The following morning, I opened again the review I had left unresolved. Nothing had changed since the night before. The guest still described arriving outside the usual hours and finding it difficult to coordinate entry; the hotel’s policy still explained that, with advance notice, an arrival outside the usual hours could be arranged; and the other reviews we had found still described experiences compatible with that possibility. All the pieces remained exactly where we had left them. The only thing that was no longer quite so simple was continuing to look at them as though they necessarily told a single story.
For a few minutes, I tried to decide which one I should believe. The question seemed reasonable, but the more I asked it, the less useful it became. Nothing required us to assume that any of those signals was false. The hotel might have successfully received several guests outside the usual hours and encountered difficulties with another; the policy might exist and, at the same time, not always be carried out in the same way. Even the negative experience could have happened exactly as described without demonstrating that the property had ceased to be flexible. Perhaps the contradiction was not between incompatible facts, but between different outcomes arising from the same intention.
Until then, we had used consistency between signals as a way of increasing our confidence. When policies, decisions and independent experiences led towards a similar conclusion, we needed to add less of our own in order to sustain it. Without realising it, that logic had created another expectation: if the identity was consistent, we expected every experience it produced to be consistent as well.
In theory, that seemed reasonable. If the logic of first understanding the person and then deciding how to look after them genuinely formed part of the hotel’s way of working, we would expect to find traces of that logic repeatedly. Yet a property was not a machine executing exactly the same response every time it encountered a similar situation. There were different people, different shifts, varying levels of occupancy, requests that bore little resemblance to one another, unforeseen events, tiredness, mistakes and circumstances no policy could anticipate completely.
Perhaps, then, the question was not whether the hotel managed to sustain that characteristic every time, but how often, under what conditions and to what extent it was able to do so.
I returned to the positive reviews about late arrivals. Until then, I had read them mainly for what they had in common; this time, I looked for the differences. One concerned a couple who had given several hours’ notice. Another mentioned that the property knew their flight time. In a third, the guest explained that they had remained in contact with reception during the journey. The negative experience, by contrast, said little about what had happened before arrival. It did not say whether advance notice had been given, what time the guest had tried to make contact or what specific difficulty had prevented the arrival from being coordinated.
It was tempting to use those absences to explain the problem. If the guest had not given notice, the contradiction virtually disappeared; if they had arrived much later than agreed, perhaps the policy had never promised to resolve that situation; if the hotel had not received the message, it might even have been a communication failure rather than a lack of flexibility.
All were perfectly possible explanations. None, however, was contained in the information we had.
The same risk we had tried to avoid throughout the previous conversations was returning in a much subtler form: filling what we did not know with the explanation that best protected the conclusion we already preferred. We did not need to invent an extraordinary attribute or ascribe to the hotel a capability it did not possess; it was enough to fill a few gaps plausibly.
But a plausible explanation was still not knowledge.
So I left the review as it was: a recent experience, clear in its outcome and still incomplete in its causes. I did not need to turn it immediately into evidence that the hotel was less flexible than we thought, but nor could I use what we did not know to diminish its importance. There were good reasons to believe that the property could adapt to certain late arrivals and, at the same time, there was at least one recent situation in which that capability had not produced the expected outcome.
The contradiction remained, but it was beginning to serve a different purpose: rather than destroying our interpretation, it was beginning to point towards the part of it we did not yet understand.
I decided to examine other attributes for the same phenomenon and did not have to look far. Tranquillity, one of the best-supported characteristics, coexisted with some reviews mentioning noise. Personalised attention had numerous concrete examples and yet one guest described an interaction perceived as distant. Breakfast received frequent praise while another person considered its variety limited. Even rooms that most guests described as comfortable were occasionally described as small.
If every difference functioned as a refutation, we would end up concluding that we could know practically nothing about the property. But if we treated every discordant comment as an irrelevant exception, we would construct an artificially orderly version of the hotel. Neither reading seemed capable of explaining what was actually happening.
It was a comment about noise that allowed me to move forward. Two guests could stay during the same week and have completely different experiences of noise simply because one had been given a room facing the garden and the other one close to a circulation area. The hotel is quiet might be too broad a formulation, not because it was necessarily false, but because it concealed conditions that affected whether it held true.
