Two stores can report the same cart abandonment rate and need almost opposite work. In one, most of those carts belong to people comparing prices across four browser tabs who were never going to buy that evening. In the other, they belong to buyers who reached the payment step and met a delivery charge nobody had mentioned earlier. Both figures sit close to the published averages. Only one of them describes a problem worth a development budget.
Most cart abandonment solutions get bought before anyone has established which of those two stores they are running, which is why recovery emails, exit popups and discount codes often produce a short lift and then settle back. The five causes below account for most abandoned carts in established stores, and each one leaves a different signature in data the business already holds.
What Actually Counts as an Abandoned Cart
Salesforce states the event plainly: shopping cart abandonment occurs when a potential customer adds items to their cart but leaves the store before completing the purchase. The simplicity is deceptive, because the rate that follows depends entirely on what the platform counts as a cart, a session and a purchase.
BigCommerce sets out the arithmetic behind the figure: divide completed transactions by initiated sales, subtract the result from one, then multiply by 100. The same source separates browse abandonment, where a visitor views products and leaves without adding anything, from cart abandonment proper, where items were added and no order followed. Reporting that merges the two inflates the apparent problem and aims remediation at the wrong stage of the journey.
Platform reporting adds its own boundaries. Shopify’s behaviour reports count sessions with cart additions as sessions where a customer added a product to a cart, and treat a session as having reached checkout when there was user input during it. These are session-level measures, not customer-level ones, so one person who builds a cart on a handset at lunchtime and pays from a desktop after work is recorded once as an abandonment and once as a sale.
A store with a long consideration cycle, heavy mobile research traffic, or a catalogue customers treat as a wishlist may therefore sit well above an industry average with nothing wrong.
The Five Most Common Causes of Cart Abandonment
Survey data is consistent enough across sources to serve as a starting hypothesis list, provided it is treated as one. Shopify’s summary of the research puts unexpected extra costs at the top, at roughly 47% of shoppers, followed by account creation requirements, slow delivery, a long checkout at around 18%, and an inability to work out the total cost at around 17%. That order will not hold in every store, which is the entire reason diagnosis is necessary.
1. Cost That Arrives After the Decision Has Been Made
The most expensive moment in most checkouts is the one where a total changes. A shopper who has mentally committed to a price sees shipping, duties or a handling fee added, and reassesses the purchase from the beginning. The size of the increase often matters less than its timing: a delivery charge shown on the product page rarely provokes the reaction the same charge produces two steps from payment.
The pattern is common where shipping is calculated by weight, zone or live carrier rate, because an accurate figure cannot be produced until an address exists, and in cross-border selling, where duties are left to surface at the last possible moment. The consequence reaches past the lost order: shoppers who leave at a cost surprise often do not return through the same channel, so a store leaning on paid traffic sees its effective cost per order rise across every campaign.
It is worth being honest about when this is not the problem. If most abandonment happens before the shipping step is reached at all, cost transparency is not where the money is, however tempting the survey data makes it look.
2. A Checkout That Asks for More Than the Order Requires
Checkout friction covers two related failures: requesting information the order does not need, and requesting it in a way that is awkward to supply. Shopify’s research attributes close to a quarter of abandonment to sites asking shoppers to create an account, which makes it one of the few causes a store can remove entirely rather than merely reduce.
Account creation deserves separate attention because the exchange is asymmetric. The retailer gains a record it could largely reconstruct from the order itself, while the shopper takes on a password, a confirmation email and a relationship they did not request. Guest checkout, with an optional account offer after payment, usually recovers most of the data without the objection.
Field-level friction is less visible and often more damaging on mobile, where every extra input means a keyboard change and another chance to mistype. Address line two, company name and a phone number are frequently mandatory by habit rather than operational need, and the cumulative effect is a form noticeably slower than a competitor’s. Friction is rarely the sole cause in an otherwise healthy funnel, but it compounds the others: a shopper already uncertain about delivery abandons a slow form faster than one who is not.
3. Doubt About the Seller, the Product or the Return Path
Trust failures are the hardest cause to see in analytics, because the visitor leaves without doing anything unusual. Shopify’s summary reports that roughly 19% of shoppers have abandoned a cart because they did not trust the site with their card details, and Salesforce lists security concerns and missing reassurance among its recurring causes.
