Abandoned Cart: What we found when we looked at the checkout process step by step

We placed a real order, using a card, in a client's shop to see why people abandon their carts at the end. On a phone, using mobile data, not office Wi-Fi.

We reached the payment stage in seven steps and with 14 fields to fill in. At the fourth step, the keyboard covered the continue button, and the page wouldn't scroll. It wasn't a marketing problem. It was a button you couldn't reach.

What we found, in the order it bothered us

Delivery cost only appeared at the very last step. The person would enter their address, card details, then see an extra 25 lei. Some would leave there.

Mandatory account. There was no option to order without an account, and the registration form required email confirmation before completion.

Useless fields: company name, company registration code, "how did you hear about us", secondary address. All optional in theory, all displayed with the same visual weight.

Error messages that cleared what you had filled in. Here we lost ten minutes and had a card declined, just to understand that the problem was the phone's format.

What we fixed first

Shipping cost displayed in the basket, with the free delivery threshold mentioned next to the total. Order without an account, with the account creation suggested after payment. The form reduced to what is strictly necessary for delivery and invoice.

Then, technically: the button that remains accessible when the keyboard pops up, validation that doesn't clear fields, payment with phone wallets, so card details no longer need to be typed.

The result was a clear reduction in abandonments at the final step. I'm not giving the percentage because in the same month the client also changed courier, so I can't separate the effects. That's how it was, there's no point pretending otherwise.

Recovery emails

Three, not seven. The first at one hour, with the product and a direct link to the rebuilt basket. The second at one day, which answers a specific objection — returns, warranty, delivery time. The third at three days.

We don't put the discount in the first email. If you do, you teach customers to abandon their basket to get a discount. We saw this happening at a shop that sent 10% immediately.

SMS works for some niches and irritates others. We test, we don't assume: how much it helps depends on the niche.

What isn't worth it

Complicated automations with many branches, for shops with few orders. You build them for a week and you don't have data to know if branch A is better than B. Better to have three simple, well-written emails.

And something we repeat: if the product is more expensive than competitors and delivery is slower, a fixed checkout won't solve that. We've had cases where the honest recommendation was to look at delivery costs, not the website.

If you like, we can place a test order in your shop and send you a list of what we found.

What we put in the table after the first month

We kept a list of what we changed and what happened, so we wouldn't rely on memory for reporting. We're not putting percentages here, because they are the client's figures and we don't publish them.

The row with the email is usually the disappointing one. It helps, but it doesn't save a poorly built checkout.

What we check on the phone, separately

Almost all serious problems were found on the phone: keyboard covering the button, phone field opening a letter keyboard, error messages displayed at the top, off-screen. We open the checkout on two or three real phones, not just in the browser simulator.

The most frequent thing we fix isn't a setting, but an error message the customer doesn't see because it's off-screen.

The order we work in, if the budget is small

First, we remove obstacles (unnecessary fields, mandatory account, hidden costs). Then we clarify what happens after the order. Only at the end do we add automated emails. The reverse makes no sense, although most requests we receive are precisely for emails.

What's not worth it for low volume

A/B tests. If you have a few orders a day, you can't draw a statistical conclusion in a reasonable time. We make changes based on logic and direct observation, and we clearly tell the client that we don't have statistical proof, but judgment.

If you have abandoned baskets and don't know where you're losing people, send us a link and we'll tell you what we'd fix in the first week.