The late arrival then began to seem like a similar problem. Perhaps we did not need to decide whether the hotel was flexible or not, but to understand when it managed to sustain that flexibility and what circumstances made it more likely to do so.
What we called identity was therefore losing some of the uniformity we had unconsciously demanded of it, but it was not becoming weaker as a result. In a sense, it was becoming more real.
The logic of first understanding the person and then deciding how to look after them still seemed a good description of what we had found, although I could no longer treat it as a guarantee of outcome. It was an orientation capable of organising decisions and producing behaviours with enough recurrence to become recognisable. Between that orientation and each concrete experience, however, stood an operation that had to turn it into reality, and that execution could vary or fail.
Until then, we had thought of identity mainly as what allowed us to explain why different decisions made by the hotel seemed to belong to the same place. Now it was becoming clear that recognising that logic was not exactly the same as establishing the property’s ability to sustain it with sufficient consistency under different circumstances. A hotel whose identity existed only in its intentions could hardly be understood that way from the outside, but neither was it reasonable to require every individual stay to be a perfect reproduction of what it aspired to be.
The real problem was beginning to emerge in the distance between the two.
I looked again at the date of the review. It was recent and therefore carried a different significance from a similar experience that had occurred five years earlier, before changes in the team or policies. I looked for later comments. There were few: one again described a late arrival resolved without difficulty, while another did not mention the subject. They were not enough to settle the question, but they were enough to prevent too hasty a conclusion.
Recurrence was beginning to matter in a different way. One discordant experience among many compatible ones might correspond to a particular circumstance or to the inevitable margin of variability in a human operation. If it began to recur over a short period, involving different guests and comparable situations, our reading would necessarily have to change. At some point, what we called an exception would cease to be an anomaly and become information about the pattern, although trying to establish that point through a fixed number seemed unreasonable.
Three negative comments did not mean the same thing in a hotel with twenty reviews as they did in one with two thousand, just as three generic complaints were not equivalent to three recent, specific accounts of exactly the same difficulty.
Before becoming a conclusion, contradiction needed context.
I therefore began to look not only at how many discordant signals appeared, but where and when they appeared. They might be concentrated in particular rooms, shifts, seasons or types of request; begin to appear from a certain point onwards; or describe as a single problem situations that, examined more closely, were actually different. There might also be later experiences capable of showing that something had changed.
A negative review thus ceased to be merely an assessment to be added to or subtracted from the whole and could become a question directed at the operation.
If several guests mentioned noise in rooms in the same area, perhaps the property was not simply “less quiet” than we thought; there might be a specific part of the building where that characteristic needed to be formulated more precisely. If difficulties with late arrivals appeared only when no advance notice had been given, flexibility remained recognisable within certain conditions. If, on the other hand, they began to recur even when guests had followed exactly the procedure indicated, the contradiction would no longer be revealing a limit to the promise, but a distance between what the property said it could do and what its operation managed to deliver.
The same discordant signal could therefore be showing quite different things. Perhaps we had formulated a characteristic too broadly; perhaps we still did not understand the conditions under which it could be sustained; perhaps the hotel had simply ceased to deliver something it continued to claim about itself.
Each possibility required a different response: better understanding what we still did not know, making more precise what we had expressed too broadly, or correcting a failure that the experience was making visible.
That sequence was almost too orderly not to be tempting, but we were still observing a single property and I preferred not to turn it into a rule. For the time being, it was enough to stop treating contradiction as noise to be eliminated in order to preserve an orderly story.
Perhaps a consistent identity was not one capable of making every signal always say the same thing, but one real enough to recognise its conditions, learn from its failures and correct the distances that were beginning to recur.
I returned to the review with which we had begun. If I worked at that hotel, I would no longer want my first question to be whether the comment was fair or unfair. I would want to know what had happened. I would look for the reservation, the time, the messages, who had been on duty, what information the guest had received and at what point the coordination had broken down—not to discredit their experience, but because that contradiction had identified an exact point where our understanding of the property and one person’s experience had ceased to coincide.
