Cart Abandonment Above Benchmark: What to Do
Measured cart abandonment is defined as the share of a stated starting population that did not complete a purchase, and the starting population is the whole problem: carts created, checkouts begun and addresses entered produce materially different rates from identical shopper behaviour. So before fixing anything, establish whether your rate is genuinely above benchmark by writing down what your number counts and comparing only against a figure built the same way. Then split the loss across three gates: the cart view, the delivery and cost step, and the payment step. Each loses shoppers for unrelated reasons and each has a different fix, so a single blended number cannot tell you which problem you have. The largest loss usually sits immediately after the screen where the total cost first appears.
- Check the denominator first. Most abandonment gaps are definition differences.
- A cart used as a wishlist was never a lost sale. Exclude what you can identify.
- Split across three gates: cart view, delivery and cost, payment.
- The biggest loss sits right after the screen where the true total appears.
- Fix the gate, then recover the residue with information rather than a discount.
Cart abandonment above benchmark is the most frequently misdiagnosed number in eCommerce, because the comparison is usually invalid before the analysis even begins. Omniconvert has measured checkout behaviour across the CROBenchmark dataset of 7,000+ websites in 15+ industries, against 248+ audit criteria, over 13 years in eCommerce, and the first finding in nearly every abandonment investigation is that the store's rate and the benchmark it was compared against counted different populations. Omniconvert Explore is the CRO platform behind that record, with A/B and multivariate testing, on-site surveys, heatmaps and session recordings, averaging a 23.2% conversion uplift across 70,000+ experiments. Last updated: October 2026.
This article is the diagnosis. The fix list it hands off to is how to reduce cart abandonment. For the wider context, a good conversion rate sets the baseline, the companion diagnosis for traffic is your bounce rate is above benchmark, and the conversion-metric version is your conversion rate is below benchmark.
Check whether your cart abandonment rate is above benchmark at all
This step feels like pedantry and is routinely worth more than the rest of the investigation.
A rate measured from carts created includes every shopper who added an item to see the price, to save it for later, or to check whether delivery was available. A rate measured from checkouts started excludes most of those and counts only people who signalled an intention to buy. A rate measured from addresses entered excludes more again and describes a population that was very nearly committed.
The same store, the same week, the same shoppers will produce three quite different percentages depending on which of those you pick. Comparing your carts-created figure against a benchmark built on checkouts-started is not a performance finding, it is an arithmetic mismatch, and it generates urgency about a problem that may not exist.
There is a second, subtler version of the same issue. Platforms differ in when they create a cart record and whether they count a cart recovered later as abandoned. Shopify's analytics documentation is worth reading on exactly what its own abandonment report counts, because that definition and not the industry average is what your dashboard is showing you.
Why the published average is close to useless as a target
The Baymard Institute's checkout research is the most-cited source here and places average abandonment around seventy percent, with a well-documented list of stated reasons. It is genuinely useful work and it is routinely misused as a target.
The misuse is this. A substantial share of that seventy percent is people using a cart as a comparison tool or a shortlist, which is ordinary consumer behaviour on every store in every category. No change to your checkout converts a shopper who added an item to compare shipping costs across three sites. Setting a goal against the headline rate therefore commits you to moving a number whose largest component is immovable.
The alternative is to work at stage level. The share of shoppers lost between the delivery step and the payment step is a specific, attributable figure about people who had clearly decided to buy. It is smaller, it is movable, and improving it shows up in orders.
One further reason to prefer the stage number: it is comparable with itself over time, which the headline rate is not. A marketing campaign that brings more browsers raises your carts-created abandonment while changing nothing about checkout quality, and a team watching only the headline figure will spend a quarter investigating a campaign's success.
The three gates, and what each one is really failing on
Taking them in the order a shopper meets them.
Gate one: the cart view. Losses here are about doubt rather than cost. The shopper is re-examining the decision, and what helps is information: a clear delivery date, a visible return window, stock certainty, and nothing that reads as pressure. Losses at this gate are the least urgent of the three, because many of them were never buyers.
Gate two: delivery and cost. The largest loss on most stores, and the most fixable. This is usually the first screen where shipping, tax and sometimes duty appear, so a shopper who committed mentally to one price meets a different one. The remedy is almost never cheaper shipping. It is earlier disclosure, so that the commitment forms against the true total.
