Most people who leave did not decide against you.
They were interrupted. The tab closed. A child needed something. The day resumed and did not hand the thread back. There was no judgement, no weighing up, no moment where they considered you and found you wanting. There was a gap, and nothing on the other side of it.
We almost never read it that way. Silence gets interpreted, and the interpretation is nearly always a verdict. She has thought about it and decided. He has gone quiet because he means something by it. The absence gets filled in with a reason, and the reason is usually about our own worth.
I have been reading research this year that says otherwise, from an unlikely place.
An unlikely place
Baymard Institute has spent years asking people why they abandoned an online purchase. Across fifty separate studies the average abandonment rate is seventy per cent. Seven in ten people who choose a thing, and add it, and get as far as the point of paying, then leave.
I went looking at it for practical reasons. I have been building a store this year and I wanted to know what to fix. What I found was a list I recognised.
Here it is, in order, once you set aside everyone who was only browsing.
Forty per cent leave because the cost changed. Twenty per cent because delivery was too slow. Nineteen per cent because they did not trust the site with their card. Eighteen per cent because they were made to create an account first. Seventeen per cent because it took too long. Twelve per cent because they could not see the total before committing.
Read that without the commerce. It is not a list about money. It is a list about surprise, safety and effort.
Why a checkout tells the truth
A shopping cart is a low stakes place and I am not going to pretend otherwise. Nobody is harmed by an abandoned cart.
That is exactly what makes it useful.
The stakes are low enough that people answer honestly. Ask someone why they stopped going to therapy, or why they let a friendship lapse, or why they did not go back to the specialist, and the answer comes back shaped by loyalty and self protection and not wanting to be the difficult one. Ask why they closed the tab and they simply tell you. There was a fee I did not expect. It wanted my details before it would show me the price. It took too long.
The checkout is a controlled experiment in the conditions under which a person quietly withdraws. It runs millions of times a day and it records everything.
Surprise
Forty per cent. The largest single cause, well ahead of anything else, is a cost that arrived later than expected.
Note what this is not. It is not that the price was too high. Someone who thinks a thing costs too much never adds it in the first place. This is a person who said yes at one number and was then shown a different number.
The injury is not the amount. It is the revision. Something was one shape when they agreed to it and another shape once they had committed, and the correct response to that is to stop. It is not an overreaction. It is calibration. If the terms moved once, quietly, at the point where leaving got harder, the terms can move again.
You see the same withdrawal anywhere the real cost of a thing arrives after the agreement rather than before it. The person who stops answering is not being dramatic. They are declining to find out what else changes.
The commerce fix is embarrassingly simple. Show the total early. Put shipping on the product page. If there is a threshold for free delivery, show a live bar saying how far away it is. Nobody abandons over a number they already knew.
Safety
Nineteen per cent left in a three month window because they did not trust the site with their card.
What actually moves that number is the interesting part. Baymard has surveyed trust seals seven times and the Norton badge has ranked first every round. DigiCert owns Norton, Thawte and GeoTrust and issues identical certifications through all three. The Norton seal was still nearly twenty times more likely to give shoppers the best sense of trust.
Identical security. Twenty times the trust. Nobody is evaluating cryptography. They are recognising a name.
Placement decides whether the badge does anything at all. Baymard found a trust signal beside the card field outperforms the same badge in the footer. Not because the footer is hidden. Because reassurance offered before the hesitation is information, and reassurance offered during the hesitation is safety. The same words at the wrong moment do nothing.
Most of us know this and forget it constantly. Being told in advance that a person is safe is not the same as being met at the point where you falter. The first is a claim. The second is evidence.
Effort
Eighteen per cent left because they were required to create an account. Seventeen because the process was too long.
Roughly a third of all departures over friction alone. Not cost. Not trust. Just too many steps between wanting the thing and having it.
This is the one that gets moralised. The language around it is impatience, short attention spans, nobody can be bothered anymore. That reading is wrong and it is worth saying why. A person with limited capacity spends it on what matters most. When a process demands more than the thing is worth, leaving is not laziness. It is triage.
The fix is to remove steps rather than add persuasion. Let people buy without registering and offer the account afterwards, when they have a reason to want it. Turn on the payment buttons that skip the forms entirely.
The pattern underneath is the one I keep running into. Almost every real improvement is a removal. The instinct is to add. Add reassurance, add an incentive, add a follow up. But the person did not leave because something was missing. They left because something was in the way.
Safety comes before contact
There is a sequence in the work I teach that I did not expect to find sitting inside checkout data.
The HOPE Method moves through four phases and the order is the entire point. Honour: meet your body and its history without shame, naming as neutral fact. Orient: find safety first, your window of tolerance and a grounding anchor, before you go any further. Presence: come back into contact, gently, by your consent, only ever as far as feels right. Embody: from tolerance to tenderness, meeting whatever arises, including no.
Now set the three findings against it.
Surprise is a failure of Honour. The thing was not named plainly at the start. What was withheld gets discovered later, at the point where leaving has already become expensive, and the discovery does more damage than the fact ever would have.
Safety is Orient, almost exactly. Baymard’s finding that a trust signal beside the card field beats the identical badge in the footer is the window of tolerance argument in miniature. Safety is not established once in advance and then banked. It is established at the threshold, in the moment of the reach.
Effort is Presence. Only ever as far as feels right. A process that demands registration before it will show you a price has skipped consent and gone straight to the ask, and people withdraw from that correctly.
Embody is the one the data cannot reach. Meeting whatever arises, including no. No store does that. But it is why I do not think a well made reminder is manipulation. It reopens a door without standing in it.
Back to the gap
Which brings me back to where I started, and to the one intervention on this list that works by adding rather than removing.
A reminder, sent a few hours after someone leaves, recovers a meaningful share of them. Across Klaviyo’s benchmark data the average abandoned cart sequence sees a fifty per cent open rate and a three per cent order rate. The best performing brands more than double that.
Sit with what that means. A significant proportion of the seventy per cent were not lost. They were interrupted, and the only thing standing between them and the thing they wanted was that nobody picked the thread back up.
The reminder does not persuade. It removes the requirement that they remember. That is a real service, and it is the piece I keep taking out of this and applying somewhere else entirely. Most people are not withholding. They are at capacity.
There is a discipline to it. Do not lead with an incentive. If the first reminder always carries a discount, you teach people that leaving is how you get a better price, and you have converted an interruption into a strategy. Remind first. Sweeten only if the reminder alone does not land.
What I took from it
I set out to fix a store and found a fairly precise description of the conditions under which a person stops.
Costs that arrive after the agreement. Reassurance offered at the wrong moment. More steps than the thing is worth.
And, more common than all three combined, no decision at all. Just an interruption that nobody followed up.
Seventy per cent leave. Almost none of them leave over price. And most of them were not leaving. They were only gone.