That shift also changed the way I thought about reviews. For years, we had looked at them primarily as reputation: something that improved or worsened a rating, influenced conversion and required an appropriate response. Seen through the logic we had been building, they could serve another purpose. Compatible experiences helped us recognise what the hotel was able to sustain; discordant ones showed where that capability encountered conditions, variations or failures.
Both spoke about identity, although from different places.
I thought again of someone trying to recommend the property without ever having been there. For that person, an unresolved contradiction should introduce caution. It did not necessarily prevent a recommendation, but it should make it more precise. Perhaps it was no longer appropriate to say that the hotel always adapted to late arrivals, but something much closer to what we could actually sustain: the property accommodated arrivals outside the usual hours by prior arrangement, and there were several positive experiences supporting that possibility, although one recent difficulty suggested confirming the procedure.
The formulation was less attractive than personalised attention, but it was far more useful for making a decision.
For much of the book, we had thought that making the hotel understandable meant helping what made it valuable to be recognised more clearly. Now I was beginning to understand that being understandable also required recognising the conditions under which those characteristics could be sustained and the limits that affected whether they held true.
A property did not become more trustworthy because we managed to eliminate every signal that complicated its story, but because the understanding that could be formed from it required fewer and fewer corrections when it came into contact with the real experience.
That also meant accepting that some contradictions might remain open until enough information became available; others would force us to make a claim more precise; some would reveal an operational failure; and there would even be differences that said less about the property than about people valuing the same experience differently. What one guest considered a small room might seem perfectly spacious to another, just as silence, warmth or variety could depend partly on expectations that could never be made uniform.
Understanding a contradiction was therefore beginning to require something surprisingly similar to what we had discovered about hospitality: before deciding what it meant, we needed to understand the situation. The same logic that had led us to look first at the person and then decide how to look after them now returned, applied to information.
Before drawing a conclusion, we needed to understand where the signal came from, the circumstances in which it had appeared and what it could actually demonstrate.
Finally, I returned to the homepage. Personalised attention was still there, exactly as before, and it had not become false. After everything we had observed, however, it seemed too small to contain what we now knew. Behind those two words lay decisions, policies, people, recurring experiences, conditions under which certain adaptations worked and situations in which they had not worked as we expected.
Identity remained recognisable; what we were beginning to abandon was the need to confuse recognition with perfection.
Then a different difficulty appeared. We had been able to get this far because we had spent hours gathering sources, comparing experiences and looking for the conditions that explained why apparently incompatible signals could coexist. But much of what we had just reconstructed might not actually be hidden from the hotel at all.
Reception probably knew what happened when someone arrived after a certain hour. The team knew which rooms were quieter, which requests could be adapted, which required advance notice and in which situations a general promise reached its limits.
The problem was that all that knowledge could exist perfectly well inside the property without leaving sufficient traces outside it.
If someone needed to understand the hotel before choosing it, the question was therefore beginning to shift: it was no longer enough to know what information they could find, but how much of what the hotel knew about itself still lived only inside the people who worked there.
Management Perspective
A contradiction does not automatically invalidate a characteristic, nor should it be dismissed as an inconvenient exception. It may reveal a particular circumstance, an overly broad formulation, conditions we do not yet understand, or a genuine distance between what the hotel is trying to sustain and what its operation manages to deliver.
Before deciding what a discordant signal means, we need to examine the context in which it appeared, its recurrence and its relationship with comparable experiences. Identity does not require the absence of variation, but sufficient consistency to remain recognisable and the ability to learn from those differences that reveal something that still needs to be understood, made more precise or corrected.
The Case Study
In the hotel we have been observing, the recent experience that contradicted its apparent flexibility around late arrivals did not erase the earlier signals, nor could it be explained using information we did not have. When we compared it with other discordant signals, we began to see that some might depend on particular circumstances, rooms or expectations, while others revealed conditions or failures in the operation.
We therefore stopped asking which signal should win and began asking what we needed to know in order to understand why apparently incompatible experiences could coexist within the same property.
A Question Before Moving On
If an experience contradicts what your hotel believes itself to be, do you know how to recognise whether you are facing an exception, a condition you never made explicit, or a distance the operation has not yet managed to correct?