NPS, CSAT, and CES all try to answer the same underlying question: was this good or bad for the customer? But they ask it at different moments and about different things. NPS asks about the relationship. CSAT asks about one visit. CES asks how much work the customer had to do. For a business running a counter, a chair, or a service bay, that difference decides which one you should actually be collecting.
What each metric is measuring
NPS (Net Promoter Score) asks "how likely are you to recommend us to a friend?" on a 0–10 scale. It's a loyalty question, not a satisfaction question. A customer can rate today's visit fine and still answer 6 because of something from three visits ago. NPS is built for tracking a relationship over time, which is why subscription software and airlines use it. It says less about what happened this afternoon.
CSAT (Customer Satisfaction) asks "how satisfied were you with [this visit / this order / this appointment]?" It's transactional: tied to one specific interaction, answered right after it happens. A star rating on a receipt or a QR card at the counter is a CSAT question in disguise, even when nobody calls it that.
CES (Customer Effort Score) asks "how easy was it to get this done?" It's the odd one out because it isn't about happiness at all. A customer can be delighted with the outcome and still report high effort if getting there involved three phone calls and a callback. CES exists to find friction, not sentiment.
Why this matters more for a local business than the definitions suggest
Most of what gets written about these three metrics is aimed at software companies measuring a relationship that plays out over months of logins and renewals. A restaurant, salon, dental office, or repair shop doesn't have that problem. The customer is standing in front of you, the visit just ended, and you have roughly the length of a payment transaction to ask anything at all.
That reality rules NPS out as your default metric, not because it's a bad question, but because it's the wrong question for the moment you actually have access to a customer. NPS wants reflection on a relationship; you have thirty seconds and one visit. Trying to run a relationship survey inside a checkout interaction gets you noisy, low-context answers to a question the customer hasn't really thought about yet.
CSAT fits the moment you have. A star rating tapped at the counter, right after the meal or the haircut or the oil change, captures the reaction while it's still accurate. It's also the format how star ratings affect conversion already covers from the public-review side. The same tap that measures satisfaction is the tap that becomes a review.
CES earns its place only in specific verticals: anywhere the process itself is the product. An auto repair shop that made a customer wait three days for a callback on an estimate has an effort problem even if the final repair was flawless. A dental practice with a confusing booking flow has the same issue before a patient ever sits in the chair. If your business runs on appointments, quotes, or multi-step service, a lightweight effort question alongside the star rating tells you something CSAT alone won't.
A practical default
For most walk-in and appointment-based local businesses, the order of priority looks like this:
- CSAT first, always. A one-tap star rating at the point of the visit ending. This is your baseline and should run on every customer, every day.
- CES second, if your business has a process step that can go wrong (booking, quoting, waiting, follow-up). One extra optional question, not a second survey.
- NPS last, and sparingly. If you want a loyalty read, run it quarterly by email to your existing customer list, separate from the point-of-visit flow. Don't ask a first-time walk-in how likely they are to recommend you — they haven't formed that opinion yet.
Stacking all three onto a single counter interaction is how you end up with the completion-rate problem every long survey runs into: the first question gets answered by almost everyone, the third by far fewer, and the responses that matter most, the quick, honest ones from an ordinary visit, never get collected at all.
Where the feedback goes matters as much as which metric you pick
None of these three metrics answers a separate question: where should a bad answer go? A CSAT tap of one or two stars is a signal worth acting on immediately, and that's a routing decision, not a metrics decision. Every customer should get the same prompt regardless of how you expect them to answer — singling out only the customers you think will rate you well and steering the rest away is review gating, which Google's policies prohibit. The fix isn't fewer prompts for unhappy customers; it's giving every customer, happy or not, a private option alongside the public one.
Starboard's rating flow is built around the CSAT moment specifically: one star tap at the point of experience, the same prompt for every customer, then a choice of where that feedback goes. A low rating routes to a private inbox instead of forcing a public complaint; a high one can become a review. Pair that with staff leaderboard attribution and you get the CSAT data broken down by the person who actually handled the visit, which is usually more actionable than an aggregate score sitting in a dashboard.
Pick the metric that matches the moment you have access to the customer. For nearly every local business, that moment is right after the visit — which makes it a CSAT question first, an effort question second, and a loyalty question you save for later.