Industry · Hospitality

AI in hospitality with a focus on revenue management.

We build AI solutions that fit the operational reality of hotels, holiday parks, campsites and restaurant chains — from dynamic pricing and demand forecasting to review sentiment analysis and upselling at the right moment. Custom solutions that sit alongside your existing PMS and RMS rather than replacing them.

AudienceHotels & chains
AudienceHoliday parks
AudienceCampsites & glamping
AudienceRestaurants & F&B

The Dutch hospitality market in numbers.

~3.500
Hotel properties in the Netherlands
~2.300
Campsites and holiday parks
~46m
Hotel overnight stays per year
~75%
Bookings via OTA channels

Source: CBS Statline Accommodation 2024, NBTC Tourism in Figures, Phocuswright distribution mix 2024.

Standard RMS packages often don't match your reality.

The big names in revenue management — IDeaS, Duetto, Atomize, OTA Insight, Cendyn — are excellent if their model fits you. For enterprise hotel chains such as Accor, IHG or Marriott with a tightly integrated Opera or Mews stack, that is often an excellent starting point. But we get calls from the businesses where it doesn't fit.

A mid-market hotel chain that can't bear the licence costs of IDeaS but still needs measurable revenue impact. A holiday park where weekend, weather and local events give more signal than a standard package can handle. A restaurant chain for which no hotel RMS exists. A boutique hotel with its own pricing philosophy that doesn't want a vendor overriding it with a black-box algorithm. A conference or MICE venue where room planning, group bookings and F&B margins all need to be part of the same optimisation.

For these cases we build a custom layer on top of your existing PMS. Demand forecasting that takes your local signals into account, length-of-stay optimisation that fits your room mix, channel-mix models that favour direct bookings over expensive OTA commissions — all fed by your own historical data, not by a black-box benchmark you can never be sure is representative of your market. We connect that layer to a real-time analytics platform so your revenue team doesn't have to wait for tomorrow's report.

The difference lies in the philosophy: a vendor like IDeaS sells to a thousand hotels and therefore has to generalise its model. We build for a single client and can calibrate model parameters, data points and decision rules precisely to how your market works, which segments you serve and which distribution channels you want to strengthen. For the clients it suits, that delivers a measurably better ADR and RevPAR than an out-of-the-box package.

AI applications by type of hospitality business.

A hotel, a holiday park and a restaurant chain each need different models, different data points and a different pace of decision-making. For each audience, we show what we typically build.

Hotels & hotel chains

The classic revenue management disciplines: ADR, RevPAR, occupancy and length of stay. For mid-market chains, boutique hotels and eco-resorts, we build a custom RMS layer on top of Mews, Cloudbeds, Protel or Opera. Models that factor in city events, flight schedules, local weather and competitor rates from OTA Insight or Lighthouse.

The output lands directly in your channel manager — SiteMinder, RateGain, Cloudbeds Distribution — so rate changes automatically flow to Booking.com, Expedia and your own booking engine. We also often build LLM components for chatbots and personalised upsell emails.

For holiday parks, such as a Roompot- or Landal-style operation, we shift the focus to weekend and school-holiday patterns, household composition and regional events. For campsites and glamping, we use weather and the start of the season as core inputs — a cold spring shifts the whole curve, and your RMS needs to be able to respond to that. Restaurants call for a completely different model: booking rhythm, no-show risk, table-turn times and ingredient purchasing, not room rates.

  • Demand forecast per room classPer day, per channel, per length of stay — fed with events and weather data.
  • Dynamic pricing and rate pushAutomatic rate updates to all channels via SiteMinder or RateGain.
  • Overbooking optimisationNo-show models that find the tipping point between vacancy and denied-stay risk.
  • Upsell engine at check-inThe right room upgrade, the right guest, the right price — triggered on the day of arrival.
  • Review sentiment analysisTripAdvisor, Booking.com and Google reviews automatically classified and routed.

Built within the rules on guest data and automated pricing.

AI in hospitality touches personal data, automated decisions and payments. From sprint 1 we work within the framework of the GDPR, the AI Act, PSD2 and the relevant consumer directives.

GDPR Art. 5 + 35

Guest data minimisation and DPIA

Guest names, passport details, preferences and booking history are sensitive. For each model we build with the minimum dataset required, deliver a DPIA with every project and set up retention so that old booking data is automatically anonymised.

EU AI Act

Transparency in dynamic pricing

Automated pricing aimed at consumers falls under the transparency and non-discrimination requirements of the AI Act. We document which variables the model uses, avoid prohibited inputs and build in logging so that a guest can see on request how a price was determined.

DSA / DMA

Platform and gatekeeper rules

For businesses that sell through OTA platforms or run their own marketplace, we map out the DSA obligations. Which ranking criteria do you use, how do you explain them, what does your complaints route look like — all within the recommender transparency requirements.

PSD2 / SCA

Payments and strong customer authentication

Pre-authorisations, no-show charges and upsell payments follow PSD2 SCA. We integrate with Adyen, Stripe or Mollie and make sure 3-D Secure flows work correctly in the booking engine and upsell emails.

