Experience as Curve, Not Score

There is a category of hotel renovation that prioritises the lobby above everything else. This is not laziness or mismanagement; it is deliberate strategy. The lobby is what photographs well on booking platforms, what appears first when someone is choosing where to stay, what creates the initial impression that converts a browser into a guest. A redesigned atrium is an acquisition instrument. So it gets built first.

The guest walks through the redesigned atrium, is genuinely impressed, opens the door to a room that has not changed since 2009, and writes a worse review than the hotel would have received before the renovation. Not because the room deteriorated. Because the experience now has a different shape: it started high and fell, and that descending pattern has written itself into the verdict. The same beds, the same breakfast, the same shower. A different trajectory, and therefore a different reality.

This is not an anecdote about poor hotel management. It is a precise description of the problem facing almost every company that measures customer experience with a score: it captures the average, manages the average, and then wonders why the average improves while churn grows.

What the Brain Actually Calculates

Christopher Hsee and Robert Abelson demonstrated in 1991, in a study whose implications are still underestimated, what actually happens when people evaluate experiences. Participants who had improved from 60 to 70 within a rating frame reported higher satisfaction than participants who had remained stable at 75. A lower absolute value, a higher perceived quality of experience. The 70 beat the 75 because it arrived with the right sign in front of it.

This is not an artifact of an unusual experiment. It is the dopamine system in its normal operation. The brain does not measure absolute values; it measures prediction errors, the gap between what was expected and what arrived. A continuous improvement generates a continuous series of small positive surprises. A stable high value generates none, because it only confirms what the brain already anticipated, and a confirmed prediction produces no signal at all.

From an evolutionary perspective this is entirely coherent. An organism that responds to change is superior to one that responds to absolute states in virtually every environment. Threats and opportunities announce themselves through deviation, not through constancy. The brain was built for a world in which direction is the signal, not height.

What Scores Systematically Destroy

NPS, CSAT, and satisfaction scores are compression operations. They take a temporal sequence of experiences, a trajectory across weeks or months, and distill it into a single number. What gets lost in that compression is precisely what determines the brain's verdict: the direction.

A customer with a stable score of 7 and a customer whose score rose from 5 to 7 are identical in the reporting dashboard. But the subjective reality of these two people is neurobiologically different. The first is experiencing confirmation of the expected. The second is experiencing a sequence of positive prediction errors that the brain registers as improvement and rewards with attachment.

Conversely, the customer who fell from 9 to 7 is not flagged as a risk in any NPS table, even though they represent the highest churn risk in the portfolio. Not because 7 is a bad score, but because it arrived with the wrong sign. Churn almost never comes from a single bad moment. It comes from the cumulative recognition that the curve is moving in the wrong direction, a recognition that forms over months before it becomes a cancellation. That formation is invisible in an average.

Acquisition and Retention Are Different Problems

Here lies a confusion that is expensive and rarely made explicit. In acquisition, absolute benefit in the present moment counts. The offer must be good enough to trigger a first-time decision, in a moment when no trajectory exists, no comparison to yesterday, no accumulated model of how the relationship has developed. You are selling to a brain with no expectation history, and absolute quality wins.

Retention operates by a fundamentally different mechanism. A customer who has been working with a product for eighteen months has built an extensive expectation history. Every new interaction is measured not against zero, but against everything that came before. If what came before was better than what is arriving now, the brain registers a downward deviation, regardless of whether the absolute value still sits comfortably above some minimum threshold.

This is why organisations need both, but cannot build both with the same tools. Acquisition is a moment problem. Retention is a trajectory problem. Trying to capture both with a single CSAT score means measuring one correctly at the expense of the other, never both.

What Trajectory Design Means in Practice

Designing a customer journey by directional trajectory rather than average satisfaction requires thinking differently about sequence. The goal is not to operate every touchpoint at maximum level, which is expensive and neurobiologically ineffective because constancy produces no signal. The goal is a curve that rises.

In subscription products, this means staging feature discovery rather than granting full access at signup. Someone who sees the full product on day one has already experienced the best version of it and can only be disappointed from there, because every subsequent phase must be worse than the maximum. Someone who discovers something that improves their workflow in each month of use experiences a rising curve, and a rising curve binds.

In consulting and service engagements, it means deliberately placing insight moments in the middle phases of the work, not only in the closing presentation. The enthusiasm with which a project begins is a high starting point. If the middle of the project falls below it, the trajectory already carries a negative sign, even if the final delivery recovers. The client remembers the shape, not only where the endpoint landed.

This is not the same as recommending that you lower the beginning to manufacture an artificial curve. That would be a manipulative reading that misses the point entirely. The recommendation is to distribute the performance, the insights, the surprises intelligently across the time axis rather than concentrating them at the start, where they have the least retention power.

The Metric That Is Missing

What follows for measurement? The question is not whether the score is good today. The question is whether the score today is higher than it was three months ago, and whether it was higher three months ago than six months ago. An organisation that asks this question systematically and acts on it operates a different architecture from one that reports a quarterly average.

The operative metric is not the value but the sign. Not where the experience sits but in which direction it is moving. That is the information the brain responds to. That is therefore the information an experience system must manage if it wants to manage the brain's experience of itself.

Loyalty is not a state produced by high absolute satisfaction. It is a state the brain attaches to a perceived upward trajectory. Understanding trajectory design means not building better individual moments. It means building better curves.