- Each metric answers a different question. NPS asks whether customers would recommend you, CSAT asks whether one thing satisfied them and CES asks how hard it was to get something done.
- Pick the metric that matches the decision. Track CES first to fix a broken process, CSAT to judge one service moment and NPS to follow the whole relationship.
- Each score has a blind spot. NPS moves too slowly to manage a team by, CSAT rewards the moment and misses the process, and CES says nothing about whether customers value what you sell.
- Where customers can't leave, effort matters more than loyalty. Reichheld found "would recommend" failed to predict growth in near monopolies, a description that fits most utilities.
- One score is never the whole picture. Safaricom reports a headline NPS of 65 points alongside separate scores for each business area, a detractor ceiling and a count of problem-driven contacts.
The customer metric a company tracks first is usually the one someone heard about at a conference, not the one that fits its problem. Net Promoter Score (NPS), Customer Satisfaction Score (CSAT) and Customer Effort Score (CES) are the three measures most companies choose between. NPS asks whether customers would recommend you, CSAT asks whether something specific satisfied them, and CES asks how hard it was to get something done. Each answers a different question, so the right first metric depends on the decision you need it to inform.
When we agree a starting baseline with the customer-facing teams we train, the number on the table is almost always whichever one the company already collects, and it is not always the one that can show whether the problem the team wants fixed is getting better. A score chosen for the wrong question still produces a number every month. It just cannot tell you what to change.
This article sets out what each metric measures and where it came from, the blind spot in each, which one to track first in four common situations, how Kenyan companies use them, what a survey score cannot see and three mistakes that make any of them useless.
What NPS, CSAT and CES Each Measure
All three are single survey questions. What separates them is what they ask about and when:
- Net Promoter Score (NPS) measures the relationship. It asks how likely a customer is to recommend the company to a friend or colleague, on a scale from 0 to 10. Customers who answer 9 or 10 are promoters, 7 or 8 are passives and 0 to 6 are detractors. The score is the percentage of promoters minus the percentage of detractors, so it runs from -100 to +100.
- Customer Satisfaction Score (CSAT) measures one experience. It asks how satisfied the customer was with a specific purchase, visit or call, most often on a 1 to 5 scale. The score is usually the percentage who answered 4 or 5.
- Customer Effort Score (CES) measures how hard the customer had to work. The original question was "How much effort did you personally have to put forth to handle your request?", scored from 1 (very low effort) to 5 (very high effort), and the score is the average.
| NPS | CSAT | CES | |
|---|---|---|---|
| What it asks | Would you recommend us? | Were you satisfied with this? | How much effort did this take? |
| What it covers | The whole relationship | One purchase, visit or call | One task the customer needed done |
| When to ask | Every few months, not after each contact | Straight after the experience | Once the issue is resolved, or not |
| What it is built to predict | Referrals and growth against competitors | How that moment landed | Whether customers stay after a problem |
One caution before comparing numbers: many companies now word the effort question the other way round, so that a high score means easy. Check which version a report uses before you read a CES of 4 as good or bad.
Where Each Metric Came From
CSAT is the oldest of the three and the default in most service teams. Asking customers whether they were satisfied predates any of the named methods, and for decades it was the main way companies listened.
Satisfaction became a national statistic in 1994, when University of Michigan researchers launched the American Customer Satisfaction Index, built on a Swedish barometer first run in 1989. It still publishes quarterly across more than 40 industries. Thirty years of benchmarks are why CSAT is easy to compare and why it became the default.
NPS came from a challenge to that default. In 2003, Fred Reichheld, working with teams from Bain & Company, argued in Harvard Business Review that satisfaction scores often failed to track sales. His team surveyed more than 4,000 customers, then waited six to 12 months to record what each one actually bought and who they referred. The question "How likely is it that you would recommend [company X] to a friend or colleague?" ranked first or second as a predictor in 11 of the 14 cases they could measure. One question replaced a long survey, and NPS spread quickly through boardrooms.
CES came from service research. In 2010, Matthew Dixon, Karen Freeman and Nick Toman of the Corporate Executive Board studied more than 75,000 people who had contacted a call center or used self-service. In "Stop Trying to Delight Your Customers", they reported that CSAT was a poor predictor of whether those customers would buy again, spend more or speak well of the company. NPS did better, and CES did best of the three for service interactions.
