Where online shops lose sales, and how to find the leaks.
In this article
When sales are flat, the instinct is to buy more traffic. It’s the most visible lever, and it feels like doing something. But for most online shops, traffic isn’t the problem. The problem is that a large share of the people who already arrive leave without buying, and nobody knows exactly where or why.
That’s a leak, and leaks are expensive precisely because you never get a bill for them. You just don’t get the sale.
This article walks through the stages where online shops typically lose sales, how to find the leaks in your own data and what to fix in what order.
More traffic vs fixing the leaks
Let’s start with the maths, because it settles the argument about where to spend.
Say your store gets 10,000 visitors a month, converts 1% of them and has an average order of £60. That’s 100 orders and £6,000 a month.
| Option | What changes | Orders | Revenue | What it costs |
|---|---|---|---|---|
| Do nothing | 10,000 visits at 1% | 100 | £6,000 | Nothing, and nothing changes |
| Buy 50% more traffic | 15,000 visits at 1% | 150 | £9,000 | An extra 5,000 paid clicks, every month, for as long as you want the result |
| Fix the leaks | 10,000 visits at 1.5% | 150 | £9,000 | A one-off piece of work, and the improvement applies to every future visitor including the paid ones |
Both routes get you to the same number. One of them is a bill you pay forever. The other is an asset. And if you do both, the traffic you buy is worth 50% more than it was before you started.
That’s why conversion rate optimisation (CRO) is usually the first job, not the last.
Stage 1: the landing
Where the leak is: the first page a visitor sees. For a lot of stores that’s the homepage, but for paid and organic traffic it’s usually a category or product page.
What goes wrong:
- Speed. If the page takes more than three seconds to load on a phone, a chunk of visitors are gone before they see anything. Oversized images, too many apps and unoptimised themes are the usual causes.
- Mismatch. The ad promised one thing and the page shows another. Someone clicks an ad for “waterproof walking boots” and lands on a general footwear page.
- No orientation. It’s not clear what the shop sells, who it’s for or why to buy here rather than anywhere else.
How to find it: in GA4, look at the landing page report and compare engagement rate and bounce by page. Run your top landing pages through PageSpeed Insights on mobile. If your paid campaigns send everyone to the homepage, that’s a leak on its own.
Stage 2: finding the product
Where the leak is: between landing and viewing a product page. People arrived, looked around and couldn’t find what they wanted.
What goes wrong:
- Navigation that mirrors the business, not the customer. Categories named after internal departments or supplier ranges rather than how shoppers think.
- Site search that returns nothing useful. Someone types “black hoodie” and gets a blank page because the product is called “Charcoal Pullover”.
- Filters that don’t work. Missing attributes mean the filter hides products that should be there.
- Bad product data. Inconsistent titles, missing specifications and wrong categories mean products simply don’t surface.
How to find it: GA4’s site search report tells you what people searched for and whether they got results. Look for searches with zero results and searches followed by an exit. Compare the number of sessions that view a product page with the number that land. If more than half never reach a product, finding things is the problem.
This one is often a data problem rather than a design problem. Stores with hundreds or thousands of lines tend to have product data that has drifted over years, and the fix is cleaning it up and setting rules so it stays clean.
Stage 3: the product page
Where the leak is: people view the product and don’t add it to the basket. This is where most of the money goes missing.
What goes wrong:
- Not enough to decide on. One photo, a supplier description copied from the manufacturer and no sizing, fitment or compatibility information. The visitor can’t be sure it’s right, so they don’t risk it.
- The price is a surprise later. Delivery costs and lead times only appear at checkout.
- No proof. No reviews, no “customers also bought”, nothing that says other people have bought this and been happy.
- The button is hard to find. Below the fold on mobile, same colour as everything else, or hidden behind a variant selector that hasn’t been explained.
- Out of stock with no alternative. The product shows but can’t be bought, and the page offers nothing else.
How to find it: look at add-to-basket rate by product. Most ecommerce platforms report this, and GA4 does if ecommerce events are set up. Products with lots of views and a low add-to-basket rate are leaking. Session recordings and heatmaps (Microsoft Clarity is free) will show you what people are doing on those pages, and it’s often scrolling around looking for information that isn’t there.
One thing worth checking if returns are a problem. A product finder or fit-check tool on the page (“will this fit my vehicle?”, “which size do I need?”) often cuts both returns and abandonment, because it answers the question the visitor was going to leave to find out.
Stage 4: the basket
Where the leak is: items are added and then the visitor leaves without starting checkout.
What goes wrong:
- Delivery shock. The delivery cost appears for the first time and it’s more than expected. This is the single biggest cause of basket abandonment in almost every study ever done.
- No sense of urgency or reassurance. Nothing says how long delivery takes, what the returns policy is or that payment is secure.
- Discount code box with no code. Visitors leave to go and search for one, and don’t come back.
- Friction. The basket page is cluttered, the quantity controls are fiddly, or the “continue to checkout” button is below a wall of cross-sells.
