At 11:42pm a woman named Allegra almost books a four-night stay at a coastal resort. She is three fields into the payment screen. Her phone rings, her kid is calling from college, she closes the browser to take it, she means to come back, she does not.
By 11:48pm an automated email lands in her inbox: “Complete your booking.” Subject line cheerful, button large, discount code modest. She reads it the next morning on her walk. The link opens her desktop browser, which does not remember the cart. The page loads fresh, the dates have to be re-entered, the room she wanted is no longer at the price she saw. She closes the tab.
Three hours later, on her phone again, she books the same resort through Expedia, which remembered everything.
That is the state of abandoned-cart recovery in hospitality in 2026. A single email. One fire. One chance. And a device boundary the direct channel cannot cross.
The travel vertical has the worst abandonment rate of any category
The most rigorous benchmark in cart-abandonment research is the Baymard Institute’s meta-analysis of 50 studies, last updated September 2025. Across e-commerce broadly, the average is:
70.19% of online shopping carts are abandoned.
Baymard Institute, Cart Abandonment Rate meta-analysis
Travel, as a vertical, is worse. According to SaleCycle’s tracking of 280 million+ online bookings, travel abandons at 81.31%, with airlines at 87.87%. SaleCycle’s explanation is structural:
Travel purchases are the biggest form of ecommerce, with the highest price points. Bookings also have to be made in an inflexible way, and many sites require payment up front, which drives abandonment higher than retail.
SaleCycle, Why Travel Sites Have Higher Abandonment Rates
Hotels, specifically, sit somewhere inside that travel aggregate, typically lower than flights because many bookings do not require upfront payment, higher than retail because the consideration period is longer. Public figures for “hotels specifically” are scarce and mostly circulate on vendor blogs without a primary source. We are going to use the defensible figure: travel at 81%, per SaleCycle.
That is the denominator.
What e-commerce has industrialized, and hotels have not
The e-commerce industry built an entire sub-economy around cart recovery. Klaviyo’s 2024 benchmark analysis of 143,000 abandoned-cart flows gives the cleanest picture of what is achievable at scale:
Abandoned cart flows generate an average placed-order rate of 3.33%. Top-performing flows hit 7.69%. Average revenue per recipient is $3.65; top performers hit $28.89.
Klaviyo, Abandoned Cart Email Benchmarks 2024
Two things are worth sitting with.
The first: 3.33% is not a big number. The entire e-commerce abandoned-cart industry, the flows, the tools, the consultants, the agencies, is built on top of a mid-single-digit recovery rate, because the gross abandoned volume is so large that even a 3% recovery is a meaningful business.
The second: the top decile is 2.3x the average. The difference is not discount depth. It is infrastructure, specifically, multi-touch sequences, cross-device identity resolution, session resumption, dynamic inventory hold, payment continuity, and channel orchestration across email, SMS, push, and retargeting.
Hotels have not built any of this. The state of the art at an independent in 2026 is: Revinate Marketing (triggered email when booking engine captures a form field), Cendyn CRM (rule-based retargeting email on engine abandonment), Inntopia or CartStack RezRecover (standalone recovery-email products).
What every one of these has in common: a single trigger, a single channel, and no cross-session memory. The email fires, it fails, the thread dies. Allegra’s story, scaled across the industry, is the product.
The TAM math, shown with sources
Let us build the number carefully, using only primary-source inputs.
Input 1: Statista’s 2025 global hotels market forecast projects global hotel market revenue at $455 billion in 2025.
Input 2: Phocuswright reports that global online travel bookings will hit $1.2 trillion by end of 2026, with roughly 65% of travel bookings now transacted online. Applied conservatively to hotels, 60% of the hotel market is online-booked. 60% × $455B = ~$273B in online hotel bookings.
Input 3: SaleCycle’s travel-sector abandonment rate is 81.31%. If $273B is the completed online hotel revenue, the abandoned side of the funnel, at an 81% abandonment rate, is mechanically larger. Solving: $273B / (1 − 0.8131) × 0.8131 = roughly $1.19 trillion in abandoned hotel-booking attempts per year.
That is the gross. Nobody gets to recover the gross. You get to recover the slice that a well-run abandoned-cart program can convert.
Input 4: Klaviyo’s 2024 e-commerce recovery rate is 3.33% average, 7.69% top decile. Applying those rates to the $1.19T abandoned-attempt pool:
- At the e-commerce average: $1.19T × 3.33% ≈ $39.6B/year recoverable
- At the e-commerce top decile: $1.19T × 7.69% ≈ $91.5B/year recoverable
Call it a $40B to $92B recoverable TAM. Pick the end of the range you want to defend. It is real money no matter which end you pick, and the hotel industry, collectively, is capturing a low-single-digit percentage of it through one-email recovery flows that fire once and die.
