Your NPS Score Went Up Last Quarter. Here Is Why That Might Be a Problem.
Your quarterly review looks clean. The NPS survey went out, the score went up, and someone put a green arrow next to it in the deck. The room feels good about it.
We have sat in that room. And we have also seen what happens six months later, when the score is still climbing but so is churn, response rates are dropping, and nobody can name a single decision that changed because of that number.
NPS is one of the most widely recognized customer metrics in retail, eCommerce, fintech, hospitality, SaaS and beyond.
It is also one of the most consistently misused ones. Not because the methodology is flawed, but because the way most companies implement it turns a useful signal into a scorecard with no one watching the game.
This is not an argument against NPS. It is an argument for using it in a way that actually changes something.
WHAT IS NPS AND WHY IT BECAME THE DEFAULT
NPS, or Net Promoter Score, was introduced by Fred Reichheld in a 2003 Harvard Business Review article as a simpler alternative to long customer satisfaction surveys. The premise was straightforward: ask customers one question, calculate a single number, and use that number to predict growth.
The question is: “How likely are you to recommend us to a friend or colleague?”
Respondents answer on a scale of 0 to 10. Those who answer 9 or 10 are Promoters. Those who answer 7 or 8 are Passives. Those who answer 0 to 6 are Detractors.
NPS = % Promoters – % Detractors.
A score above 0 means more people would recommend you than not. Above 50 is generally considered strong. The benchmark shifts by industry, so a score of 30 in retail means something different than a score of 30 in financial services.
NPS caught on because it is fast to deploy and easy to explain in a leadership meeting. A single number travels well across departments. That simplicity is also its biggest vulnerability.
THE PROBLEM IS NOT THE SCORE. IT IS WHAT HAPPENS AFTER.
Here is the pattern we see repeatedly at Canadian companies in the 50-200 employee range: the NPS survey goes out once a quarter or once a year, the results come back, the score gets reported, and that is the end of the process.
No follow-up with Detractors. No analysis of what specifically drove the rating. No change to a touchpoint, a policy, or a communication. The number exists, and then nothing happens.
When the score goes up in that context, the temptation is to treat it as evidence that things are improving. But a rising NPS without a clear link to operational change is not a signal of progress. It is noise.
There are several reasons a score can go up that have nothing to do with actual customer experience improvement:
Survey timing and sample selection. If the NPS survey goes out right after a successful promotion, a product launch, or a particularly smooth peak season, the sample is skewed toward customers who interacted with the business at its best. Those same customers might rate you very differently in month three.
New customer inflation. Newer customers tend to rate higher because they are still in the honeymoon period. If the business is growing, each survey cycle brings in more new customers who have not yet experienced the friction that longer-tenured customers have. The score goes up. Loyalty does not.
Response rate drift among unhappy customers. Detractors stop responding over time. They have already decided how they feel. If the people who would rate you a 4 or a 5 stop opening the survey, the average climbs without any improvement in the underlying experience.
None of these mean NPS is unreliable. They mean that a score in isolation tells you almost nothing. What matters is the system around it.
WHAT A CLOSED-LOOP NPS SYSTEM ACTUALLY LOOKS LIKE
The value of NPS comes from closing the loop, which means treating the survey as the beginning of a process rather than the output of one.
A closed-loop system has three components that most companies skip entirely.
Detractor follow-up within 48 hours. Anyone who responds with a 0 to 6 should be contacted by a real person within two business days. Not with a discount code.
With a genuine question: what happened, and what can we do about it. This single step recovers more customers than any score improvement effort, and it generates the most specific, actionable feedback a company can get. It also tells you where the actual problems are, which the score itself never will.
Structured qualitative capture. The one-question NPS survey is not enough on its own. The follow-up question, “What is the primary reason for your score?“, is where the signal lives. That open-ended response needs to be categorized and reviewed, not left as raw text in a spreadsheet. Common patterns across Detractor comments point directly to the touchpoints that need attention.
Operational ownership. Each category of feedback needs an owner. If Detractors consistently mention shipping delays, that belongs to operations. If they mention confusing return policies, that belongs to the team managing customer communications. Without ownership, the feedback loop closes on paper but not in practice.
This is what separates a VoC program from a survey cadence. The survey is one input. The program is the structure that turns inputs into decisions.
HOW TO RUN AN NPS SURVEY THAT GENERATES USEFUL DATA
Before the score means anything, the survey design has to be right. These are the variables that matter most for Canadian companies.
Relationship NPS vs. transactional NPS. Relationship NPS measures overall sentiment about the brand and is typically sent once or twice a year.
Transactional NPS measures sentiment about a specific interaction, such as a purchase, a return, or a support call, and is sent shortly after that interaction occurs. Both are useful. They answer different questions. Running only one and treating it as the complete picture is a common gap.
Timing relative to the experience.
For transactional surveys, the window matters. Sending an NPS survey within 24 to 48 hours of the interaction captures the most accurate sentiment. Waiting a week introduces noise from other experiences the customer has had in between.
Survey channel and response rate. Email remains the most common channel, but response rates for email NPS surveys in retail typically sit between 5 and 15 percent. That is not enough to make confident decisions if the sample is not representative. SMS delivers higher response rates in certain customer segments. In-app surveys work well for eCommerce companies with an active app. The channel needs to match where the customer actually is.
Minimum sample size for meaningful segmentation. A score calculated from 30 responses is not stable. Small sample sizes produce large swings between periods that are statistically meaningless. For a company with a few thousand active customers, a target of at least 100 responses per period is a reasonable floor before segmenting by customer type, product line, or channel.
NPS vs. CSAT: which one to use. NPS and CSAT (Customer Satisfaction Score) measure different things and are not interchangeable. NPS measures overall loyalty and likelihood to recommend, which makes it a leading indicator of long-term retention. CSAT measures satisfaction with a specific interaction, which makes it more useful for diagnosing friction at individual touchpoints.
A company running only NPS can miss localized problems that are dragging down satisfaction without yet showing up in the loyalty score. Running both in parallel gives a more complete picture. For a deeper look at how these metrics work together, this guide to customer satisfaction metrics for Canadian retailers covers the full landscape.
WHAT TO DO IF YOUR SCORE WENT UP AND YOU ARE NOT SURE WHY
This is a more common situation than most companies admit. The score improved, but nobody can trace it to a specific decision or change.
The first step is to go back and ask three questions: Did the survey timing change? Did the customer mix change? Did the response rate change? If any of those shifted between periods, the score comparison is not apples to apples.
The second step is to look at the open-ended responses from both periods side by side. Are the themes different, or are the same complaints still appearing even though the score went up? If Detractors are still writing about the same issues but there are fewer of them because response rates dropped, the problem has not been solved. It has just become less visible.
The third step is to look at what happened operationally in the same period.
If the score went up and something genuinely changed, that connection should be documentable. If nothing changed and the score went up anyway, that is a signal that the survey design or sample needs attention before the next cycle.
NPS is a useful metric. It is not a self-explanatory one. The score tells you where sentiment landed. It does not tell you why, and it does not tell you what to do next. That part requires a system.
If your team is running an NPS survey but struggling to turn the results into decisions, that is usually a structural problem rather than a measurement problem.
How Makeable Can Help
Knowing what data to look at, how to analyze it, and what to do with what you find is not always straightforward. At Makeable, we help growing businesses in Canada turn their customer data into clear, actionable insights that drive real growth.
Whether you need help identifying the right metrics to track, analyzing trends in your existing data, or building a strategy based on what your customers are telling you, we are here to help.