Effort turned out to predict what customers did next. In the same 2010 study, 94% of customers who reported low effort intended to buy again and 88% said they would spend more, while 81% of those who struggled to solve a problem said they would speak badly of the company. Removing effort protects revenue more reliably than adding delight.
Source: Harvard Business Review, Stop Trying to Delight Your Customers
The Blind Spot in Each Score
Each metric was designed for one job, and each misleads when it is asked to do another.
NPS
NPS is a relationship score. It reflects price, product, brand and every contact the customer has had, so it moves slowly, and when it falls it doesn't say why. A team that is told its NPS dropped four points this quarter has no idea what to do differently on Monday. NPS also assumes the customer has a choice. Reichheld himself found that "would recommend" didn't predict growth where customers had little choice, such as local telephone and cable TV monopolies, or where the people using a product were not the people who bought it.
CSAT
CSAT rewards the moment. A customer can rate a friendly agent 5 out of 5 and still leave, because the agent was pleasant but the problem took three calls to fix. Each of those calls may score well. The process that needed three of them never gets a score at all.
CES
CES only covers getting things done. It tells you a refund was easy to claim, not whether the customer wanted to buy from you again. A company can make it effortless to cancel an account and still lose the account. Effort is the right lens for service and processes, and the wrong one for judging whether customers value what you sell.
Which Customer Metric to Track First
Start from the decision you need the number to inform, not from the metric. Four situations cover most of the requests we see.
1. A process customers complain about
Track CES first. If customers are calling twice about the same bill, chasing a delivery or re-explaining an issue to a second agent, the problem is effort. Ask the effort question at the end of the process, not after each call, and pair it with the count of repeat contacts.
2. A service moment you are trying to improve
Track CSAT. If the question is whether new branch staff, a redesigned onboarding call or a new checkout handle one moment well, CSAT asked straight afterwards gives a fast, specific answer. It is also the score frontline teams find easiest to act on.
3. Retention and growth against competitors
Track NPS, quarterly or twice a year, and always with a follow-up question asking why. NPS is the measure a board can compare over time and against a market. Treat it as the result the other two feed into, not a target for a service team.
4. Customers who cannot easily leave
Track CES and complaints, and treat NPS with caution. A household connected to the grid in Nairobi cannot switch power distributor, and a county water customer cannot change water company. Many business-to-business suppliers in Kenya hold a similar position under long contracts. In these markets, customers stay whatever they would tell a friend. Qualtrics XM Institute's 2025 global consumer research found people were least likely to cut spending after a bad experience with a public utility. Effort shows where they are paying for the lack of choice in time and frustration.
One metric tracked well beats three tracked badly. A score that a named owner reviews every month, with the comments read and a change made, does more than three dashboards nobody acts on. Add a second metric only once the first one is changing decisions.
Not sure which customer score your teams should be held to?
Our Customer Experience program starts by agreeing a baseline in a measure you already track, then trains customer-facing teams on the behaviors behind it and reports the change at the end. Request a proposal.
How Kenyan Companies Use These Scores
Some of the largest Kenyan companies now publish their customer scores, and the most detailed example shows that one number is never enough. Safaricom's Annual Report 2024 reports a headline NPS of 65 points, which it ranks first in its market, and separate NPS scores for its network at 78, its fixed business at 61 and its enterprise business at 43. It reports leading competitors by more than 10 points in each of those three areas.
Safaricom Net Promoter Score by Business Area, FY2024
Headline score and three business areas. Source: Safaricom PLC, Annual Report and Financial Statements 2024
Network 78, headline 65, fixed 61, enterprise 43.
NPS is the percentage of promoters minus the percentage of detractors, so values are points on a scale from -100 to +100, not percentages.
A 35-point gap between the network and enterprise scores is invisible inside the headline 65. That is the case for scoring each business area separately: the overall number hides where the work is. The same report goes further than NPS. It sets a ceiling of 6% on detractors, tracks NPS on individual customer processes against a 40-point baseline, and reports cutting what it calls customer pain, measured as technology-driven demand, by 44%. In effect, that last measure counts contacts caused by problems, the operational partner of an effort score.
