Where E-commerce Conversion Actually Leaks

E-commerce conversion optimization has a reputation problem. It's been so thoroughly associated with button colours and urgency banners that serious teams sometimes skip it entirely. That's a mistake — but so is the button-colour version. The real leaks are further upstream, and they're product problems rather than page problems.

Here's where I actually find conversion in consumer e-commerce, roughly in order of how much it's usually worth.

1. Product page optimization starts with the wrong question

Most product pages are built to describe. The customer is trying to decide. Those need different content.

Description content answers "what is this?" Decision content answers the questions that are actually blocking the purchase: will it fit, how long until it arrives, what happens if it's wrong, is this the right one of the four similar things you sell, and has anyone like me bought it.

The fastest audit I know: list the five questions your support inbox gets most about a product, then check whether the product page answers them above the fold. Usually two of the five aren't answered anywhere on the page. Every unanswered question is a customer either leaving or emailing you, and most of them leave.

2. Shipping cost arrives too late

The single most reliable cause of cart abandonment in e-commerce, and it's not really a checkout problem — it's a disclosure sequencing problem.

A customer who learns about a delivery charge at step three of checkout has already invested effort and is now recalculating a decision they thought they'd made. That feels like a bait-and-switch even when nothing dishonest happened. The same charge disclosed on the product page loses far fewer people, because it's priced into the decision instead of ambushing it.

You will lose some product-page traffic by being upfront. You'll lose fewer carts, and the customers you keep will be the ones who were actually going to buy.

3. Search doesn't work and nobody knows

On-site search converts at several times the rate of browsing, because searchers have intent. Which means search failures are unusually expensive — you're losing your best-qualified traffic.

Pull your top hundred internal search queries and check the results yourself. You'll typically find a cluster of zero-result searches, which are customers telling you in plain language what they wanted to buy and couldn't find. Sometimes it's a synonym problem. Sometimes it's a product you don't stock, which is a merchandising insight you got for free.

This is dull work and it's frequently the highest-return afternoon available.

4. Mobile is treated as a smaller desktop

Most consumer e-commerce traffic is mobile and mobile conversion is almost always lower — which is often accepted as a fact of nature rather than investigated.

Some of the gap is real: mobile sessions include more browsing and idle scrolling. But a lot of it is that the experience was designed at 1440px and adapted downward. Filters buried behind two taps, image galleries that fight the scroll, forms that summon the wrong keyboard, a checkout that doesn't offer the payment methods people actually use on a phone.

Test on a real device on real network conditions, not a resized browser window. The difference is always instructive and usually embarrassing.

5. Returns policy is hidden

Counterintuitive but consistent: making the returns policy prominent increases conversion. Uncertainty about the downside is what stops people committing, and a clear, generous policy removes it.

Teams bury it because they're worried about return rates. Return rates do go up somewhat. Revenue usually goes up more, because you converted people who otherwise wouldn't have risked it. Check both numbers before deciding — but check them, rather than assuming.

6. You're optimising a page when the problem is the traffic

Before rebuilding anything, segment conversion by acquisition channel.

If one channel converts at a fraction of the others, the page is fine and the targeting is wrong. No amount of page work fixes a mismatch between what an ad promised and what the page delivers. I've watched teams spend a quarter on a product page redesign when the actual problem was a single campaign bringing in badly matched traffic at volume.

This is the same pattern as activation problems that turn out to be acquisition problems, and it's worth ruling out first because it's the cheapest thing to check.

How to work through it

The method matters more than the list, because your leaks won't be in this order.

1. Instrument the funnel properly. Every step, segmented by device and channel. Most e-commerce teams can tell you overall conversion and not much else, which makes every subsequent conversation speculative.
2. Find the biggest drop. Not the worst-looking step — the one where the most absolute revenue is lost. A 5% improvement on a high-volume step beats a 40% improvement on a rare one.
3. Watch people fail at it. Twenty session recordings of that specific step. You will see the cause within twenty minutes, and it will rarely be what the team predicted.
4. Ask a few of them why. Recordings show where people stop; interviews explain what they were thinking. You need both.
5. Fix the cause and measure. Agree beforehand what should move and by how much, so you find out whether it worked instead of arguing about it afterwards.

What this isn't

It isn't a testing programme. Testing is valuable once you have the traffic to reach significance in a reasonable window, and most early-stage e-commerce companies don't — running a two-month test on a low-traffic step is a way to spend a quarter learning nothing.

Below that threshold, the right approach is diagnosis and judgment: find the leak with data, understand it with research, fix it properly, and watch the number. Save the experimentation infrastructure for when you have the volume to earn from it.

And it isn't a substitute for the product being good. Conversion work makes an existing value proposition legible. It cannot manufacture one — that's a strategy problem, and no checkout redesign will solve it.

If your conversion has been flat for a while and you're out of ideas that aren't button colours, the intro call is free.