Conversion · E-commerce

What do you do with people who leave the checkout?

At Genomed most started checkouts never got paid. The checkout had collected an email and a phone number since the 2023 rebuild, but nobody used them. In July 2024 I built a process with the call centre to call every abandoned cart back. Checkout payments per visitor rose by about a third, roughly $55–60K a month more through the web channel.

Genomed · 2024 ·
My role
Senior Product Manager, owner of the web channel
Company
Genomed, the largest genetic testing lab in the CIS, selling tests online
Context
A year after the checkout rebuild. No paid traffic, no ad budget.
The constraint
Tests cost around $200 each, and most people want to talk to someone before paying for one.

If you only read one thing

The checkout already collected an email and a phone number before payment, but an abandoned cart was simply lost. I built a process with the call centre to work every one of them as a lead. I forecast +10–15% conversion. It came in at +22% in the first month and +35% over the quarter: about $55–60K a month more through the web channel.

The situation

By mid-2024 the biggest loss was everyone who started checkout and stopped. Genomed sells complex genetic tests at around $200 each, and many people hesitate at the payment step: they have a question, they want to check with a doctor, they get distracted. Since the 2023 rebuild the checkout asked for an email and a phone number before payment, so most of them had left a contact. Nobody used it: an abandoned cart just sat in the database.

Meanwhile the call centre was already good at exactly this kind of sale. A conversation is how most of these tests get bought.

The fork

Option A

Another round on the form

Keep polishing the checkout UX and hope fewer people leave. Safe, familiar, and with diminishing returns after the 2023 rebuild.

Option B · chosen

Treat an abandoned cart as a lead

Most started checkouts already left a contact. Work every one of them: messages, and above all a call from the call centre.

For a $200 medical test, a person who added it to the cart and left is the warmest lead the company has. The form could not answer their question; a person on the phone could. My forecast at the time was +10–15% conversion.

What shipped

Nothing changed on the form. I went to the call centre and we built a callback process together: every abandoned cart, with its email and phone number, became a queue for the operators, and people also got a message. We kept tuning it after launch.

What happened

~$55–60Ka month more through the web channel
+35%checkout payments per visitor, three months after vs three before
10–15% → 22–35%forecast vs actual

Fewer visitors, more money index, January 2024 = 100 · revenue through the site's checkout vs new visitors · 2024

Checkout revenue against new visitors, 2024, indexed to January = 100 Through 2024 new visitors fell by about a fifth while checkout revenue rose by more than a third. The callbacks launched in July. 60 80 100 120 140 launch Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Revenue138 Visitors82
Same data as a table
MonthPaid ordersRevenue, $New visitors
Jan 20241,049$212K184K
Feb 20241,205$247K179K
Mar 20241,140$239K181K
Apr 20241,110$240K179K
May 20241,192$260K165K
Jun 20241,130$234K153K
Jul 20241,432$275K160K
Aug 20241,464$285K151K
Sep 20241,253$263K135K
Oct 20241,450$303K152K
Nov 20241,396$293K150K
From my monthly tracking at Genomed. Over 2024 new visitors fell by about a fifth, while revenue through the checkout grew by more than a third. The gap widens from July, when the callbacks launched.

How the money is counted: had conversion stayed at June's level, July to September would have brought about 860 fewer paid orders. At an average order of about $190–215, that is $165–180K over the quarter, or $55–60K a month at 2024 exchange rates. Traffic over the same months was flat to slightly down, so none of it came from more visitors.

Some of this growth was cannibalisation, and that was the point. Customers who used to pay in cash at a medical office, and even call-centre operators with their own payment tools, started choosing the site's checkout. For the web channel I owned, every one of those payments was a win: faster for the customer, cheaper to process, and visible in the analytics.

The calls did more of the work than the emails, and that was the design: the change was not a message template, it was a process that put everyone who had nearly bought in front of the people best at closing these sales.

All money on this page is in US dollars, converted from roubles. Monthly figures in the chart and table use one fixed rate, 89 ₽ per dollar (Bank of Russia average for January 2024), so currency swings don’t distort the trend. The monthly totals in the text use the rates of the months the money came in.

The takeaway

An abandoned checkout is a lead with a name and a phone number. Most companies already collect them and never call. When your best sales channel is a conversation, feed it.

All cases