Hotels and restaurants, even correct data can lead to wrong decisions

From data to decision

In my experience in the Horeca sector, I've seen companies with a perfect understanding of their overall revenue, yet lacking a sufficiently precise understanding of which services generate value. A restaurant can increase seating capacity and reduce margins. A hotel can improve occupancy, but become increasingly dependent on platforms. A promotion can generate bookings without building a truly profitable customer base. The apparent result is positive; the financial impact can be much less so.

The useful question, therefore, is not simply how much we've sold. We need to ask ourselves through which channel, at what acquisition cost, with what margin, over what period, and with what reference point. It is this step that transforms management data into a decision-making tool.

Averages can hide the problem

Aggregate information is essential, but it often only tells a part of the story. An increase in average spending may be due to a few high-spending guests. An increase in guest numbers may coexist with shorter stays. A good monthly result may hide a loss of profitability in a specific channel or service. To understand what's really happening, we need to separate at least a few dimensions: Italian and international guests, new and repeat guests, direct and intermediate bookings, revenue from rooms, food and beverages, and ancillary services, and high and low season.

The same is true in the restaurant industry. The average cost of raw materials isn't enough if it isn't linked to individual preparations, waste, menu rotation, labor times, and actual sales. Average figures are reassuring; contextualized figures can force a change in decision.

Prediction and outcome are not the same thing

One of the most common mistakes is confusing what is expected with what has been achieved. An investment may indicate estimated savings, but the result only exists after measurement. An artificial intelligence project may promise greater efficiency, but purchasing the technology does not automatically demonstrate its usefulness. It's best to distinguish between four stages: planned, financed, implemented, and verified. This separation avoids presenting intentions or forecasts as consolidated results and allows us to compare the initial objective with what has actually been achieved.

A dashboard does not necessarily retain evidence

Dashboards are valuable because they reduce complexity and make a wealth of information easily readable. However, this very summary can hide essential elements: data source, calculation criteria, exclusions, updates, and responsibilities. For each indicator used in an important decision, it should be possible to reconstruct at least the source, period, definition, any changes, the person responsible, and the related decision. Without these elements, it becomes difficult to determine whether a change is due to the market, a different calculation method, or a collection error. This does not necessarily require extensive infrastructure. Even an independent organization can start with just a few indicators, as long as they are stably defined and linked to specific operational questions.

Artificial intelligence does not eliminate responsibility

Artificial intelligence can analyze large amounts of information, identify patterns, and make recommendations. However, it cannot automatically replace the operational knowledge of a facility's managers. If a system suggests changing a price, reducing a service, or targeting a specific audience, it is necessary to know what data it used, how up-to-date it is, which variables have been excluded, who can challenge the recommendation, and how the outcome will be measured. Human oversight should not be limited to the ability to approve a system's proposals. It must retain real authority: to understand, modify, reject, or abort the decision.

A few questions asked well

Not all businesses need extensive analytics systems. Often, the first step is to define a few questions and associate each with a reliable source. Which customers generate value over time? Which channels guarantee the best real margin? Where does the booking journey end? Which costs grow faster than revenue? What decision was made, and what outcome did it produce? If it's not connected to a question, data risks becoming mere noise. If it doesn't retain its source, it's unverifiable. If it doesn't lead to a decision, it's unlikely to generate value.

Technology in hospitality will continue to evolve. Decision-making responsibility will remain in the hands of those who must transform information, people, and resources into a sustainable, financially sustainable experience. Data doesn't make decisions on its own, but it can help us make better decisions, provided it remains understandable, contextualized, and verifiable.