Precovery: Winning the sale back before you lose it

Every retailer has a recovery programme. Abandoned basket emails. Retargeting ads. Exit popups that fire as the cursor drifts towards the close button.

It’s a whole industry, and it’s built on an admission: the sale is already slipping away, so let’s chase it. Here’s the problem with chasing.

By the time you’re recovering a sale, the shopper has already decided to leave. And the moment someone decides to leave is the moment you have the least influence over what they do next.

You’re not persuading any more. You’re pleading. Begging.

There is a better moment. It’s a few seconds earlier, while the shopper is still on the page, still hovering, still deciding. Acting then has a name.

We call it precovery (yes, we’re making up words and you can’t stop us).

Recovery is a confession

Think about what a recovery tactic actually says.

• An abandoned basket email says, “We lost you, please come back.”
• A retargeting ad says, “We lost you, here’s a reminder.”
• An exit pop-up says, “We can see you leaving. Here’s 10% to stop.”

Every one of them is triggered by failure. The shopper has to be on their way out before the tactic even fires.

And most of them are. On average, around 70% of online baskets are abandoned. That’s the headline number the whole recovery industry exists to claw back.

The trouble is that clawing back is hard, slow and expensive, because you’re starting from a standing position of weakness.

The leverage problem

Timing is everything in a sale, and recovery has the timing backwards.

When a shopper is mid-decision, you can still tip them. A nudge, a reassurance, a reason to act now.

Once they’ve left, that window is shut. You’re reduced to interruption: an email they might open tomorrow, an ad that follows them around the internet, a discount big enough to drag them back.

The later you act, the more you have to give away to get the same result.

Which leads to the second hidden cost. Recovery doesn’t know who actually needed saving.

It treats every leaver the same, so it hands incentives to shoppers who’d have come back on their own.

That isn’t a small leak. Nielsen estimates that around 84% of price promotions are unprofitable, largely because most promoted sales would have happened anyway. Recovery, done bluntly, pours straight into that statistic.

What precovery is

Precovery flips the tense. Instead of waiting for a shopper to leave and then chasing them, you predict they’re about to leave without buying and act while they’re still there and still persuadable.

It only became possible recently, because it depends on prediction. You need to read a shopper’s behaviour in real time and know, before they’ve gone, that this session is heading for an exit.

That’s a purchase-intent problem, and it’s exactly what modern intent models are built to solve. They watch the recent signals, dwell time, hesitation at checkout, the small tells of a shopper losing momentum, and score the likelihood of a leave as it builds.

Same goal as recovery. Opposite moment.

The bit recovery can never do

Here’s where precovery earns its keep, and it isn’t only the timing.

Because precovery is predictive, it doesn’t just spot who’s about to leave. It can be said that, apart from who was going to buy anyway. And that distinction is where the margin lives.

A recovery tactic shows its offer to everyone heading for the exit, including the shopper who would have wandered back and bought at full price.

Precovery doesn’t. It steps in for the genuine wobble and leaves the certain buyer alone. So you’re not just acting earlier, you’re acting only where it matters. The shoppers who’d convert regardless never see a discount, and the margin stays where it belongs.

That’s the part the spreadsheet loves. Fewer discounts given, not more, and every one of them aimed at a sale you’d otherwise have lost.

What it looks like in practice

This is what our new intent-driven app was built to do, and the mechanics are simpler than they sound.

It watches each shopper's behaviour as it happens, reading more than 40 signals from the session: how they move through the site, how long they dwell, where they hesitate, the small tells of someone losing momentum.

Because intent is about right now, it leans on recent behaviour rather than what they did weeks ago.

From those signals, it scores, moment to moment, how likely this shopper is to buy, and how likely they are to leave without doing so. When the score says a shopper is wavering and drifting towards the exit, that's the cue to act, while they're still on the page.

Then it does two things at once. It steps in for the shopper who needs a reason to stay, and it leaves the one who was always going to buy completely alone.

That's the whole loop. Read the signals, predict the leave, act on the few who need it; in the second, it counts. And because every nudge is a deliberate prediction, you can hold a group back and measure the lift rather than assume it.

It isn’t a real word, and that’s the point

You won't find precovery in a dictionary.

Recovery has a word because it's the old, reactive way of doing things, and it's had decades to settle in. Precovery doesn't have one yet, because until prediction caught up, nobody could actually do it. The missing word is the tell. It marks the gap between waiting for a sale to fail and stopping it from failing in the first place.

It also decodes itself on first hearing. Pre, plus recovery. You didn't need this paragraph to work out what it means, which is the test any new term has to pass.

And the prefix is doing real work. Pre is the language of acting ahead. Prediction. Preview. Prevention. Precovery belongs with them: the same forward instinct, pointed at the sale you're about to lose.

Where to start

You don’t have to rip out your recovery programme. Abandoned-basket emails and retargeting still have a job, mopping up shoppers who genuinely abandon their carts. But they shouldn’t be your first line. Your first line should be a few seconds before, while you can still change the outcome cheaply.

So the question worth asking isn’t “how do we recover more carts?” It’s “how many of those carts did we never need to lose?”

Recovery waits until they’ve gone. Precovery acts before they do.