Tourist tax

Local levies processed automatically

Tourist tax differs by municipality, by type of accommodation and sometimes by season. Our pricing layer separates the room rate and the levy transparently, so your bookkeeping and tax returns run without manual corrections.

AI in hospitality is broader than pricing alone.

Revenue management is the most valuable use case and therefore the core of what we build for hospitality clients — but it is rarely the only place where AI makes a difference. A revenue project usually involves three to five more models that together form the complete picture.

Demand forecasting by room category, channel and length of stay — not just for pricing but also for staff planning, housekeeping and F&B purchasing. Customer segmentation using RFM and value models that focus your marketing budget on guests with the highest lifetime value. Personalised marketing via Revinate or Cendyn, with AI-generated email campaigns and offers that match a guest's previous stay and stated preferences.

Chatbots for 24/7 multilingual guest support on your website and WhatsApp, powered by your own FAQ and booking data. Voice assistants in rooms via Alexa for Hospitality or similar integrations for lighting, climate, room service and local tips. Predictive maintenance on HVAC, lift and pool installations, so faults are fixed before a guest phones reception in the middle of the night.

For restaurants and F&B chains, we also build waste prediction (how much fresh produce you need tomorrow, based on weather, day of the week and events), menu optimisation (which dishes deliver both margin and satisfaction) and no-show models that determine how much overbooking is responsible. On the guest side: sentiment analysis of TripAdvisor and Booking reviews that picks up patterns per location, and complaint classification that automatically routes incoming emails or chat messages to the right department — a housekeeping issue goes to housekeeping, a pricing complaint to the revenue manager, a safety report to management.

Which models get priority depends on where your operation loses the most time or margin. We map this out in the audit phase. A project often starts with dynamic pricing and demand forecasting, and grows organically with the models that deliver the most in phase two.

Seamlessly connected to the hospitality ecosystem.

We integrate with the PMS, RMS, channel manager and booking engine you already use. Our AI layer complements them — we never ask you to replace your core system.

Mews
PMS
Cloudbeds
PMS + distribution
Opera
Oracle PMS
RoomRaccoon
All-in-one PMS
SiteMinder
Channel manager
RateGain
Distribution + intel
OTA Insight
Rate shopping
Lighthouse
Market intel
Profitroom
Booking engine
Revinate
Reviews + CRM
Cendyn
CRM + marketing
Adyen / Stripe
Payments

The right integration in the right place.

Every project starts in the same place: which data lives where, and which integrations are needed to feed the model and get its output back into your operation. A hotel chain running on Opera gets a different integration architecture from a holiday park on Cloudbeds — but the logic behind the AI model stays the same.

Our choice is almost always not to replace existing RMS packages. If you already work with IDeaS, Duetto or Atomize and they do what they should — great. We only come into the picture when there are specific demand signals (local events, weather, B2B pipeline, your own loyalty data) that the standard vendor doesn't take into account, or when the cost of an enterprise RMS is out of proportion to your scale. For those scenarios, we build a custom AI implementation tailored to your data and your operation.

A typical integration architecture for a mid-market hotel chain looks like this: the PMS (Mews or Cloudbeds) is the source of booking data, room inventory and guest profiles. A rate-shopping tool such as OTA Insight or Lighthouse supplies competitor prices. External APIs add weather, flight data and local events. Our model layer gathers all of this in a data warehouse, trains the models periodically and sends pricing suggestions back to all distribution channels via the channel manager. Your revenue team always keeps an override button — automatic updates can be manually overridden or paused, for example in the case of an unexpected event or a commercial decision.

Not sure about a large project yet?

Test your idea first: a working prototype in 1 day

With OneDayBuild, we make your idea tangible in a single day for €1,150, so you know whether further development is worth the investment. Decide to go ahead with the full build? Then we deduct the cost in full.

View OneDayBuild →

From data audit to production model in clear steps.

An AI implementation for revenue management has its own rhythm. Five phases that we go through for every hospitality project.

01 · Audit

Data & PMS mapping

Which data sits in your PMS, channel manager and booking engine. How clean it is, how far back it goes, and where the gaps are. Result: a data readiness report.

02 · Strategy

Model selection per use case

Which models deliver the most at your scale: dynamic pricing, demand forecasting, review sentiment or upsell. A scope with measurable goals.

03 · Build

Modelling sprint by sprint

Every sprint delivers one working model trained on your historical data. Revenue managers test along with us, and the first pricing suggestions emerge within a few sprints.

04 · Rollout

Shadow mode & cutover

First, the model runs in shadow mode alongside your current process. It only goes live once its results are consistently better than the baseline.

05 · Maintenance

Retraining & monitoring

Models quietly degrade as the market changes. We monitor drift, retrain periodically and adjust for new seasons or changing booking patterns.

Work for hospitality organisations.

Hotel · Mid-market chain

Dynamic pricing on Mews

A custom RMS layer on top of Mews for a chain where an enterprise RMS fell short. Demand forecasting by room class, with automatic rate push via SiteMinder.

