By Konstantin Artemyev

September 24, 2026

Small independent hotels are constantly offered new PMS platforms, new dashboards, new analytics tools and new ways to “understand the business better”. Yet a large part of the information needed to make better decisions has often been sitting inside the hotel’s existing PMS for years.

For me, the ability to work with data is actually one of the important criteria when choosing a PMS. I look not only at whether the room plan is convenient, reservations are easy to create, or the channel manager integration works properly. I also want to know what reports the system can produce and whether the developer is willing to build additional reports when the hotel needs a different view of the data. A PMS contains much of the operational history of a hotel, so using it only as an electronic reservation book seems like a waste.

The problem is usually not that the data does not exist. More often, nobody looks at it systematically, or the numbers are treated simply as reporting: occupancy, average rate, monthly revenue, comparison with last year, and then everyone moves on. For me, analysis really starts only when a number leads to the next question: what are we going to do differently because of this?

Average length of stay is a simple example. Suppose the PMS shows an average stay of 4.5 nights. On its own, that does not tell us very much, because the same average can come from very different booking patterns. But if we look deeper and see stable demand for four-, five- or seven-night stays, especially during high season, that can influence minimum stay restrictions, rate structure and the availability of shorter bookings.

Booking window — how far in advance guests normally reserve — is another figure that can translate very quickly into money.

When I managed an airport hotel, a very large share of reservations arrived only one or two days before check-in. That was completely normal behaviour for that type of property. Flights were booked or changed, connections appeared, business trips were confirmed late, and travellers suddenly needed a night near the airport.

Against that background, the standard recommendation from Booking.com to introduce a last-minute discount made very little sense to me. If a large part of the demand already arrives one or two days before check-in, why give those guests an additional discount at exactly the time when they normally book anyway? The promotion may not create any new demand at all. It may simply make existing demand cheaper.

If rooms genuinely start selling only after the discount appears, then the question is different. Perhaps the base rate is too high. Perhaps the positioning is wrong. Perhaps the issue is the room type, booking conditions or the competitive environment. But the automatic conclusion that “sales are weak two days before arrival, so we need a last-minute rate” can easily become a way of reducing ADR without understanding why.

There are many other examples hidden in PMS data. Cancellations can be analysed by channel and rate plan. A channel may look very strong by gross bookings until you discover that a large share is later cancelled. At that point, it becomes more useful to compare stayed room nights, ADR, commission and cancellation conditions than simply the number of reservations created.

Demand by day of week can also change the picture. In a city hotel, the difference between corporate weekdays and weak weekends may be very clear. A resort may show the opposite pattern. Discounts can then be used where demand is actually weaker, rather than because a promotion happens to be available.

Room type performance can be especially revealing. Which categories sell first? Which remain available until the end? Where is ADR higher? Which rooms are most often upgraded? Sometimes the room category the owner is most proud of turns out to be commercially weaker than the most basic room. In other cases, the most popular category has been underpriced for years simply because nobody has looked carefully at its performance.

I also like to look at length of stay together with distribution channel. Booking.com may generate a large share of reservations, but if those bookings are short and carry a high commission while direct guests stay longer, the channel’s share alone tells us very little about its real value.

Some of the less obvious data can influence the product itself. Guest demographics, if collected properly, can tell us whether the hotel is really serving families, couples or another segment. If families with children represent a large part of demand, then family rooms, extra beds, children’s facilities, menus or seasonal activities may be worth the cost. If families are rare and the hotel performs particularly well with couples, it may be more sensible to stop paying for services that almost nobody uses. Data can tell us not only what to add, but also what we can stop paying for.

The same applies to guest geography, repeat business, arrival and departure patterns, channel mix, cancellations and even the days when housekeeping workload is highest. None of these numbers is especially interesting on its own. It becomes useful when it changes a rate, a restriction, a staffing decision, a service or the way the hotel spends money.

Large chain hotels usually have an easier time with this. They have established reporting methods, revenue management structures, regional or central analytics and employees who are trained to work with data. An independent hotel is often in a very different position. The general manager is already overloaded with operations, there may be no revenue manager or analyst, and even good PMS reports require time and experience before they start producing useful conclusions.

This is one reason why hotel data is so often underused. Proper analysis is routine work, and sometimes quite a lot of it. Data has to be exported, periods compared, demand broken down by segment, channel, booking window and length of stay, unusual changes identified, and only then do the interesting questions begin. In a small hotel, this work is very easy to postpone because there is always something more urgent happening today.

This is also one area where modern AI tools can be genuinely useful. Not because artificial intelligence somehow knows how a particular hotel should be managed, but because it can remove a large part of the routine analytical workload. It can help organise PMS exports, compare periods, find patterns, identify unusual changes and suggest questions that deserve a closer look.

The limitation is the same as with any analytical tool. If the data is badly structured, segments and rate plans are named inconsistently, or nobody knows what they are actually trying to understand, even the most advanced system will not produce much value. AI can make analysis much easier, but the decisions still belong to people who understand the economics of the hotel.

This is also where replacing the PMS can become a legitimate question. I would never change a system simply because something newer and more attractive has appeared on the market. Moving from one PMS to another is painful, and if the existing system does its job properly, there is little reason to create a revolution.

But if a hotel wants to manage pricing, costs and demand seriously while its PMS is an old system from the early 2000s that cannot provide useful data, or a cheap cloud product that is little more than a room calendar with a few basic reports, then the system itself may be starting to limit the business. The same is true when the developer cannot provide the reports, API access or even the data exports the hotel needs.

At that point, the question is no longer whether the PMS looks modern or old-fashioned. The question is whether it allows you to see what is happening in the hotel today, understand why it is happening, and make better decisions about tomorrow. If it does not, replacing it is at least worth considering.

This is why I am generally sceptical of the idea that a small hotel needs another analytics product before it can start managing the business better. Perhaps it does. But first I would look at what the existing PMS already knows — and whether the system is capable of telling us properly.

Very often, the data has been there for years. What is missing is not more data, but the time, habit or competence to ask the next question: what are we going to do differently now?