Your Hotel Has Forty Products. The Website Sells Five.

Walk into the back office of any 120-room hotel and ask the front office manager how many genuinely different rooms the property sells. Let them think for a moment. They will name the corner suite with the silent garden. The room above the bar that nobody loves on a Friday night. The two adjoining doubles that families fight over for every school break. The high floor with the rooftop view. The accessible room with the larger bathroom that couples quietly prefer. By the time the team finishes, you will have counted thirty or forty distinguishable products in their heads. Now look at the website. Five categories. Maybe six. The most differentiated, most expensive, most emotionally loaded asset in the entire business is being sold under labels that erase the things guests would actually pay for.

I have done this exercise with hoteliers in Hamburg, in Lisbon, in the Alps, in coastal Tuscany. The numbers vary. The pattern does not.

That is the real distribution problem of our industry, and it has been hiding in plain sight for thirty years. I have sat through more keynotes on personalisation, loyalty, AI agents, and immersive booking journeys than I care to count. Almost none touch the inventory model underneath. That model was designed in the 1970s for a green-screen terminal, and we are still living inside it. Until the foundation is rebuilt, no amount of new technology in the pre, mid, or post-stay layers will deliver the experience guests now expect.

The customer journey, when you strip away the conference jargon, has three moments that matter. The moment the guest decides. The moment the hotel actually delivers. The moment the memory forms and either pulls the guest back or quietly does not. All three are held back by the same inheritance. We are still selling ingredients in an industry that has been asked to deliver products.

Where the Guest Decides

It starts long before anyone walks into reception. The guest is typing into a metasearch box, scrolling through Instagram, asking a chatbot, or whispering a request to a voice assistant. None of them are searching for “a hotel in Paris” anymore. They are searching for something specific. Quiet. South-facing. A workspace with a real desk. A bath as well as a shower. Walking distance from a particular bakery. The questions are detailed, contextual, and often emotional.

Our standard answer to those questions remains a list of room categories that say almost nothing. A category is a container, originally invented so a property management system could organise housekeeping rotations and rate plans. It made inventory legible to back-office staff and to the first wave of online travel agencies. Over time we mistook the container for the product.

Our industry has been talking about a fix for almost a decade. Attribute-Based Selling, usually shortened to ABS, has dominated conference keynotes since at least 2017. The promise has always been straightforward: let guests filter for a higher floor, a quieter side, a walk-in shower, a morning balcony, rather than pick between Standard and Superior. The label gets richer. Mapping each unit’s real characteristics becomes visible inside the booking flow (Stegmayer, 2021). I have been on the optimistic side of that conversation the whole time.

What I have come to admit is that ABS has barely been built where it matters most. The chains that announced pilots in 2018 are still announcing pilots. Where ABS does show up in production, it is almost always bolted onto post-booking upselling. The guest is offered the chance to pay extra for a balcony or a higher floor after committing to a Standard Double. The point of sale itself, the moment that decides whether the guest books at all, still presents the same five tiles it presented twenty years ago. And even in its purest form, ABS stops at the ingredient. It describes the room. It does not yet design the product.

That sharpens the argument considerably. The hotels that move next will skip past the half-built ABS layer altogether and design real products at the moment of sale. Each physical unit becomes the source of multiple offerings, each defined by features that a real human values, each with its own narrative, its own price logic, and its own channel placement. Same ingredients, different products. The same room can become a quiet retreat for a writer on deadline, a compact city base for a midweek consultant, or, paired with the adjacent twin, a family reunion suite. Customers everywhere else in commerce buy products. Hospitality remains one of the very few categories that still asks them to buy ingredients.

There is a name for this approach now. Some of us have started calling it "Dynamic Inventory": the practice of treating each physical unit as the source of multiple bookable products, each with its own name, story, features, and price (Mueller, 2026). The first hotels operating this way already report measurable uplifts in direct conversion and in the share of reservations the system can complete without a human in the loop. This is not theory anymore.

McKinsey reports that 71% of consumers now expect personalised interactions, and 76% feel frustrated when they do not receive them (McKinsey & Company, 2023). Search engines and AI travel agents have something concrete to match against once products exist. A query for a hotel in Lisbon with a real workspace and a south-facing balcony then returns a real product, with a name, a story, and a price that reflects what those features are worth (GauVendi, 2025). Phocuswright finds that direct bookings continue to grow primarily for properties that give travellers a reason to switch (Phocuswright, 2024). A genuinely different product is a reason. A two-euro discount on the same shelf item is not.

