Hotels are susceptible to anomalies since service delivery always should be perfect. Therefore, we spend vast amounts of time checking and double-checking to ensure that we deliver what we have promised. What if a system could automate these endless checks, and once the system finds an anomaly, it sends an alert to the person that can prevent or fix the problem?
In data analysis, anomaly detection identifies rare items, events, or observations that raise suspicions by differing significantly from the majority of the data. Thus, anomaly detection is what a system does much better than a human being, easing overworked managers' burdens.
Systems can process vast amounts of data in seconds and instantly detect an anomaly. The same work would take hours or days for a hotel manager. If the system presents what it found, the hotel manager can immediately use human judgment to assess the potential impact and take appropriate action. Why is this logic not used to its fullest extent in hotels? The immediate answer is lack of knowledge, accessible data, and definitions of hotel anomalies. Systems that can detect anomalies have been available in other industries for many years.
Modern systems such as the new fully and real cloud-based hotel PMS will replace the old legacy systems and let hoteliers access data through open APIs. Easily accessible data is the first step in setting up robust anomaly detection systems that alert the general manager when something is off-limits. The second step is to define hotel anomalies, where things get a bit more complicated.
Hotels use "standard operating procedures" (SOP), so the straightforward definition of an anomaly is a deviation from the SOP. Here is one example. The hotel would like to get feedback from every guest and therefore collects the NPS score after check-out. The response rate is typically around 30 %, and the average NPS score is 65 %. The hotel has an SOP to send a follow-up email to each guest one day after the stay, ask for the NPS score, and check the actual score. Based on this information, three significant potential anomalies could occur.
The NPS is a leading indicator where a high NPS will attract more guests in the future, and a low NPS will drive potential guests to the competition. The guest experience delivery is vital in creating the overall hotel reputation, so the NPS score is vitally important to keep track of for general managers in hotels.
The general manager wants to know if there is any problem collecting the data. If the hotel does not send the email, the system automatically alerts the general manager. However, too many alerts do not make sense, so it is better to set some boundaries, such as if the hotel sends the email to less than 80 % of guests, the system alert the general manager.
The next part is the response rate. The system alerts the general manager if the response rate is less than 25 % or higher than 40 %. Higher is good, but a significant deviation could indicate a problem.
The final part is the score. The system alerts the general manager if the score is below 55 % or above 75 %. Alerts also have to be positive to inspire and confirm success.
In this case, no alerts mean that everything is normal and there is no need for urgent actions. Instead, the general manager can focus on other more pressing issues.
Finally, when setting the boundaries, it is crucial to calibrate the system to keep the number of alerts to a minimum and only send alerts when the impact of an action is significant. For example, the hotel has calculated that if the NPS score falls below 55 %, the average rate starts to drop by USD 10 per percentage point. The impact for a 100 room hotel with 75 % annual occupancy would be USD 270 000. The general manager needs to assess the risk of losing such a high amount before the drop in the NPS score accelerates.
A well-defined alert system would ease the stress from general managers and let them focus on those actions that have the highest impact on the profits.