How to increase ecommerce conversion rate

The benchmark problem

The first thing most people do is look up an average and compare themselves to it. Do not. Published ecommerce averages vary by roughly an order of magnitude depending on category, how much of the traffic is branded, and whether the source counts sessions or users, so a figure lifted from a report built on a different mix tells you nothing about your store.

The only comparison that means anything is your own store last quarter, segmented by device and by traffic source. A site-wide number hides the two things that matter: mobile is usually much worse than desktop, and paid traffic usually converts differently from organic.

Where it is actually lost

Two places, on almost every store we look at. Between the collection page and the product page, where people cannot work out which of your products is the one they want. And at the delivery step, where a cost or a timescale appears later than the buyer expected.

Neither is a design problem in the way the word is usually meant. Both are information problems, which is why a store can be redesigned beautifully and convert exactly as badly afterwards.

The order worth fixing

1. 1. Whatever is broken on mobile Not "optimise for mobile", but find the thing that is actually failing. Most stores have at least one, and it is usually invisible on a desktop browser resized to look like a phone.
2. 2. Delivery cost and returns, earlier The most common abandonment reason is a cost appearing later than expected. Moving it earlier loses you some sessions and wins you orders, because the people who leave were never going to buy at that price anyway.
3. 3. The objection specific to your category Every category has one question that decides the purchase: fit, ingredients, compatibility, how long it lasts. Find it by asking whoever answers your customer emails, then answer it on the product page before it is asked.
4. 4. Steps out of the checkout Every field is a chance to leave. Guest checkout, address autocomplete, and not asking for anything you do not need to fulfil the order.
5. 5. Collection pages that help people choose The least-designed template on most stores and where people actually shop. If a visitor cannot tell your forty products apart, the product page never gets a chance.

Mobile, separately

Worth its own section because it is where most of the loss is and where the least attention goes. How to improve ecommerce conversion rate on a phone starts with using your own store on one, on a bad connection, having not seen it for a month.

The recurring faults barely vary between stores: tap targets crowded together, a filter interface that covers the products it is filtering, imagery that pushes the price below the fold, and a delivery estimate that only appears once an address has been typed.

What not to do first

Do not start with a redesign, do not start with a testing tool, and do not start by copying a store you admire. All three are ways of spending money before you know what is wrong, and the third is actively risky because their design is a solution to a problem you may not have.

And if you are considering split testing, check the maths on whether your traffic can support it before buying anything. Most stores cannot, and underpowered tests are worse than judgement because they carry false authority.

When outside help pays

When the obvious is exhausted. Everything above is doable in-house and plenty of stores should do exactly that before paying anyone. Once what remains needs development capacity you do not have, or genuine test volume to resolve, that is the work we sell. What that engagement actually looks like is a separate read, written for the moment after this one.