Gate three: payment. Losses here divide into friction, trust and failure. Friction is a long form or a forced account. Trust is an unfamiliar payment screen or a missing reassurance at the moment of entering card details. Failure is a genuine technical error, which is far more common than teams believe and invisible unless somebody looks for declined and errored attempts specifically.
| Gate | Usual cause of loss | What actually helps | Share of the loss |
|---|---|---|---|
| Cart view | Doubt, or never a buyer | Delivery date, returns, stock certainty | Large but mostly not winnable |
| Delivery and cost | The total appears for the first time | Disclose the cost earlier, not lower | The biggest winnable share |
| Payment: friction | Long forms, forced account creation | Guest checkout, fewer fields | Moderate and cheap to fix |
| Payment: trust | Unfamiliar screen at the worst moment | Recognisable methods, visible reassurance | Moderate, often underestimated |
| Payment: failure | Genuine declines and technical errors | Log errored attempts and read them | Small, and almost never measured |
| All three blended | Reported as one abandonment rate | Splitting it, which costs one report | The reason the wrong fix gets shipped |
The payment-failure row deserves a specific recommendation, because it is the one nobody checks: ask whoever owns your payment integration for the count of declined and errored attempts last month. On a meaningful number of stores that figure is large enough to be the cheapest win available, and it is not in any abandonment report.
What an operator should do this week
- Write down what your rate counts. Carts created, checkouts started or addresses entered. Then find a benchmark built the same way, or stop comparing.
- Split the loss across the three gates. Cart view, delivery and cost, payment. One report, and it usually relocates the whole investigation.
- Walk a real purchase on a phone. Note the exact screen where shipping, tax and any duty first become visible. The largest loss is normally just after it.
- Pull the declined and errored payment count. From whoever owns the integration. This is the one line of data that is never in the abandonment report and is sometimes the answer.
- Get your free benchmark score. CROBenchmark scores your storefront against the criteria set behind the CROBenchmark dataset, which shows whether checkout is genuinely your weakest area or a distraction from a larger one.
Once the failing gate is named, the fixes themselves are ranked in how to reduce cart abandonment, and the page-level examination is in audit it. Where your cart, order, payment and delivery-cost data sit in four systems and have to be assembled before the gate split is even possible, that assembly is what Nexus by Omniconvert addresses as an AI for eCommerce growth engine, with every proposed change staying something a person approves before it goes live.
FAQ: cart abandonment above benchmark
Is your cart abandonment rate really above benchmark?
Check the denominator before believing the comparison. A rate measured from carts created, from checkouts started, or from addresses entered will differ by a wide margin for exactly the same shopper behaviour, and published benchmarks rarely state which they used. Write down what your own number counts, then compare only against a figure built the same way. A great many abandonment problems turn out, on inspection, to be a definition difference rather than a performance difference.
What are the three gates in checkout?
The cart view, where a shopper decides whether to proceed at all; the delivery and cost step, where the total becomes real and where the largest single loss almost always sits; and the payment step, where friction, trust and technical failure decide the outcome. Each gate loses shoppers for unrelated reasons, so a single abandonment number averages three separate problems and cannot tell you which one you have.
Why is the delivery step the biggest loss?
Because it is usually the first screen where the true total appears. A shopper who has mentally committed to a price meets shipping, tax and sometimes duty at that moment, and the decision reopens. The fix is rarely cheaper shipping: it is showing the cost earlier, so the commitment is made against the real figure. Moving the information is cheap, and it moves the loss out of checkout and into a stage where it costs you far less.
Should you discount to recover abandoned carts?
Sparingly, and never as the first response. A recovery discount works, which is the problem: it teaches a measurable share of your customers to abandon deliberately, and that learned behaviour persists after the campaign ends. Fix the gate first, then recover the residue with a reminder that carries information rather than money, such as the delivery date or the stock position.
What is a realistic improvement to aim for?
A few points at the gate you fixed, not a transformation of the headline rate. Published averages place abandonment around seven in ten, and a large part of that is browsing behaviour that was never a lost sale. Aim at the stage-level number instead: the share lost between the delivery step and the payment step is specific, attributable and genuinely movable, and improving it shows up in orders rather than only in a dashboard.
The bottom line
Two questions, asked in order, settle almost every cart abandonment investigation. First, what does your number count? Carts created, checkouts started and addresses entered are three different populations, and comparing one against a benchmark built on another produces an urgent problem out of arithmetic. Second, which gate is losing them? The cart view, the delivery and cost step, or payment. Split it, because a blended rate averages three unrelated problems and the fix for each has nothing in common with the fix for the others. On most stores the answer is the middle gate, and the remedy is to show the true total earlier rather than to make it smaller. Then pull your declined-payment count, which nobody does and which is occasionally the cheapest order growth available to you. Change one thing at the gate that failed, and you will be able to tell whether it worked.