.svg){kind=small}
.webp)
Marketing attribution for hotels: which model to trust and when
Your Google Ads account reports 140 bookings this month. Meta reports 90. Your PMS shows 160 rooms actually sold. Every platform is counting the same guest, none of them agrees, and Monday's budget conversation will be decided by whichever screenshot gets opened first.
Settling that argument will not help. Even a perfectly reconciled attribution report cannot tell you what a room-night costs to fill through each channel, which is the only number a budget decision needs.
Marketing attribution assigns credit for a booking across the touchpoints before it. It answers one question: which spend appeared on the path to this revenue.
Two things make hotels harder than retail. The booking window: a guest sees an ad in July, books in September, stays in October, and most attribution windows have closed. The revenue event: the ad platform records a confirmation. Your PMS records the stay, the cancellation and the settled amount.
So marketing reports growth, revenue reports flat pace, and neither report is wrong.
Last-click is the default in most hotel dashboards. Brand search wins every time, because a guest who already decided to book types your hotel name into Google. You pay to capture demand you created, then cut the channel that created it.
Data-driven models are trained inside the platform selling you the media. Google grades Google, Meta grades Meta, and added together they overstate performance.
All five are reporting instruments. None was built to say what should happen next.
Cross channel marketing attribution is the attempt to see one guest journey across every platform at once. In hotels it fails at six points.
These are structural, not tracking bugs, which is why data fragmentation keeps independent hotels below 10% direct share.
Well: which campaigns appear on booking paths most often, which channel a guest touched last, and how performance shifts when you change creative or rate.
Badly or not at all: whether a booking would have happened anyway, which channel drove a booking that completed on an OTA, and whether spend on a date matched the occupancy need on that date.
The second list is the one that decides budgets.
There are three cost numbers behind every channel. Most hotels have one, use it for everything, and are wrong in a predictable direction.
Two consequences follow.
Organic is not free, and you cannot buy more of it. It has a volume ceiling, not a cost curve. Once you have taken the demand that was going to find you anyway, there is no more at that price. Treat it as cheap and it wins every comparison it enters.
Cancellation is asymmetric. A cancelled booking reverses the revenue. It does not reverse the media cost that bought it. OTA commission does reverse in full. Any model netting cancellations from both sides overstates paid performance every month.
Incrementality cannot be observed. It has to be created, which means suppression: paid switched off in one set of matched markets, left running in another, and the difference read.
Two conditions decide whether the result is worth anything. It has to outlast your booking window, because if your median lead time is 45 days a two-week test only measures demand you had already generated. And it has to run in parallel. Suppress in March, compare with February, and you have measured seasonality.
Most properties cannot suppress every market. If one carries half the house it stays on, and the honest position is to state that number as a judgement, not a measurement.
Touchpoints 1 and 6 matter commercially, and they sit in systems that never speak. Everything between is proxy measurement, which is why Meta attribution understates what Meta creates.
Standard tools were built for e-commerce, where click and purchase sit in one session and revenue is final at checkout. Hotels break all three assumptions, but fixing that only answers a reporting question more precisely. The commercial question stays open. Marketing optimises ROAS. Revenue optimises RevPAR. Distribution optimises channel mix. All three hit target in the month net margin per room-night falls, because nobody compares what a room-night costs to fill through each route.
That is a missing denominator, not a missing report.
pulse. puts every demand source on one number: incremental net margin per demand cell, where a cell is a property, demand source, market and stay month.
What changes is the conversation, not the dashboard. Brand search stops being flattered by its position at the end of the path. Organic stops looking free. The revenue leak was never inside a single tool. It sits in the gap between them, and it is denominated in margin.
pulse. opens 30 September. Get early access and join the first properties running on one commercial brain.
.webp)
.webp)