The Untapped $47 Per Guest Night: How AI Unlocks Hotel Ancillary Revenue

Occupancy is fine. ADR is fine. RevPAR is fine. The asset is performing exactly the way the comp set is performing, which is to say that it is performing exactly the way the macro environment lets it perform. And yet the owner wants another four points of GOP margin, the brand wants another two points of loyalty contribution, and the lender wants debt service coverage that the rate environment is not going to deliver from rooms alone. The next meaningful dollar of margin isn't coming from rooms. It's coming from everything else - parking, EV charging, pre-arrival upgrades, F&B attachment, spa, late checkout, experiences, retail, and the dozen other categories that the industry collectively calls ancillary and individually under-resources.

The numbers are harder to ignore than they used to be. In CBRE's sample of 1,178 U.S. hotels, parking brought in $11.53 per occupied room in 2023, up 23.1% since 2019 and more than four times the growth in total hotel revenue. CBRE's H1 2025 data has F&B revenue per occupied room up 3.8% year over year, against 3.0% for total revenue. CBRE calls F&B a bright spot. Operators who treat ancillary as a discipline are the ones positioned to collect it.

AI can now do the four things ancillary revenue needs at scale: identify which guests are likely to buy which products, price those products against demand and inventory, time the offer to the moment the guest is most likely to say yes, and measure attribution cleanly enough for the learning to compound. That last one is what most operators have lacked, and it's what turns ancillary from an art into a system. This piece assumes you already know RevPAR and TRevPAR. It adds the AI layer.

The Ancillary Revenue Heat Map: Where the Margin Actually Lives

Most hotel revenue plans treat ancillary as a single line item - "other operated departments" on the P&L. That accounting convenience hides the operational reality that ancillary is really ten categories, each with its own demand pattern, margin profile, customer overlap with rooms, and AI surface area. The heat map below is a working taxonomy for scoping an ancillary program, sorted roughly by capture potential: the combination of margin, addressable percentage of guests, and improvement runway available through AI today.

Category	Revenue scale	Margin profile	AI capture potential
Parking & EV charging	$8.35 (limited-service) to $14.85 (resorts) per occupied room, CBRE 2023	61.3%, CBRE 2023	High - pricing, allocation, EV demand
Pre-arrival room upgrades	Depends on inventory and pricing	Very high; the room already exists	Very high - propensity + dynamic pricing
F&B attachment (breakfast, in-room dining, banquet add-ons)	Largest ancillary category by dollars	29.1%, CBRE H1 2025	High - bundling, menu engineering
Spa & wellness	Concentrated at resorts and luxury	Labor-heavy	Very high - slot pricing, propensity
Late checkout & early check-in	Small per stay	Very high; little marginal cost	Very high - almost zero marginal cost
Experiences & activities	Much larger at resorts	Depends on own-operated vs. commission	High - recommendation + cross-sell
Retail & merchandise	Small	Moderate; inventory cost	Moderate - inventory-driven
Resort & destination fees	Varies by market	High, now under disclosure rules	Low - regulatory headwind
Day-use & co-working	Small, emerging	Varies	Moderate - emerging category
Pet, transportation, ancillary services	Small	Varies	Moderate - bundling opportunity

Parking and F&B figures in the table come from CBRE; the other cells are qualitative, because no published benchmark covers them reliably.

The right way to read the heat map is to look where margin and AI potential overlap. The two highest-margin categories - late checkout and pre-arrival upgrades - also happen to be the two with the most AI capture potential, because the product is digital, the marginal cost is near zero, and the buying decision is highly time-sensitive. The two weakest categories - resort fees and retail - share the opposite profile: either the regulatory environment is moving against them (resort fees) or the inventory friction is real (physical retail). A coherent ancillary program leans toward the high-potential corner of the matrix and treats the rest as long-tail.

Why Ancillary Revenue Has Quietly Become the Margin Story of 2026

The backdrop is a split market. HotelData.com's Q4 2025 profit report has U.S. TRevPAR down 8.8% from 2024 to 2025, from $165.95 to $151.34. It says luxury and independent properties leaned into F&B, events, parking and resort fees to prop up income, while economy and parts of midscale faced guests who shopped harder, shortened stays and resisted add-on spend. Leaning in didn't save everyone: the same report has independents missing ADR budget by 10% and losing 4.5 points of margin.