+ADR
in low-occupancy weeks
shadow mode
before going live
Holiday park · Family-owned

Demand forecasting with local data

A forecasting model that factors in events, school holidays and weather per region. The output feeds into central planning for expansion and staffing.

lower
forecast error vs manual
multiple parks
centrally managed
Restaurant chain · Nationwide

No-show and waste forecasting

Two linked models: no-show classification on reservations and waste prediction on daily menu purchasing. For a chain where a hotel RMS doesn't fit.

lower
food waste per location
12+ locations
in production

The hospitality context we work in.

Phocuswright distribution report 2024

"Mid-market hotels in Europe lag behind enterprise hotels in adopting AI-driven pricing, mainly because of the licence costs of the major RMS packages. Custom solutions are gaining ground."

Hospitality Net trends 2024

"Alongside pricing, the use of AI in review sentiment analysis, predictive maintenance and personalised marketing is growing quickly — often built outside the standard RMS."

CBS accommodation statistics 2024

"The Netherlands has around 3,500 hotels and 2,300 campsites and holiday parks, with OTA dependency in the mid-market exceeding seventy per cent."

Answers for revenue managers and hospitality leadership.

The questions we hear most from hotel chains, park operators and restaurant organisations.

What exactly does revenue management AI do?
In short: a set of models that uses your historical booking data, local demand signals (events, weather, flight schedules) and competitor prices to forecast how much demand there will be per day per room category, and then advises which price, which length-of-stay restriction and which channel mix will work best. The output goes automatically to your channel manager. It doesn't replace a revenue manager: it takes over routine decisions so your team can focus on strategy.
Do you replace IDeaS, Duetto, Atomize or OTA Insight?
No. For enterprise hotel chains that already run these packages, it rarely makes sense. These vendors have years of experience, a broad customer base and validated models, and building your own layer to replace them only pays off if there is a very clear business case. We only come into the picture when licence costs are out of proportion to the scale of the organisation, when there are specific data signals (local events, B2B pipeline, your own loyalty data, multi-property pricing strategy) that the standard package doesn't take into account, or when you run a specific segment, such as a restaurant, glamping site, multi-format holiday park, conference venue or cruise line, for which no ready-made hotel RMS exists. We also build custom for white-label implementations for industry organisations that want to offer their members their own tooling. In all these cases, we build a solution tailored to your data, your operation and your pricing philosophy.
Is automated dynamic pricing actually legal when it comes to consumers?
Yes, provided you are transparent. The EU AI Act sets requirements for automated decision-making affecting consumers: no prohibited inputs, explainability, and guests must be able to find out on request how a price was determined. We build that transparency in as standard: a log of which model recommended which price and based on which variables, plus an explanation component for consumer complaints. Dynamic pricing in itself is entirely legal — airlines, OTAs and almost all chains have been doing it for years.
How do you handle GDPR for guest data?
For revenue models we almost never need to process personal data of individual guests: aggregated booking data per room category per day is enough. For personalisation use cases (upselling, personalised marketing), we work with the minimum dataset necessary, deliver a DPIA and set up retention so that old booking data is anonymised automatically. We work on a privacy-by-design basis from sprint 1, with pseudonymisation at your data warehouse layer and an audit log of who viewed what. If your data protection officer sets additional requirements, such as a specific encryption standard or local data residency, we build towards them within the project.
Can you integrate with our PMS?
In most cases, yes. Mews, Cloudbeds, RoomRaccoon, HotelRunner and Hotelogix have decent APIs and are relatively easy to integrate. Opera (Oracle) takes more work but is possible via OHIP or a middleware layer. We handle Protel and older systems via batch export or a direct database integration if the vendor allows it. For holiday parks, we often see proprietary or semi-proprietary PMS implementations; there too, we build the integration to fit. In the audit phase, we map out the integration options before you commit to anything.
What does a custom revenue AI project cost?
That depends on the scope, the number of use cases and the state of your data. A focused project for a single target group (for example dynamic pricing for one hotel brand on Mews) is more compact than a chain-wide AI layer with dynamic pricing, demand forecasting, no-show and upsell models all at once. We work with fixed sprint budgets and, after the audit phase, give you a concrete scope and timeline for the entire build. A common way of looking at it among our clients: don't compare the expected budget with a one-off cost, but with what an enterprise RMS would cost in licences over three to five years. At mid-market scale, a custom project often fits comfortably within that. Call or email us for an indicative price based on your specific situation.
For which other hospitality use cases do you build AI?
Besides pricing and demand forecasting: review sentiment analysis across TripAdvisor, Booking.com and Google, no-show prediction for restaurants, food waste prediction in the kitchen, upsell engines at check-in, personalised email campaigns via your CRM (Revinate, Cendyn), chatbots for 24/7 multilingual guest enquiries, predictive maintenance on HVAC and lift installations, and complaint classification with automatic routing to the right department. We also work on travel and booking apps and on AI strategy through our AI strategy practice.

Ready to strengthen your revenue strategy with AI?

A half-hour introduction with your revenue manager or management team. We listen to your distribution mix, your data situation and your goals, and give you an honest first direction. No obligation.

Is this topic relevant within your own organisation too? On applatenmaken.com you can read more about building hotel revenue management software.

Edit content