Most Kenyan organizations are not Safaricom and don't need its dashboard. The structure is still worth copying at any size: one relationship score for the board, a score per area for the managers who can change it, and an operational count that shows where customers are working too hard.
What Survey Scores Miss
Every survey score is built from the customers who answer, and the unhappiest ones often don't. A customer who gave up on a problem rarely fills in the survey about it. So a CSAT that looks steady can sit on top of a rising number of customers who quietly stopped calling.
Three things belong next to any survey score:
- Repeat contacts. How many customers contacted you more than once about the same issue. This is the closest operational partner to CES.
- Escalations and formal complaints. What reached a supervisor, a regulator or social media, and how long it took to close.
- The verbatim comments. What customers wrote in the free-text box, read by someone with the authority to fix what it describes.
Forrester's 2026 CX Index rates brands on three qualities of experience: effectiveness, ease and emotion. It drew on more than 224,000 customers' views of 462 brands in 13 countries. Those three map closely onto CSAT, CES and the loyalty NPS tries to capture, which is why no single score covers the whole experience.
Three Common Customer Metric Mistakes
Surveying after every contact
A customer who gets a survey after every call, message and branch visit stops answering, and the ones who keep answering are not typical. Survey fatigue thins the sample until the score describes a small, unusual group. Ask NPS a few times a year, and ask CSAT and CES on a sample of contacts rather than all of them.
Paying staff on the score
Once a bonus depends on a score, staff learn to manage the score. Agents ask customers for a 10, branches choose who gets the survey and difficult cases go unrecorded. If a score is going to be tied to pay, tie it to a team over a long period and check it against repeat contacts and complaints, which are harder to manage.
Reporting the number without the reasons
A score without its comments tells leaders something changed but not what. The follow-up question, "What is the main reason for your score?", is where the work is. A monthly report that shows the number and the five most common reasons beside it will get acted on. One that shows only the trend line won't.
Frequently Asked Questions
What is the difference between NPS, CSAT and CES?
NPS (Net Promoter Score) asks how likely a customer is to recommend the company, so it measures the whole relationship. CSAT (Customer Satisfaction Score) asks how satisfied the customer was with one purchase, visit or call. CES (Customer Effort Score) asks how much effort the customer had to put in to get something done. Each answers a different question, so they are not interchangeable.
How is NPS calculated?
Customers answer how likely they are to recommend the company on a scale from 0 to 10. Those who answer 9 or 10 are promoters, 7 or 8 are passives and 0 to 6 are detractors. NPS is the percentage of promoters minus the percentage of detractors, so it runs from -100 to +100 and is reported in points, not percent.
Is CES better than NPS?
For service interactions, the 2010 Harvard Business Review study that introduced CES found it predicted whether customers would buy again and spend more better than either CSAT or NPS. NPS remains the stronger measure of the overall relationship and of growth against competitors. CES is better for fixing processes; NPS is better for tracking loyalty.
Which customer metric should a company track first?
The one that matches the decision you need to make. Track CES first if customers complain about a process, CSAT if you are improving one service moment, and NPS if you want to follow retention and growth against competitors. One metric with a named owner who acts on it each month is worth more than three nobody uses.
Does NPS work when customers cannot switch provider?
Not well. Fred Reichheld, who introduced NPS, found that willingness to recommend did not predict growth in monopolies and near monopolies, where customers have little choice. For a utility or a supplier on a long contract, customer effort and complaint data show more about where customers are struggling.
How often should you survey customers?
Ask NPS a few times a year rather than after every contact, and ask CSAT and CES on a sample of interactions shortly after they happen. Surveying after every call or visit causes survey fatigue, and the customers who keep answering stop being typical of the rest.
Conclusion
NPS, CSAT and CES are not rivals. They measure the relationship, the moment and the effort, and each is useful only for the decision it was built for. Choose the one that matches your problem, and give it an owner.
Think of the customer score your company reports most often. Can the people who see it each month name one thing they changed because of it this year?
If the honest answer is no, our Customer Experience program can help you choose the right measure and train the teams behind it. Request a proposal and we will start from the score you already collect.