Labor is the other pressure, and HotelData.com credits "tighter labor deployment" with protecting what margin operators kept in 2025. That makes department margin the right filter for where to push. A dollar of parking revenue at CBRE's 61.3% parking margin leaves 61 cents. A dollar of F&B at CBRE's 29.1% leaves 29. Both matter. They don't matter equally.

The third pressure is regulatory. The FTC's junk fees rule took effect in May 2025, requiring hotels to display total price upfront including mandatory fees. Resort fees themselves remain legal; the rule requires disclosure, not elimination. Hidden mandatory fees as a revenue strategy are on their way out, though, and properties that built ancillary capture on opaque fees have good reason to shift to optional, value-perceived ancillaries: pre-arrival upgrades, experiences, F&B, spa, and the rest of the category where the guest actively chooses to pay. Doing that profitably across thousands of guests is where AI earns its keep.

The Four AI Capabilities That Make Modern Ancillary Capture Work

Stripping the jargon out of the vendor pitches, an AI-driven ancillary program needs to do exactly four things well. Plenty of platforms claim all four. Use the framework below as a diligence lens when evaluating an upsell or ancillary platform.

Capability	What it does	Signal of maturity
Propensity modeling	Predicts the probability a specific guest buys a specific product at a specific moment	Per-segment propensity scores updated in real time, not batch
Dynamic pricing	Adjusts ancillary prices by demand, inventory, season, channel, and competitive position	Rates change at least daily; explicit elasticity model behind them
Channel orchestration	Routes the offer to email, app, SMS, online check-in, kiosk, or front desk based on guest behavior	One offer system feeds all touchpoints; no duplicate or conflicting offers
Attribution & learning	Cleanly measures lift, feeds outcomes back into the propensity and pricing models	A/B testing native; control groups available; clear incrementality reporting

Of the four, attribution is the capability hotels most consistently under-evaluate at the point of vendor selection and most consistently regret 12 months later. A platform that drives offers but cannot tell you which offers were truly incremental - versus which would have happened anyway - produces revenue growth that looks great in board decks and dissolves under owner scrutiny. The diligence question to ask every vendor in the pitch is: "Show me your incrementality methodology." The vendors who have one will show it instantly. The vendors who do not will pivot to a different topic.

Propensity modeling is the second place vendor quality varies a lot. Stronger engines train models on booking history, past ancillary spend and real-time context, and produce per-guest, per-offer scores that keep updating. Weaker platforms apply rule-based segments (business vs. leisure, weekday vs. weekend) and call it AI. Scale decides how much that matters. A 50-room independent may not need a propensity model at all; a 500-room luxury property probably does.

Pick ancillary categories by department margin and by how many of your guests can actually buy, not by how good the vendor demo looks.

The Pre-Arrival Window: Where Upsells Are Won or Lost

If a property could invest in only one moment of the stay, it should be the pre-arrival window, roughly the week or so before check-in. Oaky, an upsell vendor, reports from its own platform that a first upsell email sent 7 to 10 days before arrival drew a 48% click-through rate and 10.6% conversion, and a follow-up three days out converted at 12%. Its site cites an ROI of 41 at Hard Rock Hotel New York. Vendor numbers, so treat them as a ceiling. The logic holds up anyway: pre-arrival is the only window where the guest has committed to the hotel but has not yet committed their wallet to the rest of the trip. They are receptive, they have time to decide, and they have not yet spent on the dinners and excursions that will compete with the hotel offer once they arrive.

The mistake most properties make in the pre-arrival window is treating it as a generic email send. A better design is a four-touch sequence matched to each guest's propensity score and arrival timing, like the one below.

Touch	Days before arrival	Channel	Offer type
1 - Confirmation enrichment	Booking day	Email	Practical info + 1 highest-propensity offer
2 - Soft upgrade nudge	7 days out	Email + SMS for VIP	Room category upgrade with dynamic price
3 - Experience bundling	3 days out	Email + app push	Spa, F&B, activity based on stay purpose
4 - Online check-in conversion	1 day out	Web check-in flow	Late checkout, parking, breakfast bundle

The single biggest performance lever inside this sequence is the room upgrade offer at touch #2. Pre-arrival upgrades carry very high margin, since the room is already empty, the operating cost is already sunk, the only incremental cost is housekeeping prep for a different category - and the conversion rate scales with how well the price-to-perceived-value ratio is tuned. Static upgrade pricing (e.g., $30 to upgrade from a king to a king with balcony, every night, every guest) leaves money on the table on high-demand nights and produces no conversions on low-demand nights. Dynamic upgrade pricing, driven by available inventory, paid upgrade history and channel mix, fixes both ends of that.