For established retailers the doubt is rarely about fraud in the crude sense. It is about consequences: whether an item can be returned without argument, whether the size will be right, whether anyone answers if the parcel arrives damaged. A returns policy sitting three clicks from the cart is functionally absent at the moment it is needed.
This cause concentrates in identifiable segments. New visitors, first-time buyers, higher order values and categories where fit or authenticity is uncertain carry most of it, so a store with strong repeat purchase behaviour can have a serious trust problem among new customers and never see it in a blended figure. The remedy is usually informational rather than technical, which makes it cheap to test and easy to under-prioritise: return terms, a delivery promise and payment reassurance placed where the hesitation occurs will often outperform a redesign of the same page.
4. Delivery Terms That Do Not Match the Purchase Occasion
Delivery is where a purchase decision meets a real-world constraint. Slow delivery sits behind close to a quarter of abandonment in the research, and even that understates it, because slow is relative to the occasion rather than to a benchmark. Three days is unremarkable for a replacement filter and useless for a birthday on Saturday.
Stores tend to discover this cause seasonally. Abandonment climbs in the weeks before a gifting peak while nothing on the site has changed, because the estimate that was acceptable in October stops being acceptable in December. Where the estimate is generic rather than calculated, shoppers tend to assume the worst case.
The remedy is not always faster fulfilment, which may be commercially impossible. A specific, dated estimate shown early, an expedited option at a price the shopper can choose, and a collection route where physical locations exist all answer the same objection without changing the warehouse. When delivery is genuinely the cause, abandonment usually concentrates at the shipping step and in the geographies furthest from the warehouse rather than spreading evenly across the funnel.
5. Purchase Intent That Was Never There
Salesforce includes browsing and saving for later among its listed causes, and Shopify notes that some shoppers use an online cart as a shopping list before buying in a physical store. Neither behaviour is a defect. The cart has become a comparison tool and a way of holding a price, and a share of every store’s abandonment is simply that being recorded.
Treating it as failure leads to real waste. Discounting a segment that intended to buy anyway transfers margin for nothing, and pressing researchers with urgency messaging can damage a relationship that would have converted a week later at full price.
Separating this group needs no specialist tooling. Carts never revisited, sessions with no checkout input, long gaps between addition and return, and comparison patterns across sessions all point to research rather than blocked intent. The share is worth quantifying once, because it sets a realistic floor for everything else. The useful question then is not how to convert these visitors immediately, but whether returning is easy: a cart that persists, a price that holds, a reminder that informs rather than pressures.
How to Diagnose Which Cause Is Yours
Sequence matters, because each step narrows what the next has to explain. Behavioural research run before the tracking is validated tends to produce vivid findings about a funnel that was never measured correctly.
- Confirm the measurement before trusting it. Reconcile analytics orders and revenue against the platform’s own reporting for the same period. A gap of any size makes every percentage that follows negotiable, and consent banners, unfiltered internal traffic and duplicated events are the usual causes.
- Separate browse abandonment from cart abandonment. Establish how many sessions reach a cart addition at all. If the loss is concentrated before that point, the problem sits on product and category pages, and no amount of checkout work will reach it.
- Locate the exact step where people stop. Cart to checkout entry, contact details to address, address to shipping selection, shipping to payment. Each transition maps to a different set of causes, and the largest drop rarely sits where the team assumed.
- Split the number by segment before interpreting it. A blended rate is an average of several different problems and usually describes none of them accurately.
- Watch what the numbers cannot explain. Microsoft’s Clarity team puts the limitation directly: funnels can show where users drop off, but they do not explain why. Repeated clicks on a submit button, inputs corrected several times, and sessions that stall at one step before exiting are the observable evidence that turns a drop-off into a cause.
- Ask the people who left. A short exit survey at the cart, or one question added to a recovery email, often resolves an ambiguity weeks of analysis cannot. Self-reported answers are imperfect, but they are the only source that states the reason directly.
- Size each candidate cause in money. Convert every hypothesis into recoverable revenue using the affected segment, its order value and a conservative recovery assumption. A cause affecting 4% of sessions at a high order value may outrank one affecting 20% at the lowest.
Step four carries more weight than its length suggests. The splits that most often change the conclusion are:
- Device. Mobile and desktop behave differently enough that a single figure often hides the entire problem, particularly where forms and payment methods differ between them.
- New versus returning. Trust and delivery objections concentrate among people who have not bought before, so a store with loyal repeat customers can look healthy in aggregate while losing most first orders.
