How to Create an NPS Survey (Step by Step, With Real Examples)

Net Promoter Score only works if you build the survey behind it correctly, and most people learning how to create an NPS survey skip straight to the question wording without fixing the parts that actually break the metric: the trigger, the follow-up, and who it's sent to. Get any of those wrong and you'll get back noise dressed up as insight. This guide walks you through building an NPS survey that actually produces a usable score, from the one required question to where you send the results once they land.

By the end, you'll have a working NPS survey: the core question, a follow-up that explains the number, a plan for who gets it and when, and a way to read the results without fooling yourself about what they mean.

What an NPS survey actually needs

A proper NPS survey needs one required question, plus one more that does the real work of making it useful.

The scoring question is fixed, and you shouldn't change it: "On a scale of 0 to 10, how likely are you to recommend [company/product] to a friend or colleague?" That single line is the entire basis of the metric. Answer 9 or 10 and you're a promoter. Answer 7 or 8 and you're a passive. Answer 0 to 6 and you're a detractor. The score itself is just the percentage of promoters minus the percentage of detractors.

The follow-up is where it gets useful: "What's the main reason for your score?" You can calculate NPS without it, technically, but skipping it is the most common mistake people make. The 0-10 number tells you how someone feels. The open text tells you why, and why is what you can actually act on. A 6 with "support took three days to reply" is a different problem than a 6 with "the product does what I need, nothing more."

Resist the urge to add more. NPS surveys are short by design, and every extra field is a reason for someone to abandon it before finishing. If you need demographic or account-level context, pull it from your CRM instead of asking the customer again, they've already told you once.

Step 1: Decide the trigger and the timing

Before you write a word of survey copy, decide when the survey fires. This matters more than the wording of the question itself.

Relationship NPS is sent on a fixed schedule, usually quarterly, to your full active customer base, regardless of what they've just done. It tracks overall sentiment over time, and it's what most people mean when they say "our NPS." Transactional NPS works differently: it fires right after a specific moment, like an onboarding call finishing or a renewal going through, and it tells you how that one moment performed rather than how the customer feels overall.

Mixing the two without labeling them is how companies end up confused about why their NPS jumps around from quarter to quarter. A transactional score collected right after a smooth onboarding will run higher than a relationship score collected mid-contract, and comparing the two as if they're the same metric will mislead you. Pick one purpose per survey and keep the timing consistent, so the trend line actually means something.

Step 2: Build the question in SurveyKar

SurveyKar has a dedicated NPS question type, so you're not faking the 0-10 scale with a generic rating field. It renders the standard horizontal selector, applies the promoter/passive/detractor split automatically, and reports the score without you building a spreadsheet formula by hand.

Here's how to set it up:

1. Create a new survey in SurveyKar and add a question.
2. Choose the NPS (0-10) question type from the picker. You'll find it in the "Rating" category, next to star rating, CSAT and CES.
3. Keep the standard wording ("How likely are you to recommend...") unless you have a real reason to change it. Deviating breaks comparability with published NPS benchmarks and with your own past surveys.
4. Add a second question, a long-text field, for "What's the main reason for your score?" Mark it optional rather than required, but most people will fill it in anyway if the scoring question just made them think about why.
5. Consider conditional logic that routes detractors (0-6) to a slightly different follow-up, like "What would it take to improve your experience?" That sends unhappy customers straight into something your team can act on instead of a generic complaint box.

One thing worth knowing before you build: SurveyKar's dedicated NPS question type is available from the Starter plan upward, not on the free tier. If you want to test the concept first, you can still build the logic with a free-tier number rating field and calculate the split yourself, but for ongoing NPS tracking, the dedicated question type earns its keep.

Step 3: Choose how you'll send it

An NPS survey is only as good as how it reaches people, and SurveyKar supports several distribution paths. Which one fits depends on where the relationship already lives: a shareable web link works well dropped into an email or a dashboard notification, a QR code suits a physical trigger like a checkout counter or an event, and an embed placed inside a product or support portal catches people mid-session. In markets where WhatsApp is the default channel, which in India is most customers, the survey can be answered as a conversation inside your business's own WhatsApp number instead of a separate app or page.

For relationship NPS sent to an existing customer list, email with a direct link is usually simplest. For a transactional trigger right after a support interaction or a purchase, an embedded or WhatsApp-delivered survey catches people while the moment is still fresh, and that timing matters: NPS collected three weeks after the fact measures something different than NPS collected the same day. See the full breakdown of channels on the surveys and feedback page.

Step 4: Set a response target before you send

Decide how many responses you need to trust the number before you launch, not after. A score built on twelve responses swings wildly from one quarter to the next for reasons that have nothing to do with how customers actually feel, it's just small-sample noise wearing a business metric's clothes. Run your numbers through the NPS calculator once responses come in, and size your send so a quarterly business decision isn't riding on a sample too small to mean anything.

Step 5: Read the results correctly

Once responses are in, don't treat the single NPS number as the whole story. Segment before you average: a blended score across free-tier and enterprise customers, or new and long-tenured accounts, hides more than it reveals, so break it down before reporting the headline figure. Read the comments before the score reacts to it, the number tells you there's a problem, but the open-text answers tell you what it actually is, and a dip from 42 to 35 is just a number until you've read what detractors wrote.

Be wary of chasing a single published benchmark, too. "Good" NPS varies enormously by industry, and a lot of the numbers circulating online aren't built on comparable methodology. For a more grounded look at what counts as reasonable for your situation, see what is a good NPS score. And watch the trend rather than the point: one quarter's number is a data point, but three or four consecutive quarters moving in the same direction is something you can actually act on.

If you want help wording the core question and its follow-ups for different situations, the NPS survey questions guide has twenty examples you can adapt directly.

A worked example

Say you run a subscription software product with 2,400 active customers. You settle on relationship NPS, sent quarterly by email to the full base, with the standard 0-10 question plus the open-text follow-up.

First quarter, 410 responses come back, a reasonable 17% response rate for this kind of send. Of those, 180 score 9-10 (promoters), 140 score 7-8 (passives), and 90 score 0-6 (detractors). The score works out to (180/410 − 90/410) × 100 = 22.

That number alone doesn't tell you much. What's useful is reading the 90 detractor comments, which cluster around slow support response times. That's an operational fix, not a product one, and it's the kind of insight a raw score never surfaces on its own.

Try it

SurveyKar has a ready-made NPS survey template with the scoring question and follow-up already built, so you can launch in minutes instead of starting from a blank canvas. Once responses start coming in, run your score through the NPS calculator to see where it lands and how much confidence you can place in it.

Start your NPS survey on SurveyKar today. Build it from the template, send it through whichever channel your customers actually use, and get a score you can defend in the next business review.