Parking and EV Charging: The Quiet $11+ Per Room

Of all the categories on the heat map, parking is the one most chronically under-managed by hotel revenue teams. CBRE's data is clear: $11.53 average parking revenue per occupied room across its 1,178-hotel sample in 2023, $14.85 at resorts, $8.35 at limited-service. Department profit margin: 61.3%, materially higher than the 58.7% margin produced by every other operated department combined. Growth from 2019 to 2023: 23.1%, more than four times the growth rate of total hotel revenue. Parking is a department that, when ignored, quietly delivers a high-margin annuity, and when actively managed, becomes one of the strongest levers in the whole ancillary program.

The active management starts with dynamic pricing. A flat $40 per night parking rate is the equivalent of selling rooms at a flat rate every night of the year - a strategy hotels abandoned a generation ago for rooms and somehow still tolerate for parking. AI-driven parking pricing modulates by event calendar, weather, occupancy mix (group vs. transient), and competitive lot rates within a half-mile radius. The same software handles the EV charging conversation that every property must now have. A 2022 Greenview and AHLA report, cited by CBRE, found 26.6% of U.S. hotels had EV charging, and CBRE expects that share has grown since. The economic question is no longer whether to install EV charging but how to price it: bundled with parking, separate paid utility, or amenity-free as a loyalty perk. The right answer varies by property and demand mix, but in every case the question deserves an explicit pricing strategy rather than a default.

The F&B Attachment Engine

Food and beverage is the largest ancillary category by absolute dollar volume and the one with the most operational complexity. CBRE's H1 2025 analysis documented F&B revenue per occupied room up 3.8% year over year and department margin expansion from 28.7% to 29.1% in the same period. The bright-spot framing is fair. The operating challenge is real too. F&B requires labor, inventory, square footage, and a brand-aligned product, and the AI levers available to it are narrower than in pure-digital categories like pre-arrival upgrades.

Where AI moves the needle most in hotel F&B is on attachment, not on operations. The single biggest gap between leading and trailing properties in the F&B P&L is the percentage of room nights that produce any F&B spend at all. It's usually far higher at luxury properties than at select-service hotels. The AI lever is the personalized pre-arrival and in-stay nudge that converts the guest who would have eaten in their room or walked to the nearest restaurant into the guest who eats in the hotel restaurant or orders in-room dining.

The table below maps the F&B attachment opportunities by guest segment and the AI capability that unlocks each one.

Segment	Default F&B behavior	AI-enabled conversion lever
Business traveler - short stay	Skips breakfast, eats out for dinner	Pre-arrival breakfast bundle at 15 to 20% discount
Group attendee - conference	Group F&B only	Evening cocktail offer + retail upsell
Leisure couple - weekend	One in-property meal max	Tasting menu + sommelier pairing pre-arrival
Family - vacation	Kids' menu + casual dining	Activity-bundled F&B (poolside lunch, dinner&movie)
Extended-stay business	Grocery store offsite	Stocked-room option + delivered meal credits

Menu engineering is the second AI layer that pays back inside hotel F&B. Modern systems analyze item-level profitability against item-level guest demand and surface menu rationalization opportunities, lifting average check by repositioning the highest-margin items in the highest-attention positions and demoting or eliminating low-margin items the guest would not have ordered anyway. The work is mechanical, the upside is real, and and the effort is a regular chef-and-finance session, not a project.

Spa, Wellness, and Late Checkout: The Highest-Margin Ancillaries Most Properties Mis-Price

Spa and wellness is the under-discussed engine of luxury and resort ancillary economics. In CBRE's 2022 data, 139 U.S. hotels that operate their own spa averaged $7,097 in spa revenue per available room, 12.6% above 2018. The category is concentrated in resorts and luxury, but the discipline behind those numbers (slot inventory management, per-treatment pricing, therapist scheduling against forecast demand, pre-arrival booking) travels down the chain scale and is under-used at plenty of upper-upscale properties.

