Product page CRO: the complete guide

Start with the pages that already get traffic and already lose money. A store with four hundred products does not have four hundred conversion problems. It has a handful of pages carrying most of the sessions, and the loss is concentrated there.

Pull a report of sessions by landing page for the last ninety days, with add to cart and orders for the same pages and the same dates. Sort by sessions and read from the top. A page with thousands of sessions and a weak add to cart rate is worth far more of your attention than a page with ninety sessions and no orders, even though the second page looks worse on paper.

Then sort the queue into three groups, because they need different work. Pages with plenty of traffic and a weak add to cart rate have a page problem: something on the page is not answering a question the buyer has. Pages with a healthy add to cart rate and few completed orders have a cart or checkout problem, which is a different piece of work entirely. Pages with almost no traffic have a visibility problem, and no amount of copy will fix that.

Keep paid landing pages in their own list. When Shopping ads or social ads send people straight to a product page, that page is doing two jobs at once. It is the first impression of your brand and the place the decision gets made, and it usually needs more of the trust material than a page a returning customer lands on.

Write the queue down somewhere permanent, with today's numbers beside each page. You will want those before figures in three months, and nobody remembers them accurately once the page has changed. This list is the plan for the rest of the guide: everything below gets applied to the top few pages first, then rolled out to the rest of the catalogue as a template once you know what worked.

Watching is cheaper than guessing and it is faster than arguing. Before you rewrite anything, sit beside five people while they try to buy from your store on their own phone, and record what they say out loud. Three can be existing customers. Two should be people who match your buyer but have never seen the store, because they will trip on things your customers have already learned to ignore.

Give each person a real task, never instructions. Something like: you need a winter jacket for an eight year old, find one you would actually buy and get as far as the payment screen. Then stay quiet. The moment you explain something is the moment you stop learning, because in real life nobody is sitting there to explain.

Write down every pause, every pinch to zoom, every scroll back up the page and every question asked aloud. That list is your defect list, and it is more useful than any opinion in the room. A pinch to zoom means the photograph is too small or the detail is not shown. A scroll back up means the information they wanted was not where they expected it. A question asked aloud is a line of copy that is missing from the page.

If you cannot get five people, session recordings do for a first pass. Watch twenty recordings of your busiest product page at phone width and skip to the moments where the scroll stops or reverses. Heatmaps are the weakest of the three, because they show where people looked, not what they were looking for.

Do not fix anything yet. Collect the list, count how many people hit each wall, then start with the wall that stopped the most people. Two people struggling with the same thing is a pattern worth money. One person struggling is one person.

The first screen on a phone has one job: tell the visitor this is the right thing for them and that they are in a real shop. Everything else can wait for a scroll. If a buyer has to scroll twice before they know what the product is and what it costs, you have already lost some of them.

Six things belong in that first screen. What the product is, in words a stranger would use. One photograph that shows it clearly. The price. The variant choice if there is one, so the price is not a mystery. One line that says who it is for or what it does that a similar product does not. And the add to cart control, either visible or one short scroll away.

What pushes those six things down the page is almost always something you added for a good reason. A stacked announcement bar, a promotional banner, a large logo, a slow carousel, a cookie prompt that covers the price, a chat bubble that sits over the button. Load your own page on a mid range phone and take a screenshot of the first screen. If the price is not in that screenshot, the page starts with a problem.

The product name matters more than most stores treat it. A name that only makes sense inside your brand, like the Aurora Mark Two, forces the visitor to keep reading just to work out what they are looking at. Put the plain description next to it. Aurora Mark Two, insulated stainless steel water bottle, one litre. That single line also helps the page get found, because it is close to what people type.

One line of positioning under the name does the rest of the work. It should answer the question a buyer is really asking, which is whether this fits their situation. For a bottle that might be whether it keeps drinks cold all day. For a jacket it might be whether it is warm enough for a real winter. Say the specific thing, not that it is premium quality.

Photography is the part of a product page that does the work a shop floor used to do. In a shop the buyer picks the thing up, turns it over, checks the size against their hand and reads the back of the box. Your photo set is the only way they can do any of that, so build it as a sequence of answers rather than a gallery.

A set that works usually runs like this. A clean shot of the whole product on plain background. The product in use, held or worn or in a room, so scale is obvious. A scale reference against something familiar. Close ups of the parts people worry about, which is stitching on clothing, the seal on a container, the connector on electronics, the finish on furniture. The back of the packaging where there are ingredients, materials or care instructions. What is actually in the box, laid out. And for anything with moving parts or assembly, a short video.

