Share article
The board wants to know one thing: how are our customers doing? Yet the answer usually arrives as three, five or twelve numbers. A composite index promises to merge them into one. The question is what gets lost along the way.
The history of customer experience measurement is, to a large extent, a history of the search for a single number. In December 2003, Frederick Reichheld of Bain & Company published an article in Harvard Business Review with a telling title: “The One Number You Need to Grow.” In it, he introduced the Net Promoter Score (NPS), a metric built on a single question: how likely is a customer to recommend the company? The idea was seductive. Forget long questionnaires, track one number, and you will know whether you are going to grow.
Twenty years later, it is clear that one number was not enough. Alongside NPS, companies now track satisfaction, customer effort, churn and dozens of operational indicators. And leadership is asking again: can all of this be summed up in a single number? A composite index, a summary indicator built from several metrics, promises that it can. Its real value, however, depends on details that rarely make it into board presentations.
Why Leadership Wants a Single Number
The desire for a single indicator is not a sign of superficiality. It is a rational response to data overload. A typical mid-sized company tracks NPS for the overall brand relationship, CSAT (Customer Satisfaction Score, a measure of satisfaction with a specific interaction) after contact with support, and CES (Customer Effort Score, a measure of how much effort the customer had to expend) in digital channels. Each number has its own owner, its own report and its own story. The board ends up with five charts and no answer.
The data on this point are consistent: customer experience measurement convinces very few people. According to a McKinsey survey of more than 260 CX leaders (2021), 93% of companies measure experience with survey-based metrics, yet only 15% of them are fully satisfied with how they do it. When a CX team cannot translate its numbers into the language of the P&L, leadership takes matters into its own hands and starts demanding a single number.
A composite index answers this with a compromise: leadership gets one number, while the CX team keeps the richer picture. One key condition, however, tends to be overlooked. The compromise works only if the index is built honestly and everyone understands what is in it and what is missing.
Three Metrics, Three Different Questions
Before anything is merged, you need to know what is being merged. NPS, CSAT and CES appear side by side in reports as if they were three views of the same thing. They are not.
NPS measures the relationship. The customer rates the brand as a whole, including price, product and everything they have ever experienced with the company. The score ranges from −100 to +100 and moves slowly. CSAT and CES, by contrast, measure the moment: how a specific interaction went and how much work it cost the customer. They react quickly and can be tied to a specific process. That this is no minor detail was shown by the article “Stop Trying to Delight Your Customers” in Harvard Business Review (2010): among customers who expended only little effort, 94% reported an intention to buy again.
The first lesson follows. These metrics do not operate at the same level, they use different scales and they react on different timelines. Adding them up without thought is like a doctor averaging blood pressure, cholesterol and a stress test result and telling the patient that their health is “seventy-two.”
The opposite is also true. Where metrics measure related things (willingness to recommend, overall satisfaction, willingness to repurchase), they often overlap heavily. An index made of three such items does not deliver three times the information. It delivers one piece of information at three times the cost.
How Several Numbers Become One
Composite indicators were not invented by the CX field. Economists have worked with them for decades, from human development indices to national competitiveness rankings. The process is always similar and rests on three decisions.
Normalization. The metrics are converted to a common scale, usually 0 to 100. NPS, the percentage of satisfied customers and an average on a seven-point scale thus start to “speak the same language.”
Weighting. Each component receives a weight. Equal weights are the simplest option, but they quietly assume that all metrics matter equally to the business. A better approach is to derive the weights from data, that is, to find out which metric best predicts repeat purchase or customer churn.
Aggregation. The components are combined, most often through a weighted average. That average, however, has a treacherous property: a weak result in one area can be “paid for” by a strong result in another. Should an excellent willingness to recommend make up for the fact that a customer had to fight through three channels to get a problem solved?
A well-built index looks like the Forrester CX Index. It evaluates the effectiveness, ease and emotional quality of interactions and derives its weights from what actually builds loyalty in a given industry. Forrester now also places it within a broader indicator that, since 2026, includes employee experience (EX) as well. Composite logic is thus moving from the customer to the organization as a whole.
What this looks like in practice is shown by a simple illustrative calculation with equal weights. The figures are hypothetical, not taken from a real company.
| Metric | Q1: measured | Q1: rescaled 0–100 | Q2: measured | Q2: rescaled 0–100 |
|---|---|---|---|---|
| NPS (−100 to +100) | +30 | 65.0 | +10 | 55.0 |
| CSAT (% satisfied) | 82% | 82.0 | 90% | 90.0 |
| CES (average 1–7) | 5.4 | 73.3 | 5.6 | 76.7 |
| Composite index | 73.4 | 73.9 |
In the second quarter, the index rose slightly. The board sees stability, perhaps even a modest improvement. In reality, NPS fell by twenty points, a signal that in a standalone report would have triggered a crisis meeting. Improvements in support (CSAT, CES) concealed a deterioration in the brand relationship. The index did not lie. It simply answered a different question from the one leadership believed it was asking.
Clarity at the Cost of Blindness
The arguments for a composite index are real. A single number can be placed in an annual report, linked to financial results and tracked over time without explaining five charts. It dampens noise, because a random swing in one metric gets lost in the whole. And if the weights are derived from customer behavior, the index can predict business results better than any of its components.
The price of clarity, however, is high, and it takes three forms.
The first is loss of diagnosis. When the index drops by three points, nobody knows why. The savings in reporting turn into extra work searching for causes.
The second is masking of opposing trends, as the illustrative calculation above showed. An improvement in one part of the experience and a deterioration in another cancel each other out. The number stands still while tectonic plates shift beneath it.
The third and most serious is Goodhart’s law: when a measure becomes a target, it ceases to be a good measure. A composite index is exceptionally vulnerable to this effect because it is opaque. As soon as bonuses are tied to it, teams quickly find the component that is cheapest to improve and start optimizing that, not the customer experience. NPS creator Fred Reichheld himself acknowledged in Harvard Business Review in 2021 that tying the score to rewards had turned NPS into a vanity statistic at a number of companies. Not even the author of the most famous “one number” claims today that one number is enough.
When One Number, and When the Whole Dashboard
The real question, then, is not whether to have a composite index or not. It is: who is it for, and what decision is it meant to support?
A single index makes sense where the audience needs to track a long-term trend and compare, not manage. Typically this means the board, investors, or a comparison of branches with similar customer journeys. The precondition is enough data for the weights to come from customer behavior rather than from a meeting.
Several metrics side by side are the better choice at the operational level, where a manager needs to know which lever to pull. The same applies to companies whose CX program is just starting and that do not yet know which metrics actually predict their business. And it applies whenever the components measure different things, namely the brand relationship and the quality of individual interactions.
If a company opts for an index, it pays to follow four principles:
- The index as the top of the pyramid, not a replacement. Anyone who sees the summary number must be able to see its components at a glance.
- Weights validated by behavior. Derived from repeat purchases or churn and reviewed once a year.
- A safeguard against masking. A drop in any component beyond a set threshold triggers a separate alert, even if the index holds steady.
- Never tie rewards to the index alone. Always pair it with the components and hard business data.
Leadership’s desire for a single number is legitimate, and the CX profession will do better to serve it than to ignore it. A composite index, however, is a communication tool, not a diagnostic one. It will tell you that something is happening. It will not tell you what. Companies that understand this difference have a compass. Everyone else has a thermometer showing the average temperature of a patient whose head is on fire and whose feet are freezing.