Late checkout is the simplest and most frequently squandered high-margin ancillary in the hotel P&L. Its value shows up the morning of departure, when the guest is deciding when to pack, not at booking, so the natural time to sell it is in-stay. The right system for late checkout is an in-stay AI nudge sent the evening before departure, dynamically priced against next-day arrival pattern (no charge if arrivals are light, premium charge on heavy-arrival days), with one-click acceptance and automatic PMS update. The marginal operating cost is a shift in the housekeeping schedule. The category is essentially free margin for any property that bothers to systematize it.

Experiences and Activities: The Resort-Calibrated Lever

Experiences are the category where the gap between resort properties and city properties is widest and the AI play is most distinctly segmented. At a resort, experiences can be a large revenue line (golf, spa packages, snorkeling, fishing charters, cooking classes, off-property excursions), and the AI lever is recommendation density and cross-sell across the on-property activity calendar. A guest who books snorkeling on day one is a better prospect for a sunset sail on day three than a guest with nothing booked; the AI surfaces that propensity in real time and the front desk or concierge app delivers the offer.

At a city property the experiences lever looks different. The property is rarely operating its own activities; it is brokering third-party experiences (museum tickets, walking tours, dining reservations, theater) and earning a commission. The AI value here is in the matching algorithm - the speed and precision with which the property can match an arriving guest with the right experience for the right night with the right operator. The economics are smaller per transaction but the addressable percentage of guests is much higher, because every city guest has open evenings that the hotel can either monetize or surrender to the city's general inventory.

The implementation question both segments share is whether to build, partner, or buy. The build option (proprietary activity-booking platform) makes sense at large resorts with established activity operations; the partner option (integrating with a third-party experiences platform) is the right answer for most city hotels and smaller resorts; the buy option (white-labeled experiences marketplace) sits somewhere in between. Hotels that are scoping the right architecture and integration depth for their ancillary technology stack often benefit from an outside diagnostic that maps current systems to revenue use cases and finds the biggest gaps. explore our AI Revenue Optimization & Forecasting service →

The Implementation Sequence That Actually Pays Back

The ancillary build-out is the kind of project that fails when sequenced wrong. The instinct is to deploy everything at once because the operating committee approved everything at once. The discipline is to deploy in stages calibrated to capability dependency, organizational bandwidth, and short-term ROI visibility. The 12-month sequence below assumes a property starting with essentially no ancillary tooling.

Phase	Timeline	What you deploy	What to look for
Phase 1 - Foundations	Months 1 to 2	Parking dynamic pricing, pre-arrival upgrade engine, late checkout automation	Measurable lift in parking and upgrade RevPOR against your baseline
Phase 2 - Personalization	Months 3 to 5	Propensity model deployment, segmented pre-arrival sequence, F&B attachment nudges	Higher offer conversion in the targeted segments
Phase 3 - Experience layer	Months 6 to 8	Activities marketplace integration, spa dynamic pricing, in-stay AI nudges	Experience and spa RevPOR lift, larger at resorts than city hotels
Phase 4 - Attribution & optimization	Months 9 to 12	Incrementality testing, channel orchestration, propensity model retraining	ROI reporting you can defend to ownership

Phase 1 has to fund the rest. Illustrative arithmetic, not a benchmark: at 300 keys and 70% occupancy, every $1 of incremental RevPOR is worth about $77,000 a year in revenue, so $3 is about $230,000 and $8 about $613,000. Put your own vendor quotes against that and you'll know quickly whether the payback case for Phase 2 is there.

Governance: Who Owns Ancillary Revenue

A common implementation failure is governance ambiguity. Rooms revenue has a clear owner (the revenue manager). F&B has a clear owner (F&B director). Spa has a clear owner (spa director). But ancillary as a category - parking pricing, pre-arrival upgrade pricing, late checkout offers, F&B attachment campaigns - typically has no single owner, which means it has no consistent strategy and no consistent reporting cadence. The structural fix is to designate ancillary as a named GM-level discipline with an explicit owner: usually the director of revenue management at smaller properties, sometimes a dedicated commercial director at larger groups.

The reporting cadence is the second governance lever. Ancillary should appear in the weekly revenue meeting alongside rooms, with the same rigor of forecast vs. actual, segment mix, channel attribution, and forward-looking pipeline. Monthly reporting is too slow to correct a pricing rule or a bad offer before it costs a season. The reporting discipline is the cultural change that locks in the operational gains.