Scale is the single most common gap. Buyers return products because they arrived smaller or larger than expected, and almost every one of those returns was avoidable with one photograph. If the product sits on a desk, photograph it on a desk with a laptop beside it. If it is jewellery, photograph it on a wrist or a neck, not only on white.

Video earns its place when the product moves, folds, pours, stretches or has to be assembled. Ten seconds with no music and no titles is enough. Do not autoplay a heavy video file above the fold, because that becomes the speed problem later in this guide.

Keep the treatment consistent across the catalogue. Same background, same crop, same lighting, first image always the same type of shot. Consistency reads as competence to a visitor scanning a collection page, and it makes the feed you send to shopping channels much easier to keep clean. Then compress every file properly and give each one honest alt text, because both the search engines and screen readers depend on it.

Unexpected delivery cost is the most common reason a cart is abandoned, and the cheapest place to deal with it is the product page. If a buyer only finds out the postage at the last step, you have paid for the visit, earned the interest and then given it away at the checkout.

Put three things beside the price. What delivery costs, or the order value that makes it free. When it will arrive, expressed as a date range rather than a number of business days. And what happens if they want to send it back. A line that reads free delivery over sixty dollars, arrives Tuesday to Thursday, thirty day returns does more for conversion than most redesigns.

If you have a free shipping threshold, show the gap. Telling someone they are twelve dollars away from free delivery while they are still on the product page gives them a reason to look at a second item, which is a different lever doing useful work on the same page.

Delivery dates need a cut off to be honest. Order in the next four hours and it ships today is fine when it is true and the timer resets properly. A permanent countdown that restarts every time the page loads is a trick, and buyers who spot it stop believing the rest of the page.

If you ship internationally, say what happens with duties and taxes. Buyers who have been stung by a customs bill once are careful forever, and a plain sentence saying duties are collected at checkout so there is nothing to pay on delivery removes a real objection.

Finally, make the returns promise specific. Thirty days, unworn, free return label, refund to the original payment method. Vague reassurance about a hassle free policy does nothing, because every store says that. The specifics are what change a decision.

Every unanswered question on a product page is a chance for the buyer to leave and look for the answer somewhere else. Some of them come back. Most do not, because the place they went to find the answer also sells the product.

You already own the list of questions. It is in your support inbox, your chat transcripts, the questions section under your reviews, and the search box on your own site. Spend an hour reading the last two hundred pre purchase messages and tally what people ask before they buy rather than after. The same six or seven questions will appear again and again, and they will be different for every category.

For clothing it is sizing against a known reference, fabric weight, whether it shrinks, and how it fits someone of a given height. For supplements it is dosage, ingredients, allergens and how long a container lasts. For electronics it is compatibility, what is in the box, power and warranty. For furniture it is dimensions, assembly time and whether it fits through a standard door. Write your own list, not a generic one.

Then put the answers where the question occurs, not all of them in one accordion at the bottom. Sizing belongs beside the size selector. Compatibility belongs beside the variant picker. Battery life belongs under the headline claim about battery life. A buyer should never have to hunt for the answer to a question your page provoked.

A short specification table pays for itself on any technical product. Plain rows, real units, no marketing language. It is scannable for a person in a hurry and it is the part of the page that answering engines and search results tend to quote, so it earns visits as well as orders.

Keep the description underneath for the people who read. Written well it does two jobs: it convinces the careful buyer and it gives search engines the vocabulary your buyers actually use.

The add to cart step should be the one obvious action on the page, and it should never fail silently. Most stores get this almost right and lose orders on the details.

One primary button, in one colour used for nothing else on the page. On a phone, keep a sticky bar with the price and the button visible as the visitor scrolls, because long pages otherwise bury the action a thousand pixels away. Every other action on the page, wishlist, share, compare, size guide, should look secondary. When three buttons compete, the visitor picks none of them.

Variant selection is where the quiet losses happen. If a buyer has to choose a size and a colour, show both clearly with the unavailable combinations visibly out of stock rather than silently missing. Never let the page throw an error that says please select an option after the button is pressed, when you could have defaulted the choice or made the requirement obvious before the press. If a colour is sold out in one size, say so on the swatch.

Be honest about stock. Low stock indicators work when they are real and stop working the moment a buyer sees the same two left message on a page for a fortnight. If your stock system knows the number, use it. If it does not, say nothing.