Where the Hotel Delivers

What the website promised is either honoured or quietly broken between check-in and check-out. Every operator I speak to wants the moment of arrival to feel effortless and the stay itself to feel personal. Very few are set up to deliver it.

The reason is the inheritance again, this time wearing operational clothes. I have stopped finding it funny. When a hotel sells categories, it postpones the actual decision of which physical room a guest will receive. Assignment is made the morning of arrival, often by a tired front office team balancing occupancy, late check-outs, group movements, and engineering blocks. The guest who chose the property because of the courtyard view ends up in a street-facing room with double glazing that almost works. The guest who wanted the high floor is downstairs because the lift is being serviced. Operations did its best with the inventory model it inherited.

There is a more honest way to think about assignment. Treat it as a continuous job across the booking lifecycle, rather than a single event at check-in. Every new reservation should be allowed to trigger a reshuffle. Specific feature requests should be honoured before generic ones. The objective can shift between revenue, occupancy, and guest experience depending on what the property needs that week. None of this requires generative AI. It requires an operating system that treats inventory as fluid. I keep being told this is the hard part. In my experience the harder part is admitting the model we inherited has quietly become the bottleneck.

Orchestration has a second dimension. The mid-stay is also where the ingredients of an experience get assembled. A pillow menu, a late lunch, a yoga mat, an early breakfast, a quieter table at dinner. Each is a feature, with a value the guest will often pay for if asked at the right moment. Cornell research has shown for years that ancillary revenue per guest grows when offers are contextual rather than generic (Anderson, 2017). The technology already exists. The bottleneck, once again, is the underlying data model. If we have only ever sold categories, we have only ever known categories.

A property that captures feature-level signals turns itself into a learning system. By checkout, the hotel knows more about that traveller than the OTA does, and the knowledge belongs to the hotel.

Where the Memory Forms

Memory is selective. Guests forget the rate they paid within days of departure. They remember whether the experience matched the promise. Bain reports that emotionally engaged customers are between two and four times more valuable than merely satisfied ones (Bain & Company, 2022). A stay that delivered exactly what was sold, down to the feature, lodges in memory as a “right place at the right time” experience. A stay that fell short, even slightly, leaves a quiet reservation that surfaces the next time the guest is choosing where to go.

Loyalty, as a concept, has been over-engineered and under-delivered. Most programmes are discount mechanisms dressed in tier names. They reward frequency. They rarely reward fit. Anyone who has worked the front desk on a wet Tuesday in November knows the difference. The evidence is clear that emotional loyalty, the kind that drives unprompted recommendation, comes from feeling understood (Reichheld, Darnell and Burns, 2021). A guest who once booked the quiet garden room does not need a generic “we miss you” email twelve months later.

They need a note that says the same room is available next month, the new chef has added a vegetarian tasting menu, and the bakery they liked has just opened a second location around the corner. Re-engagement built on feature memory is more useful than re-engagement built on stay history alone.

Another quieter benefit. Every booking made against a feature-defined product produces a new class of data. It tells the hotel what was valued, by whom, in what context, and at what price. Categories destroy this signal because every guest is funnelled through the same label. Products preserve it. Over a year, an independent property accumulates buying intelligence that no chain headquarters or OTA can easily replicate. That dataset feeds the next round of product creation. Pricing sharpens. Marketing sharpens. The loop compounds.

Conclusion

The deciding, the delivering, and the remembering are usually treated as three separate disciplines, each with its own software, its own KPI, and its own conference track. They share a single foundation. If that foundation collapses dozens of unique units into a handful of categories, every downstream layer will struggle to feel personal, no matter how sophisticated the AI sitting on top of it.

Hospitality has always been an industry of detail. A particular light through a particular window in the late afternoon. A croissant from a specific oven. A welcome that lands because someone read the file. The technology stack we have used for three decades was never designed to honour those details. The technology stack we are now building can be. ABS opened the door. The next move is to walk through it and start designing real products.

I keep coming back to the same thought, and I will leave you with it. The room category was never the product. It was a workaround we forgot to replace. Once we admit that, the rest of the stack starts to make sense.