- Traffic source. Visitors from broad prospecting campaigns arrive with weaker intent than those from branded search, and mixing them makes the store look worse than it is.
- Market and shipping zone. Cost and delivery causes are often geographic, appearing only where duties apply or where the warehouse is far away.
- Order value band. Higher-value carts attract more deliberation, so their abandonment may be a normal consideration cycle rather than friction.
What Each Cause Looks Like in Your Data
The table below pairs each cause with the evidence that tends to accompany it. No row is conclusive alone, and two causes often operate together, but a location in the funnel plus a matching behavioural signal is usually enough to justify a first fix.
| Likely cause | Where abandonment concentrates | Behavioural signal | Cheapest confirming test |
|---|---|---|---|
| Late cost reveal | The step where shipping or duties are first calculated | Recordings show a pause, a scroll back to the totals, then an exit | Show an estimated delivery cost earlier and compare the same step |
| Checkout friction | Contact and address entry, heavier on mobile | Repeated corrections, stalled sessions, repeated clicks on the continue button | Make optional fields optional and offer guest checkout |
| Trust and returns doubt | Spread across the cart and early checkout, mostly new visitors | Visits to returns, delivery or policy pages immediately before exit | Place return and payment reassurance at the cart, measured on new visitors only |
| Delivery mismatch | The shipping method selection, and distant markets | Shipping options viewed repeatedly without one being chosen | Publish a dated estimate and add one faster paid option |
| Research intent | Even across steps, with many carts never revisited | No checkout input at all, repeat product comparison across sessions | Quantify the segment, then test an informative reminder without a discount |
Reading it in the other direction is equally useful. Where recordings show no hesitation, no repeated corrections and no policy-page visits before exit, the store is probably looking at a research population rather than a broken journey, and cart abandonment solutions aimed at friction will have little to work with.
Key takeaway: the abandonment rate is a symptom shared by five unrelated diseases. Until the store knows which one it has, every remedy it buys is being prescribed by a supplier rather than by evidence.
Matching Cart Abandonment Solutions to the Cause You Found
Once the cause is identified, the choice among the available cart abandonment solutions becomes narrow. Choose cost transparency work when the drop concentrates at the shipping calculation, accepting that this may mean rebuilding how rates are estimated rather than adding a banner. Choose checkout simplification when the loss sits in form completion and recordings show hesitation. Choose reassurance content when new visitors leave after reading policy pages, and delivery changes when the shipping selection is where sessions end.
Recovery messaging is the one remedy that helps in nearly every scenario, but its design should still follow the diagnosis. Omnisend recommends staggering incentives across a sequence rather than leading with a discount, and segmenting the messaging by customer history, since a first-time buyer and a repeat customer who spends heavily are not abandoning for the same reason. Where research intent dominates, a reminder that carries delivery information and return terms may outperform one that carries a discount code, and it costs nothing in margin.
The delivery model deserves as much thought as the fix. An off-the-shelf application is often sufficient when the cause is a single, well-understood friction point the platform supports natively, and an internal developer is the right choice when the fix is small and unlikely to recur. External support earns its cost when the causes are entangled, when the fix touches shipping logic or payment configuration, or when nobody internally has capacity to run the analysis, implement the change and validate the result.
Before committing to any of these, a short list of questions tends to expose whether the plan rests on evidence:
- Which funnel step does this change affect, and what share of abandonment currently occurs there?
- Which segment is expected to respond, and how large is it in orders rather than in sessions?
- What behavioural evidence supports this cause, as opposed to a survey finding that it is common generally?
- How will the result be validated at the store’s traffic volume, and how long will that take?
- What is the margin cost if the change works, and is that cost acceptable at the volume expected?
A store that cannot answer the second and fourth questions is not ready to buy anything yet. That is not a delay, it is the cheapest stage of the work, and a structured ecommerce CRO audit exists largely to produce those answers before money is committed to a fix.
Where Cart Abandonment Diagnosis Commonly Goes Wrong
The most frequent error is importing a survey ranking as a conclusion. Extra costs top most published lists, so extra costs get addressed first, regardless of whether the store’s own abandonment concentrates anywhere near the shipping step. The work is not wasted in an absolute sense, but it is funded from a budget that had one clear priority available and chose a different one.
A second error is treating the rate as a target. Pushed hard enough, abandonment falls if cart additions are discouraged, which improves the ratio and reduces revenue at the same time. Instruments stop being useful once they become objectives.