What happens after the click matters as much as the click. A cart drawer that slides in, confirms what was added, shows the running total and progress to free shipping, and offers one clear way to continue shopping and one to check out, is the pattern that works. A full redirect to a cart page interrupts anyone who wanted to buy two things. A silent add with a tiny number changing in the corner leaves people wondering whether it worked, so they press again and end up with two.

One relevant add on can sit in that drawer, and only one. A wall of unrelated products in the drawer is noise, and noise at the moment of decision costs orders rather than raising them.

Proof works when it appears at the moment of doubt, not when it is stacked at the bottom of the page. The doubt on a product page has a shape: it rises when the buyer sees the price, again when they choose a variant, and again just before they press the button.

So put a rating and the review count near the price, where it answers the question of whether this is worth the money. Put fit or performance reviews near the variant picker, where the buyer is asking whether this particular one is right. Put the returns and warranty line next to the button, where the buyer is asking what happens if this goes wrong.

Reviews are worth more when they are specific and worth less when they are all five stars. Leave the critical ones visible. A run of perfect scores reads as filtered, and buyers know it. A three star review that says the colour is darker than the photo, followed by a reply from you saying the photo has been updated, builds more confidence than another line of praise. Ask reviewers for the details that matter in your category, which might be height and usual size for clothing, or how long they have owned it for anything durable. Photos in reviews do a job your studio photography cannot.

If the product is new and has no reviews yet, use what you do have. The number of units sold, a note from the person who chose the product, the testing you did, the materials you specified, the returns window. Silence is worse than a small amount of honest proof.

Risk reversal is the other half. The clearer and shorter your returns promise, the less the buyer has to think about the downside. Say the window, say the condition, say who pays for return postage, and put it in a sentence rather than a link to a policy page. If your policy is genuinely generous, that sentence is one of the highest value pieces of copy on the whole page.

A slow product page loses buyers before any of the copy gets a chance. Test on a mid range phone on a mobile connection, not on your laptop on office wifi, because the gap between those two experiences is where most stores are quietly losing orders.

On nearly every product page the largest element to load is the main product image, so that is where the speed work starts. Serve it in a modern format, size it for the device rather than sending a two thousand pixel file to a phone, and let it load before anything below the fold. Lazy load everything under the first screen, including the review widget and the recommended products carousel.

Then count your apps. Every installed app that touches the storefront adds script to the page, and stores accumulate them: a review widget, a currency switcher, a popup tool, an upsell app, two analytics tags, a chat bubble, a wishlist, a sizing tool, an abandoned browse tracker. Open your theme and list every third party script the product page loads. Remove the ones you no longer use, and load the ones you keep after the page is usable rather than before.

Fonts are a small and easy win. Two weights of one font family, loaded locally, with a fallback that displays immediately, beats four weights fetched from elsewhere while the text sits invisible.

Do not let the page jump around while it loads. Reserve the space for images, banners and the review stars so nothing pushes the add to cart button down as the visitor is reaching for it. Layout that shifts under a thumb is one of the fastest ways to lose a sale you had already won.

Measure with real user data over a few weeks rather than a single lab score. Lab tools tell you what to fix. Field data tells you whether the fix reached the people who matter, on the devices and networks they actually use.

The number that tells you whether product page work is succeeding is add to cart rate on that page, not store wide conversion rate. Store wide conversion moves for a dozen reasons that have nothing to do with your page, including traffic mix, a sale, a public holiday or one influential post. The page metric is the one you can actually attribute.

So measure per page: sessions, add to cart rate, and then orders and revenue for sessions that saw the page. Write the numbers down before you change anything, note the date of each change in a simple log with one line saying what you did, and annotate your analytics so the timeline is readable a year from now.

Give changes enough time and enough orders to mean something. A page that gets a handful of orders a week cannot tell you anything in three days. If your busiest page produces a steady flow of orders, a fortnight before and a fortnight after is a reasonable read. If it does not, group similar pages, apply the same change to all of them, and look at the group.

Split testing is worth it when the page has enough traffic for a test to reach a conclusion in weeks rather than seasons. Below that, testing tools mostly add script to the page and give you results you cannot trust. Make the change, watch the page metric, keep it or reverse it.

Beware of the two traps. The first is judging a change during an unusual period, such as a sale week or the run up to a holiday. The second is stacking five changes on the same day and then being unable to say which one worked, which matters because you want to apply the winner across the rest of the catalogue.

When a change works on your top page, roll it into the template so every product page inherits it. That is where product page work stops being a project and starts being compounding.