Cart recovery is the most over-discounted flow in eCommerce. Most of the revenue is recoverable without a code at all — if the timing, the order of the messages and the definition of “recovered” are right. Here is the structure we build, and the mistakes that flatten it.
The Dazzle Commerce Team
eCommerce Management & Marketing
Updated
5 September 2026
Reading time
7 min read
Most abandoned carts are not a pricing objection. They are an interruption.
Almost every store has an abandoned cart email. Far fewer have an abandoned cart flow, and the difference between the two is usually worth several per cent of revenue.
The single email version assumes the shopper decided not to buy. Most did not decide anything. They were interrupted, they wanted to check the delivery date, they were on a train, or they wanted to think about it for a day. Those are four different problems, and one message cannot address them all.
KEY TAKEAWAY
The first job of a cart flow is not persuasion. It is putting the shopper back where they were, with the question that stopped them already answered.
The twelve to eighteen per cent range in the title is the band we see across the programmes we run, and it needs a definition attached or it means nothing. We count it as: revenue from orders placed by shoppers who abandoned a cart and then returned through a flow message, within the attribution window, minus the orders those shoppers would have placed anyway.
That last clause is where most reported figures fall apart. A shopper who was always coming back tomorrow will click your email on the way, and a naive report will credit the whole order to the flow. Any number that has not been tested against a holdout is a ceiling, not a result.
of abandoned value recovered
The band we see once holdouts are accounted for
in the core sequence
A fourth helps rarely; a fifth almost never
in message one
Discounting the interrupted is paying for orders you had
Figures reflect Dazzle Commerce engagements and depend heavily on category, price point and how recovery is defined. They are not industry benchmarks.
Three messages, each with a different job. The timing matters more than the copy.
Timing
Message
The job it does
45–90 minutes
Cart contents, delivery date, one link back
Catches the interruption while the intent is still live
20–24 hours
The objection answered: returns, sizing, stock, support
Removes the reason they went to think about it
48–72 hours
Social proof, and only now an incentive if the margin allows
Converts genuine hesitation without discounting the first two
Optional, 5–7 days
Back-in-stock or price-change trigger only
Event-driven, not scheduled — and only if something changed
Windows are starting points. High-consideration and higher-priced categories stretch the whole sequence; impulse categories compress it.
It is the single most common structural mistake, and it costs money in three separate ways.
1
You discount the people who were coming back anyway. In most carts they are the largest group in the flow.
2
You train the behaviour. Shoppers learn quickly that abandoning produces a code, and a measurable slice of them start doing it deliberately.
3
You spend your only lever first. If the third message has nothing new to say, the sequence has no reason to exist.
They get collapsed into one programme constantly, and they are not the same event. Someone who reached payment and stopped has a specific, usually solvable objection. Someone who looked at three products has not decided anything yet.
A shopper who stops at payment has a question. A shopper who stops at the grid has not finished deciding. The same email cannot serve both.
1
No suppression against the rest of the programme. A cart message landing in the same hour as a campaign send halves both.
2
A generic subject line. The item name in the subject reliably outperforms the word “cart”, because it tells the reader what this is about.
3
Sending to a stale cart. If the item is out of stock or the price changed, the message needs to say so or you have created a support ticket.
4
Ignoring the SMS and push versions. The 45-minute message is often better as a push notification than an email, where you have consent for it.
5
Never rebuilding it. Cart flows decay. The objection your copy answers today is not the one shoppers had eighteen months ago.
EXPERT VIEW
“The best cart flow improvement we made last year was not an email at all. It was putting the delivery date on the product page, which removed the objection before the cart existed.”
Dazzle Commerce · Retention team
If the flow, the suppression rules and the holdout testing are more than the team has time for, that is the shape of our retention work.
Three rules keep the number defensible in front of a finance team:
THE TEST THAT SETTLES IT
If your cart flow reports a recovery rate but has never run a holdout, you do not have a recovery rate. You have an upper bound with a discount attached.
Three services that turn a cart sequence into a lifecycle programme.
Lifecycle flows, segmentation, holdout testing and the reporting that keeps recovery honest.
Removing the objection before the cart exists, which beats recovering it afterwards.
The wider programme the cart flow has to be suppressed against and measured inside.
WRITTEN BY
eCommerce Management & Marketing
Dazzle Commerce builds lifecycle and retention programmes alongside acquisition, which is the only way to tell whether a flow recovered revenue or simply took credit for it. The structure below is the one we start every cart programme from.
More from the retention and conversion side of the practice.
Most cart abandonment is caused earlier, on the page that failed to answer the question.
The same double-counting problem, one layer up in the reporting.
Where lifecycle coverage sits in the wider order of operations.
We will review your current sequence against this structure and show you where the recovery is being left behind or double counted.
