.

The version to refuse

The shadow version of the product page is the one that tries to rush the decision instead of supporting it. “Only 2 left” when there are hundreds in the warehouse, a countdown on an offer that resets every time you visit, reviews filtered so only the five-star ones appear, or reviews the business wrote or paid for. In the UK the law on this is clear. The Digital Markets, Competition and Consumers Act 2024 bans falsely claiming an offer is only available for a limited time, and bans submitting or commissioning fake consumer reviews, with publishers required to take reasonable steps to prevent and remove them. Real stock levels, stated for the specific variant that’s running low, are useful information. Invented ones will usually lift a short test and cost you trust after it, as the earlier lesson on social proof and herding covers.

What to ask whoever builds the test

Ask which customer question the change answers, and where the evidence for that question came from. Ask whether the primary metric is orders or only add to basket, and what the guardrail is for returns. Ask how the test handles products with very different traffic, because a change tested across the whole catalogue can be driven by a handful of best sellers. Ask whether any stock, urgency or review content in the variant is pulled from live data, and how they know. If an agency proposes a template product page redesign tested as one big change, ask how you’ll know which part of it worked, and whether a few smaller tests on the biggest unanswered questions would teach you more.

Put it to work

Pick one high-traffic product page and write down the five questions a first-time buyer would ask before paying. Check whether the page answers each one above the add to basket button, below it, somewhere else on the site, or not at all.

Your answer

Show the worked example

There is no right answer. Write yours first, then compare.

Sources

1. Baymard Institute: Product Page UX, 10 Pitfalls and Best Practices
2. Spiegel Research Center: How Online Reviews Influence Sales
3. Digital Markets, Competition and Consumers Act 2024, Schedule 20 (banned practices)