Discovery 27 July 2026, 6 min read
Website metrics that actually matter in hospitality
Most hospitality reporting measures things that cannot be acted on. Here are the few numbers that change a decision, and what to do when each one moves.
The monthly report from most agencies opens with sessions, then bounce rate, then a list of the top ten pages. All three are real numbers and none of them tell you whether to do anything. Sessions go up in August because it is August. Bounce rate goes down when the site gets slower in a particular way, which is the opposite of what you would want it to mean.
A hotel or restaurant has a small number of numbers worth watching, and each one has an action attached to it. That last part is the test: if no plausible movement in a metric would change what you do next week, stop reporting it.
Direct share of room revenue
The number that matters most, and it does not come from the website analytics at all. It comes from the property management system: what proportion of room revenue arrived through your own channels rather than through a platform or an agent.
Watch it monthly against the same month last year, never against last month, because seasonality will otherwise tell you a story that is not there. If it moves up two points and stays there, whatever you changed worked. If it drops while total revenue holds, you have not lost business, you have started renting it, and that is the moment to look at what changed on the site or in the rate structure.
Booking journey drop-off, step by step
Overall conversion rate is close to useless on its own, because it averages together the person idly browsing in February and the person with a card in their hand. What you want is where people stop.
Reaching the availability check, seeing rates, choosing a room, entering details, completing. Five numbers. The step with the largest fall is your entire to-do list. A big drop between seeing rates and choosing a room usually means the rates surprised someone, or the room descriptions are not doing their job. A drop at entering details is almost always the form asking for too much, or the moment where the design changes and the guest feels handed to a stranger.
Brand search against category search
In Search Console, split the people who searched for your name from the people who searched for a category, such as hotels in a region or somewhere to eat near a town.
Brand search rising means your reputation, press, and the OTAs are doing their job of making people aware of you. Category search rising means the site itself is being found by people who did not know you existed, which is the harder and more valuable growth. If brand search is high and direct bookings are not, you have a website problem rather than a marketing problem: people are looking for you specifically and still not booking.
Mobile speed, on the pages that matter
Not a single site-wide score. The load time of the home page, the room or menu pages, and the first step of the booking journey, measured on a phone connection rather than an office one.
The threshold that matters in practice is around two and a half seconds to the main content appearing. Beyond about four, losses become obvious in the drop-off numbers above. It is worth checking after every content update, because the usual cause of a site getting slower is somebody uploading a photograph straight from a camera.
Enquiries that turn into something
For the parts of the business that are not instant bookings, weddings, events, private dining, large parties, the metric is not enquiry count. It is the proportion of enquiries that become confirmed business, and how long it takes.
Twenty enquiries a month with a five per cent conversion is a worse outcome than eight with forty, and it usually means the site is attracting the wrong people, most often by being vague about price. A page that states a realistic starting point turns away the people who were never going to book and lets your team spend their attention on the ones who will.
Returning guests
From the property management system rather than the website: the share of stays from someone who has stayed before. It is the cheapest revenue you have, it is invisible in almost every marketing report, and it is the number that tells you whether the place is genuinely working.
The numbers to stop reporting
Worth saying plainly, because they take up most of the space in most reports.
Sessions. Rises in August, falls in January, and tells you nothing you did not already know from the diary. Useful only as a denominator.
Bounce rate. Widely misunderstood and, since the measurement changed, close to meaningless on its own. Someone who arrives, reads the opening hours, learns you are open on Sunday and leaves satisfied is a bounce. So is someone who arrives, cannot find anything and leaves furious. The number cannot tell them apart.
Average time on page. A long time on the menu is usually good. A long time on the booking form is very bad. Averaged across a site it is noise.
Keyword rankings as a list. Positions vary by person, place and device, so a table of ranks is a snapshot of one machine's opinion. Impressions and clicks in Search Console are the real thing and are already free.
Social followers. A real number about a real audience, and almost never connected to whether rooms were sold. Worth watching as its own thing. Not worth putting in a report about the website.
Setting it up so the numbers are true
Half of hospitality analytics is wrong before anyone reads it, usually for one of three reasons.
The booking engine sits on a different domain, so the visit gets counted as two people and the booking is attributed to the engine rather than to whatever brought the guest in. This is the big one, and it is fixable in an afternoon by whoever maintains the site.
The team's own visits are counted, which for a small property with a reception computer can be a noticeable share of everything. And bookings are recorded without value, so the report shows twelve conversions with no idea whether they were a Tuesday single or a fortnight in the suite.
Get those three right and the rest of this becomes worth reading. Leave them and you have a monthly report about a fictional hotel.
What to look at, and when
Once a month is enough for direct share, brand against category search, enquiry conversion and returning guests. Once a quarter for mobile speed unless something has changed. Booking journey drop-off is the exception, worth looking at whenever anything in the journey is altered, because that is where a small mistake does the most damage fastest.
Everything else is context. Useful for understanding what happened, rarely a reason to act.