How to find it: compare sessions with an add-to-basket event against sessions with a begin-checkout event. The gap is the leak. If you don’t have an abandoned basket email sequence running, this is also where the quickest recovery is. A three-step reminder sequence typically recovers a meaningful share of these baskets without a discount being needed until the last email, if at all.
Stage 5: the checkout
Where the leak is: checkout started, order not placed.
What goes wrong:
- Forced account creation. “Create an account to continue” is still one of the most reliable ways to lose a sale. Guest checkout should always be available.
- Too many fields. Asking for a phone number, a company name and a fax number for a £15 order.
- Payment options. No PayPal, no Apple Pay or Google Pay, no option the visitor trusts. On mobile especially, a wallet option removes most of the typing.
- Errors that don’t explain themselves. A red box that says “invalid” with no indication what’s wrong.
- Mobile layout breaks. The checkout was designed on a desktop and tested on a desktop. Half your customers are on a phone.
- Address lookup fails. Or there isn’t one, and typing a full address on a phone is enough to make people give up.
How to find it: checkout funnel reports in your platform or GA4 will show the step-by-step drop. The biggest drop is the thing to fix first. Test the checkout yourself on a real phone, with a real card, from a slow connection. You’ll find things.
Stage 6: the silent leaks
Some leaks don’t show up in the funnel because they sit underneath it.
Broken tracking. If conversion tracking isn’t set up properly, you can’t see any of the above accurately, and your ad platforms are optimising towards the wrong signal. We’ve seen stores where orders were being double counted, and stores where a whole payment method was never recorded. Check the numbers in your analytics against the orders in your platform. If they don’t match, fix that before anything else.
Wrong traffic. A store shipping only to the UK that’s spending on ads reaching overseas visitors. They arrive, they can’t buy, they leave. The conversion rate looks terrible and the problem is targeting, not the site.
Stock and price mismatches. Shopping ads showing a price that doesn’t match the site, or products that are out of stock. Google disapproves them, or shoppers arrive and feel misled.
Nobody owns the site. It was built, launched and left. Apps have been added, the theme hasn’t been updated, speed has crept up and nobody has looked at the checkout in a year.
How to find your own leaks: a short method
You don’t need an expensive tool to do this. You need an afternoon and a willingness to look.
- Check the tracking first. Compare orders in your analytics with orders in your platform for last month. If they’re within a few percent, carry on. If not, fix tracking before you trust any other number.
- Map the funnel. Sessions, product views, add to basket, begin checkout, purchase. Write the five numbers down. Work out the percentage drop at each step.
- Find the biggest drop. That’s your biggest leak. It’s usually product page to basket, or basket to checkout.
- Watch ten sessions. Set up Clarity or similar, filter to sessions that dropped at that step and watch what people did. You’ll see the problem within ten recordings, and it’s often not what you expected.
- Fix one thing. Change the most obvious problem. Measure for two to four weeks. Then move to the next one.
The temptation is to redesign everything at once. Resist it. One change at a time means you know what worked, and that knowledge is worth more than the change itself.
What to fix first
If you want a rule of thumb for priority, it’s this.
| Priority | Fix | Why first |
|---|---|---|
| 1 | Tracking | Everything else depends on accurate numbers |
| 2 | Delivery costs shown early | Biggest single cause of abandonment, quick to fix |
| 3 | Guest checkout and wallet payments | Removes the most friction for the least effort |
| 4 | Mobile speed on top landing pages | Affects every visitor from every source |
| 5 | Product page information and proof | Biggest leak by volume, takes longer to fix across a range |
| 6 | Abandoned basket emails | Recovers sales from leaks you haven’t fixed yet |
| 7 | Product data and site search | Slower, but compounds across the whole catalogue |
Frequently asked questions
What’s a good ecommerce conversion rate?
It varies a lot by sector, price point and traffic source. A luxury furniture store and a discount stationery store aren’t comparable. Measure against your own baseline and aim to improve it, rather than chasing a figure from a benchmark report.
How much traffic do I need for CRO to be worth doing?
A/B testing needs enough visitors to give a reliable result, usually thousands of sessions a month. On lower-traffic stores you skip the testing and fix the known problems directly. The leaks above are so common that fixing them without testing is almost always worth it.
Should I redesign the site?
Probably not yet. Most leaks are specific and fixable without a rebuild. A redesign that doesn’t address them just moves the leaks to prettier pages. If the platform itself is the limit (it can’t do guest checkout, it can’t show delivery early, it’s slow no matter what), then a rebuild is on the table.
Is this different for a trade or B2B store?
The stages are the same. The leaks tend to sit around pricing (trade customers can’t see their prices until they log in), account approval and minimum order rules. The method for finding them doesn’t change.
Want someone to look?
If you’d rather have us find the leaks, talk to us about CRO and we’ll go through your funnel together. Or for a quick first read, our free 5-minute marketing audit scores your business across seven foundations, website conversion among them, and tells you what to fix first.




