The cart-abandonment recovery industry has, over the previous decade, established itself as a routine fixture of retail commerce procurement. The industry’s core proposition rests on a small number of widely-circulated statistics, of which the most consequential is the claim that approximately eighteen billion dollars of revenue is, each year, abandoned at the checkout stage of retail commerce sites and is therefore available for recovery by an appropriately-deployed remarketing programme. The figure is presented in essentially every procurement document the industry produces; it is reproduced, with minimal critical examination, in the trade publications that cover the sector; it is presented to retail decision-makers as an authoritative description of the commercial opportunity the industry’s products are intended to address.
The figure is, on careful examination, substantially fictional. The fiction is constructed in a manner that is, in some sense, more sophisticated than a simple invention; the industry has, over the years, developed a methodology for arriving at the figure that produces a plausible-sounding number while obscuring the assumptions on which the number depends. This post is an account of how the figure is constructed, what the actual abandonment-recovery opportunity looks like on the data, and what works in the recovery space when the marketing layer is set aside.
How the headline figure is constructed
The eighteen-billion figure, in its various annual restatements, is derived from approximately the following calculation. The retail commerce sector’s total annual revenue is taken from a published source. The proportion of carts that abandon before checkout completion is taken from a separate published source, typically a survey of retail platforms or a study of session-recording data. The two figures are multiplied to produce the “total value of abandoned carts,” which is then presented as the abandonment-recovery opportunity.
The mathematical operation is straightforward; the conceptual error is decisive. The “total value of abandoned carts” is the value that visitors had assembled in their carts at the moment of abandonment; it is not the value of the orders those visitors would have placed had the abandonment not occurred. The two figures differ for the following reasons.
First, the cart contents at the moment of abandonment are, in a substantial fraction of cases, not what the visitor would have purchased had the visitor proceeded. Many visitors use the cart as a kind of saved-for-later list, adding items they are considering but never intended to purchase in a single transaction; the cart at abandonment therefore reflects the visitor’s exploration rather than their purchase intent. The proportion of cart contents that the visitor was actually intending to purchase is, on the data I have access to, considerably less than half.
Second, the visitors who abandon and do not return have, by hypothesis, decided not to proceed with the purchase. The decision is rarely a result of the recovery-industry’s preferred narrative — the visitor was distracted, encountered a temporary obstacle, would have proceeded had they been gently nudged — and is, on the qualitative interview data, more frequently a substantive decision that the purchase was not, on reflection, what the visitor wanted. The “recovery opportunity” is therefore not the entire abandoned cart value; it is the small fraction of the abandoned value that corresponds to abandonments where the substantive decision had not yet been made.
Third, the visitors who would have proceeded had they been recovered would, in a non-trivial fraction of cases, have proceeded without the recovery intervention. The recovery campaign is, in this fraction, taking credit for revenue that would have arrived regardless. The proportion that genuinely required the recovery intervention to convert is, on the controlled-experiment data I have access to, considerably smaller than the recovery campaigns’ headline reports suggest.
The combined effect of the three corrections is that the genuine abandonment-recovery opportunity is, in my analysis, between five and twelve per cent of the headline figure. The recovery industry’s headline statistic, on the same analysis, is therefore inflated by a factor of approximately ten over the genuine commercial opportunity it claims to describe.
Why the industry continues to circulate the figure
The reason for the figure’s continued circulation is, in my view, straightforward and not particularly mysterious. The recovery industry’s commercial proposition depends on retailers believing the recovery opportunity is large; the larger the believed opportunity, the larger the budget retailers are prepared to allocate to the recovery products the industry sells. The eighteen-billion figure is the figure that produces the budget allocation the industry’s commercial model requires; the figure’s relationship to the actual commercial opportunity is, for the industry, a secondary consideration.
The retailers who allocate budget on the basis of the figure are, accordingly, allocating against a believed opportunity that is approximately ten times the actual opportunity. The recovery products the budget purchases produce, in the genuine recovery rates the products achieve, returns that fall short of what the believed opportunity would predict; the gap between the predicted and actual returns is, in the procurement reviews that follow, frequently attributed to implementation issues rather than to the headline figure that determined the budget in the first place.
What actually works in recovery
The genuine recovery opportunity, while substantially smaller than the industry claims, is real; the techniques that produce measurable recovery within the genuine opportunity are worth describing.
The first is the inline recovery within the same session. A visitor who abandons the cart and remains on the site can, in some non-trivial fraction of cases, be recovered by a small contextual prompt that addresses the proximate cause of the abandonment — a payment-method indicator that surfaces additional payment options, a delivery-window adjuster that provides a faster shipping option, an account-creation alternative that permits guest checkout. The inline recovery produces, on the data I have collected, considerably higher recovery rates than the email-based recovery the industry typically sells; the inline pattern intervenes before the visitor has lost the context of the purchase decision, while the email pattern intervenes after the context has been lost.
The second is the targeted email recovery for the specific subset of abandonments where the recovery is genuinely available. The subset is identifiable from the abandonment context — visitors who progressed past the payment stage but did not complete the purchase, visitors whose cart values exceeded the median, visitors whose session duration suggests considered purchase intent rather than exploration. The targeted email to this subset produces recovery rates approximately three times those of the broad-population email recovery, with correspondingly more favourable economics for the retailer; the broad-population email recovery is, on the data, the principal source of the gap between the industry’s claimed recovery rates and the rates retailers actually achieve.
The third is the structural intervention on the underlying cause of the abandonment, where the cause is identifiable and addressable. The cart-abandonment data, examined carefully, frequently surfaces specific patterns in the session-recording data — visitors abandoning at the shipping-cost surface, visitors abandoning at the account-creation prompt, visitors abandoning when the delivery window does not match their need. Each of these patterns is, in operational terms, a structural problem with the checkout that the retailer can address directly, and the structural address consistently produces a larger reduction in abandonment than the recovery intervention against the same baseline. The recovery industry, in the procurement conversations I have, rarely advocates for the structural address despite its substantially better commercial outcomes; the structural work is, by hypothesis, work the industry’s products do not perform.
An advisory close
The cart-abandonment recovery industry’s marketing materials are, in 2026 as in 2018, presenting a commercial opportunity that is approximately ten times the size of the actual opportunity that exists. The retailers who allocate budget against the marketing materials’ figure are, accordingly, allocating against a fictional baseline; the budget produces the recovery rates the underlying mechanisms can support, which are considerably below the rates the marketing materials predict. The procurement reviews that follow consistently attribute the gap to implementation issues; the gap is, in my analysis, attributable to the headline figure that determined the budget rather than to the implementation of the products purchased with it.
It is recommended that retailers evaluating recovery products do so with reference to the genuine recovery opportunity rather than to the industry’s headline statistic; the genuine opportunity is identifiable from the retailer’s own session-recording and cart-abandonment data, and is considerably smaller than the industry’s procurement conversations imply. The structural address of the underlying causes of abandonment is, on the data, consistently a more productive use of the budget than the recovery intervention against the resulting abandonments. (For a related view on the broader habituation of conversion-psychology techniques, see Free shipping is the only psychology hack that still works.)