Ancillary gains come mostly from operating discipline, not capital. Give the category one owner before you buy anything.

Measuring What Works: TRevPAR, RevPOR, and the Attribution Trap

The metrics that matter for ancillary are not the metrics that matter for rooms. TRevPAR (total revenue per available room) is the headline number, but the diagnostic that actually drives operating decisions is RevPOR (revenue per occupied room) broken down by ancillary category. A property whose TRevPAR is up 4% may be capturing genuine ancillary lift or may simply be running higher occupancy with flat ancillary capture; the per-occupied-room breakdown reveals the truth.

The attribution trap is the second measurement issue every program runs into. A property that deploys pre-arrival upgrade software and sees ancillary revenue rise 8% will typically credit the platform for the full lift. The honest answer is more complicated. Some of the lift would have happened anyway through walk-up upgrades at check-in; some of the lift is incrementally captured by the platform; some of the lift is shifted from in-stay (where it would have happened) to pre-arrival (where it is now captured). Clean incrementality measurement requires control groups, randomized hold-outs, and statistical rigor that most vendors are uncomfortable providing because the answer is always less impressive than the headline number. Build attribution discipline early and you won't spend two years renewing the wrong vendor.

Frequently Asked Questions

We are a 120-room independent without a dedicated revenue manager. Is this realistic for us?

Yes, but the sequencing changes. At independent scale skip the full propensity-modeled stack and start with two high-impact, low-complexity deployments: dynamic parking pricing (or paid parking introduction if you currently bundle) and a pre-arrival upgrade tool from an independent-friendly vendor. Neither is a long project, and both show up quickly in RevPOR if they're working. That contribution then funds the next two or three deployments. Independents that try to deploy everything at once typically get stuck implementing nothing; independents that sequence to ROI compound their capability quarter by quarter and look very different at the 24-month mark.

How do the new FTC junk fee rules affect our ancillary strategy?

The rule changes pricing transparency, not legality. Resort fees and mandatory service charges remain legal but must be displayed in the upfront total price the guest sees at the moment of booking. Operationally this means properties that have been using opaque mandatory fees to inflate yield are now disadvantaged against properties that compete on transparent rates plus optional ancillaries the guest actively chooses. The right strategic response is to migrate revenue capture away from mandatory fees and toward optional ancillaries - upgrades, experiences, F&B bundles, late checkout - where AI-driven personalization can produce comparable yield with a better guest experience and a defensible regulatory posture.

What is the right vendor strategy - best-of-breed or all-in-one?

For properties under 250 keys, an integrated all-in-one platform (Canary, Oaky, Duve, or similar) usually beats best-of-breed on total cost of ownership. The integration friction of stitching multiple vendors together exceeds the incremental feature advantage of any single specialist tool at that scale. Above 500 keys the math flips: best-of-breed for pre-arrival upsell (Oaky, IRIS), parking (ParkHub, ABM), F&B attachment (in-PMS or specialized tools), and spa management (Book4Time, Mindbody) typically produces measurably better outcomes per category, and the integration cost is justifiable. Between 250 and 500 keys the call depends on technical capability and group buying power. At any scale, don't assume the PMS vendor's bundled ancillary module matches the specialists. Test it.

How do we measure incrementality honestly?

The gold standard is randomized hold-out testing: a fixed share of eligible guests, large enough to be statistically valid, is excluded from the AI-driven offer flow at random, the spend of the hold-out group is compared against the offered group, and the difference is the genuinely incremental lift attributable to the platform. Most vendors resist this design because it makes the headline number smaller. The vendors who embrace it are the ones worth long-term partnership. Where randomized hold-outs are operationally impractical - at small properties where the statistical sample size is insufficient - the workable substitute is rigorous A/B testing on offer content, timing, and price points, combined with year-over-year same-property analysis controlling for occupancy and rate. The methodology is less clean but more honest than the unconditional "before vs. after" reporting most platforms default to.

What is the ROI timeline a CFO can credibly underwrite?

There's no published benchmark a CFO should take on faith, including vendor case studies. Underwrite from your own baseline instead: current RevPOR by ancillary category, the lift you're assuming in each, quoted costs, and a payback month. Then tie the case to three operating milestones the CFO can watch monthly, typically pre-arrival upgrade conversion, parking RevPOR and F&B attachment rate. If the milestones slip, the financial case gets revised before the money's gone.