The hotel commercial glossary: seven terms your AI tools cannot read

Share on

Sit in a commercial meeting at any hotel group and you will hear four vocabularies in one room.

Marketing talks in clicks, cost per acquisition and return on ad spend. Revenue talks in pace, pickup and RevPAR. Distribution talks in channel mix and commission. The owner listens to all three and tries to work out whether the property made money.

Everyone is fluent. Nobody is speaking the same language.

The seven terms below are the ones that actually decide margin. Each one is defined here in plain English, along with what it quietly hides. Most AI tools for hotels can read a click. Very few can read a stay date, and that gap explains why so much of the current wave of AI in hospitality produces confident output that nobody can act on.

Why the vocabulary decides the decision

Retail measurement assumes a click and a purchase in the same session. Hotels break that assumption at every point, and three things are making the gap expensive.

Demand is getting harder to win.

Deloitte's 2026 Travel Industry Outlook reports rising financial caution across all income levels. Frequent corporate travellers expecting three or more trips a month fell to 53%, from 63% the year before.

Less demand behind every room-night means less room to absorb a bad decision.

Discovery is moving somewhere hotels barely appear.

Lighthouse research, reported at Skift's 2026 Data and AI Summit, finds 82% of AI hotel recommendations draw on online travel agents and editorial media.

The channel deciding where guests look is the one hotels have least presence in.

So groups reached for AI, and most of it stalled.

McKinsey found nearly two-thirds of enterprises have experimented with AI agents, and fewer than one in ten have scaled them into tangible value. Eight in ten name data limitations as the roadblock.

The reason is specific to hospitality.

A general-purpose tool pointed at a hotel cannot read the vocabulary the business runs on. It sees a click and a conversion. It does not see a stay date, a booking window, or a room-night that could have been sold twice.

That is a vocabulary problem before it is a technology problem, and we set out the mechanics in AI won't fix hotel marketing until your data is connected.

The seven terms

Each term below follows the same four lines: what it is, how it is measured, what it hides, and why it matters to margin.

1. Pace

• What is pace: How quickly rooms are booking for a future stay date, judged against a comparable point last year. The word carries two meanings. Pace as position is where the book stands today. Pace as velocity is how fast it is filling. A property can be ahead on one and behind on the other.

• What pace hides: A November running 6% ahead on rooms can carry flat direct bookings and a 20% rise in online travel agent bookings. Volume is ahead and net rate per room-night is down, because the extra rooms carry commission the direct rooms did not.

• Why pace matters: Pace is the first number quoted in most commercial meetings and the one most likely to close the conversation early. A green pace report is entirely compatible with falling margin.

2. Pickup

• What is pickup: Rooms added for a future stay date across a defined interval, usually the last seven days. It shows what changed, where pace shows where the book stands.
  

• ‍What pickup hides: Which demand source produced the movement, and what it cost. A date picking up 40 rooms while absorbing 12 cancellations has net pickup of 28. If the 40 came through paid channels and the 12 that cancelled were direct, you paid to acquire 40, kept 28, and lost 12 that cost nothing.
  

• Why pickup matters: Pickup is the figure most used to judge whether a campaign is working. Two identical numbers can carry completely different acquisition costs, so the figure alone cannot support a spend decision.

3. On the books

• What is on the books: Confirmed rooms and revenue already held for a future stay date, shortened to OTB. The closest thing a hotel has to a forward order book, with the difference that much of it can disappear before arrival.
  

• ‍What on the book hides: Cancellation and group wash exposure. Transient bookings are individually cancellable, group blocks may still be tentative, non-refundable bookings will hold, and all three land in one total. A December date showing 180 rooms against 200 available reads as nearly sold. If 60 are an unsigned block on an account that washes at 20%, realistic exposure is closer to 132.
  

• Why on the book matters: Rate and spend decisions taken against OTB are taken against a number that has not settled. Closing availability against unsettled inventory is an expensive routine error, and it stays invisible until arrival.

4. Booking window

• What is booking window: The lead time between a guest booking and staying. It determines whether your measurement tools can see your own bookings.
  

• What booking window hides: The mismatch with the attribution window on your advertising accounts. Platforms credit conversions inside a click window, commonly 7 or 30 days. Where the booking window runs longer, the platform never connects the advertisement to the booking. A resort with a 45-day median window on a 7-day click window loses the guest who watched in July, booked in September and stayed in October.
  

