Cart abandonment is the metric everyone quotes and few diagnose properly. The headline figure — often quoted around seven in ten carts abandoned — lumps together genuine lost sales and shoppers who were never going to buy on that visit. Before you spend on recovery, it’s worth separating the two, because most of the recoverable revenue sits in a smaller, identifiable slice. AI helps you find that slice and intervene without becoming a nuisance.
Why shoppers actually abandon
Survey after survey lands on the same reasons, and almost all of them are about friction or trust rather than fickleness:
- Unexpected costs — shipping, taxes or fees that only appear at checkout. This is consistently the single biggest cause.
- Forced account creation before purchase.
- A long or confusing checkout with too many steps or fields.
- Payment friction — a missing preferred method, a declined card, a clunky mobile form.
- Comparison and research — many “abandoners” are mid-research and intend to return.
- Trust gaps — unclear returns, no reassurance on delivery, an unfamiliar brand.
The practical takeaway: a large share of abandonment is fixable with structural changes, and only a portion needs recovery messaging. Spend on the structure first.
Fix the checkout before you chase the cart
Recovery campaigns paper over a leaky checkout. The cheaper win is almost always reducing the friction that caused the abandonment in the first place.
Show the full cost early
If shipping shocks people at the final step, surface it sooner — a delivery estimate on the product page, a threshold message in the cart (“£8 from free delivery”). Predictable cost is the most effective abandonment reducer there is.
Strip the checkout to essentials
Offer guest checkout. Remove every field you don’t strictly need. Use address lookup, support digital wallets, and validate inline rather than after submission. On mobile especially, each removed field measurably helps — which is why this overlaps so heavily with closing the mobile conversion gap.
Build trust in the moment
Returns policy, delivery date, secure-payment cues and stock status, shown at the point of hesitation, do real work. This is the same reasoning behind a strong PDP optimisation checklist — reassurance belongs where the doubt lives.
Where AI earns its place
Once the structural basics are sound, AI improves both the targeting and the timing of recovery — and helps you avoid annoying the people you most want to keep.
Predict who is worth recovering
Not every abandoned cart deserves the same treatment. A model trained on session behaviour, basket composition and history can score the likelihood a cart is recoverable and the expected value of recovering it. High-intent, high-value carts justify a prompt, personal nudge; low-intent browsers don’t need an email at all. This stops you spending margin on discounts for people who’d have returned anyway.
Decide whether to discount — and how much
The lazy reflex is to fire a 10% code at every abandoner. That trains customers to abandon deliberately and erodes margin. A better approach predicts who actually needs an incentive versus who only needs a reminder or a reassurance about delivery. Reserve discounts for carts where the model says price is the likely blocker, and keep them modest.
Get the timing and channel right
AI can optimise send timing per shopper rather than firing a fixed “one hour later” email to everyone. It can also pick the channel — a real-time on-site nudge for a hesitating session, email for a research-mode shopper, and a restrained follow-up sequence rather than a barrage.
Resolve the doubt directly
If the abandonment signal points to a question — sizing, delivery, compatibility — an on-site assistant or a well-placed helpdesk automation flow can answer it in the moment, which often recovers the sale without any discount at all. Many abandoned carts are really unanswered questions.
A recovery sequence that doesn’t annoy
A restrained, well-targeted sequence outperforms an aggressive one. A pattern that works:
- In-session nudge. When exit intent or hesitation is detected, surface the specific reassurance the behaviour suggests — delivery date, free-returns badge, or a help prompt. No discount.
- First reminder (a few hours later). A clean email or push: the cart contents, current stock, and a one-click return to checkout. Still no discount.
- Second touch (around a day later). Add genuine value — reviews, a delivery guarantee, an answer to a likely objection.
- Incentive, only if warranted. For carts the model flags as price-sensitive and high-value, a modest, time-bound offer. Suppress this for everyone else.
Cap the sequence. Two or three touches is plenty; beyond that you’re harvesting unsubscribes.
Measuring it without fooling yourself
Recovery campaigns are notorious for claiming credit they didn’t earn — many “recovered” customers were always coming back.
- Use a holdout. Withhold recovery from a random slice of abandoners and compare completion rates. The difference is your true incremental recovery, not the headline number the tool reports.
- Track net revenue, after discount cost. A campaign that recovers carts only by giving away margin may be unprofitable.
- Watch unsubscribe and complaint rates as guardrails. Recovery that costs you the relationship isn’t worth it.
- Segment by the structural causes. If most abandonment traces to shipping cost, the fix is pricing strategy, not email volume.
For the wider funnel picture these numbers sit within, see eCommerce funnel analysis.
Pitfalls to avoid
- Discounting everyone, training deliberate abandonment and eroding margin.
- Treating abandonment as one problem when it’s several — research, friction, cost shock and trust each need a different response.
- Over-messaging. Three emails and two retargeting channels for a £20 basket is a way to lose the customer.
- No holdout, so you can’t tell recovery from coincidence.
- Ignoring mobile, where the structural friction is worst and the recovery upside is largest.
Reducing cart abandonment is mostly unglamorous: predictable costs, a short checkout, clear reassurance, and a restrained, well-targeted nudge for the carts that genuinely need one. AI sharpens the targeting and timing so you recover revenue without spending your margin or your goodwill. It fits within a broader conversion optimisation programme rather than standing alone.
If you’d like help separating recoverable abandonment from noise — and building a sequence that respects your customers — get in touch.