The shape of the gap, per booking
The average hotel direct booking, per SiteMinder’s analysis of 125 million reservations, is $519. One recovered abandoned booking, at that average value, pays for the entire cart-recovery infrastructure a mid-size independent would deploy in a year. Ten of them pays for the receptionist who is supposed to take the inbound call the recovery email is trying to replace.
And remember: the guest who abandons is, by definition, already in-funnel. They searched, they clicked, they chose a rate, they filled in a name, they stopped. The intent is documented. The only question is whether the hotel can keep the thread alive long enough to close.
What the e-commerce top decile actually does
The anatomy of a top-decile e-commerce recovery program is now well-documented. A condensed version of what separates 7.69% from 3.33%:
- Multi-channel, multi-touch: an email at 1 hour, an SMS at 24 hours, a retargeting ad at 48 hours, a “price drop” email at 72 hours. Not a single fire.
- Cross-device identity resolution: the phone cart is recognized on the desktop, and the desktop cart is recognized on the phone, because the identity is not the browser cookie, it is the authenticated user.
- Session resumption: the cart state is persisted server-side and restored on next visit, with the same items, the same selected options, the same prices honored for a limited time.
- Inventory hold: the cart items are reserved for a finite window so the customer does not return to find the room sold or the rate changed.
- Contextual re-engagement: the second message is not “you left something,” it is a specific reference to what you left and what has changed about it.
None of this is speculative. It is the operating playbook of every serious DTC brand over $20M GMV. It has not arrived, in any integrated form, at the independent hotel.
The memory problem is the recovery problem
The reason e-commerce recovery works at 3-8% and hotel recovery does not is not email creative. It is memory architecture.
An e-commerce cart is a persistent object tied to a resolved identity. A hotel booking attempt, in most independents’ stacks today, is a transient form-field capture at the booking engine. The identity is whatever email address the guest happened to type into the first field. If Allegra types allegrapark@gmail.com on Monday night and a.park@work.com on Tuesday morning, she is two people to the system. Neither profile has her cart. Both will be emailed to inboxes she does not monitor.
The cross-device, cross-session, cross-channel identity problem is the same problem we wrote about in Your Returning Guest Has 2.3 Profiles in Your PMS. It is the same problem that leaves Thomas the night auditor as the de facto CRM. It is the problem underneath all the other problems.
A recovery flow that actually captures the top-decile 7.69% has to live on top of a memory layer that can recognize Allegra across sessions, across devices, across the rate-search-then-price-bump-then-return loop, and resume the cart where she left it. Not email her a reminder. Restore the state.
What FlowStay builds against this
FlowStay’s abandoned-booking recovery is not a trigger-and-send email product. It is:
- A memory-resident cart that persists across session, device, and channel boundaries, tied to the resolved guest identity, not the browser cookie.
- A multi-touch re-engagement sequence that blends email, SMS, and, where the guest is a voice-channel prospect, an outbound callback at the right hour of the day.
- Real-time rate honoring within the recovery window, so Allegra’s Tuesday morning link does not open a fresh search, it opens her Monday night cart at Monday night’s rate.
- Attribution and NDC accounting so the recovered booking lands in the direct-channel column of the revenue manager’s Monday morning spreadsheet at a $3 acquisition cost, not in the OTA column at $22.
None of this is a marketing claim against the existing vendors. It is a structural claim about what the vendors were built to do. A trigger-and-send email tool, whether from Revinate or Cendyn or CartStack, is a well-made version of the product e-commerce stopped shipping in 2016. The $40B to $92B gap is what sits between that product and the multi-touch, memory-resident infrastructure the top-decile e-commerce programs run on.
Why this is the most undervalued lever on the GM’s dashboard
Every independent GM we have talked to in the past year has a direct-booking budget, a paid-search budget, and, if they are sophisticated, a metasearch budget. Most do not have an abandoned-booking recovery budget, because the product category, as a line item, has not been priced for independents.
The economics are inverted from paid acquisition. Paid acquisition spends money to acquire a new intent. Abandoned-booking recovery spends money to convert an intent the property already paid to acquire. Every recovered booking is a double-counted win: the acquisition cost has already been sunk, so the recovery revenue accrues at nearly pure margin.
At a 100-room property with a $519 average direct booking and even a 4% recovery rate on an abandoned-attempt pool of 2,000 per month (a conservative estimate for an independent with meaningful web traffic), that is 80 recovered bookings, ~$41,500 in recovered revenue, per month. At close to zero marginal acquisition cost.
Allegra’s story is not unusual. It is the default. The phone rings, the browser closes, the link in the follow-up email opens a dead session, the OTA wins by remembering what the direct channel forgot.
The industry has been writing emails about it for a decade. It is time to start remembering instead.