Third, stores frequently stack cart abandonment solutions on one another without measuring between them. A recovery sequence, an exit popup, a countdown timer and a free shipping threshold launched in the same month produce a combined result nobody can attribute, so when the discount starts costing real margin there is no basis for deciding what to keep.
Fourth, the diagnosis often stops at the cart when the cause sits earlier. Weak product information or unclear variant selection can send visitors into the cart already uncertain, where a small objection ends the session. Our analysis of why product pages receive traffic but fail to convert covers that stage in more depth.
Finally, some teams stop measuring once a fix ships. A shipping table changes, a payment provider updates its interface, a theme update alters a form, and a cause that was resolved returns without announcement. Internal expert input required: add a verified example from a WD Market engagement where a resolved checkout or cart issue reappeared after a platform or app update, and how it was detected.
Deciding What to Fix First
The abandonment rate on its own supports almost no decision. What does is knowing which step loses people, which segment they belong to, what recordings show them doing before they leave, and what the affected orders are worth. Those four answers turn a headline percentage into a ranked list of work.
Start by separating the shoppers who were researching from those who were blocked, because the first group sets a floor the store cannot go below and the second group is where the recoverable revenue sits. Then match the remedy to the cause the evidence actually supports, rather than to the one that appears first in published research, and change one thing at a time so the result remains readable. Cart abandonment solutions are cheap to buy and expensive to buy in the wrong order.
From Abandoned Carts to Recovered Revenue
If your store has a cart abandonment figure but no explanation for it, the useful next step is a diagnosis rather than another application. WD Market’s CRO and growth support covers that work: validating the tracking, isolating the step and segment where orders are lost, reading the behavioural evidence behind the drop, and returning a prioritised list of changes with the expected value of each and a way to verify it. Where the causes reach into shipping logic, payment configuration or a multi-market setup, implementation is part of the engagement, as in our Evelatus Baltic rebuild, where checkout was localised across three markets.
To discuss where your own abandonment is concentrated and what it is worth recovering, get in touch with the team. We also publish shorter conversion and ecommerce analysis on WD Market’s LinkedIn page for teams working through the same questions.
Questions Ecommerce Teams Ask About Cart Abandonment
What is a normal cart abandonment rate for an established store?
Published averages cluster around 70%, but the comparison is less informative than it looks. Rates depend on how the platform counts sessions and carts, how much traffic is researching rather than buying, and the consideration cycle of the category. A retailer of considered, high-value goods will sit above one selling routine repeat purchases without anything being wrong. Your own trend, split by device and by new versus returning visitors, is the more reliable signal.
How much traffic do we need before a diagnosis is reliable?
Diagnosis and testing have different requirements. Locating the step where sessions stop, and watching recordings of them, produces useful conclusions at modest volumes because the evidence is qualitative and patterns repeat quickly. Proving a change worked through a controlled test needs considerably more traffic and a longer window. Smaller stores can still act, using before-and-after measurement over a stable period and accepting less certainty in exchange for progress.
Do abandoned cart emails fix the underlying problem?
They recover a share of the revenue without addressing what caused the loss. That is worth having, and recovery sequences are among the most reliable of the available cart abandonment solutions, but a store relying on them alone pays repeatedly for the same defect. Treat the messaging as both a revenue stream and a research instrument: click behaviour and any question attached to the email say something about why the cart was left.
Should we offer a discount to recover abandoned carts?
Sometimes, and later in the sequence than most stores place it. A discount recovers buyers blocked by price, but also pays buyers who would have returned anyway, so its real cost includes the margin given to that second group. Regular abandoners can learn the pattern and start waiting for the code. Leading with delivery detail or return terms, and holding any incentive for a later message, keeps the cost proportionate to what it buys.
Our abandonment is worst on mobile. Is that a real problem?
Partly real and partly structural. Mobile carries more browsing and comparison, and some sessions end because the shopper moves to a laptop, which session-based reporting records as an abandonment. What separates that from a defect is behaviour inside the session: corrected inputs, repeated taps on an unresponsive button, and stalls at one field point to an interface problem rather than a change of device.
When is it worth bringing in outside help rather than fixing this internally?
Internal work is usually better value when the cause is known, the fix is contained and someone owns the follow-through. External support earns its cost when several causes are tangled, when analysis keeps stalling because nobody has protected time for it, or when the fix touches shipping, tax or payment configuration, where a mistake creates operational problems rather than a lost test.