• Why booking window matters: This mismatch is the most common reason marketing and revenue produce irreconcilable versions of the same month, and budget then gets cut from the channel that created the demand. We set out the mechanism in marketing attribution for hotels: which model to trust and when.

5. Length of stay

• What is length of stay: Nights held by a single reservation, reported as average length of stay, or ALOS. It reads as a descriptive statistic and functions as a control, since minimum stay rules, closed to arrival and closed to departure are all length of stay levers.
  

• What length of stay hides: Displacement across the stay pattern. A Friday selling out at a high rate on one-night stays looks excellent alone. The four-night booking arriving Thursday, refused because Friday showed no availability, appears nowhere. One high-rate night protected, three moderate-rate nights given up.
  

• Why length of stay matters: Inventory is fixed, so every sale is a substitution rather than an addition. The highest rate on a single night is regularly not the best outcome across the pattern, and no nightly report will tell you when.

6. Rate parity

• What rate parity is: Consistency of your published rate for the same room, date and conditions across every channel a guest can find you on. In most groups it runs as a compliance task, scored on violations closed.

• ‍What rate parity hides: The margin behind each channel. Two bookings for the same room, the same night, at the same published rate produce different outcomes once commission, payment processing, incentive tiers and cancellation propensity are counted. Parity can be perfect while margin differs materially.

• Why rate parity matters: Parity keeps the shop window honest and protects direct from being undercut. It does not tell you which window is worth selling through. Most properties only have an answer to the first question.

7. Incremental margin

• What is incremental margin: What a booking adds after the cost of acquiring it, counting only demand that would not have arrived anyway. The one term here that is a decision unit rather than a measurement, and the only one that can rank two channels honestly.
  

• ‍What incremental margin hides: Nothing, provided the test was designed properly. It must outlast the booking window, since two weeks against a 45-day window measures demand already generated. And it must run in parallel across matched markets, since suppressing in March and comparing with February measures seasonality. Where a market carries too much of the house to switch off, state the figure as commercial judgement rather than measurement.
  

• Why incremental margin matters: It is the number a budget decision needs, and the number almost no hotel has. Without it, marketing, revenue and distribution compare figures never denominated in the same currency, which is how all three hit target in a month the group earns less per room-night than the year before.

The three cost numbers sitting behind every term

A cancellation reverses the revenue. It does not reverse the media cost that bought it. Online travel agent commission does reverse in full.

Net cancellations from both sides and you overstate paid performance every month.

Why generic AI tools stumble on this vocabulary

Four structural reasons, none fixed by a better model.

• Wrong calendar. Marketing reports by click date, revenue by stay date.

• Wrong event. Ad platforms count a confirmation. The property counts an arrival and a settled amount.

• Filters instead of objects. Stay date and length of stay get applied after the analysis. They belong at the centre of it.

• No shared unit. Connecting the data still leaves three arguments in a nicer interface.

That is the gap between a measurement and something a general manager can act on, which we set out in every hotel dashboard shows metrics, none shows opportunity.

How pulse. reads these terms

pulse. by dhi Hospitality is built on hospitality primitives rather than retail ones.

• Stay date, booking window, length of stay, source market, parity and cancellation behaviour are first-class objects.

• Every demand source is expressed in one currency: incremental net margin per demand cell.

• Outcomes are reconciled to the property record and measured at settled revenue.

• Every figure carries its origin. Measured, judgement, or fitted and therefore direction only.

• Every recommendation is approval-gated. Nothing moves without an explicit click.

• pulse. reads the systems you already run. Nothing is replaced.

One honest limit at launch. Where a cost figure rests on judgement rather than a controlled test, guidance is marked direction only. Direction is useful. False precision is expensive.

Apply that standard to any hospitality digital strategy tool you are shown.

Five questions for your next commercial meeting

1. Pace is ahead. Is margin per room-night ahead with it?

2. What share of on the books is exposed to cancellation?

3. Which demand source produced last week's pickup, and what did it cost?

4. Would those bookings have arrived anyway?

5. Are we comparing channels in the same unit?

If the room cannot answer question five, the first four cannot be settled either.

pulse. launches on 30 September. Get early access or book a demo to see it running against your own data.

The vocabulary was never the problem. Running four of them at once is.