NPS survey software fields the 0 to 10 recommend question, sorts answers into promoters, passives, and detractors, and tracks the score wave after wave. Wavefield runs NPS as a real study: an AI agent builds the questionnaire, you field it to your own customers, and every score ships with the margin of error that says how much of it is signal. From $99 per study, $99 per repeat wave.
Build an NPS study freeMost Net Promoter Score software is the always-on kind: an in-app widget or triggered email that prompts whoever shows up, streams scores into a dashboard, and bills monthly. For transactional NPS at volume, wired into a CRM with detractor alerts, that class of NPS survey platform is genuinely the right machine. Bain & Company, which created the Net Promoter System, has always framed the score as an operating system for loyalty, and the always-on tools are built in that image.
The study kind treats NPS as a survey estimate. You choose the sample from your full customer list, field in a defined window, weight the results if segments answer unevenly, and read the score against its margin of error. That version answers the question executives actually ask, which is whether loyalty went up, with a number that can survive the follow-up question of how sure you are. Wavefield is an NPS survey tool of this second kind, so weigh our bias accordingly. The honest tell is your own cadence: if NPS reaches your board deck quarterly, you need a measurement, and a measurement has a margin.
The pricing follows the split. Always-on NPS tools bill monthly, commonly a few hundred dollars a month on annual terms, which is fair if surveys fire every day and the dashboard gets read. The study model charges per measurement: a Wavefield NPS study is $99 to $299 depending on complexity, and each repeat tracker wave is $99. For a company that measures relationship NPS quarterly, that is four waves a year at a few hundred dollars total, against a subscription that costs that much by February and does not print a margin of error in December.
NPS is the difference between two percentages, which makes it roughly twice as noisy as either percentage alone. From 150 responses, a score of 37 carries a margin near ±12 points at 95% confidence. A 4-point rise on 200 responses is a coin flip wearing a trend line. NPS tools almost never print this number; our free NPS calculator does, and so do our reports. If you want to size a wave before fielding it, the sample size calculator turns a target margin into a response count.
Relational NPS asks about the whole relationship on a steady cycle and is the score that belongs in a board deck. Transactional NPS fires after a specific event, a closed ticket or a delivery, and is really process feedback wearing the NPS scale. Both are useful. Blended, they are neither: if support interactions spike in a quarter, a mixed trend line moves even when loyalty did not. The quiet failure mode is response bias, and it is the same in both modes: the delighted and the furious answer, the middle does not. AAPOR's survey best practices put it plainly: when few people respond, you risk missing some types of respondents entirely. Chasing non-responders and weighting the results is unglamorous work, and it is the work that makes the score mean something. Raising the share who answer at all is covered in survey response rates.
Where NPS sits among its sibling metrics is its own decision. CSAT reads a specific interaction, CES reads effort, NPS reads the relationship, and the tooling follows the same split: customer satisfaction survey software covers the CSAT side, and customer feedback software, by kind maps the whole category. The craft of asking, question wording and survey timing and follow-ups, lives in customer feedback surveys, done properly.
Describe your customers and what you want to learn, and the agent drafts the questionnaire: the standard 0 to 10 recommend question, the open-ended why, and the driver questions that make the score explainable. You review everything before it fields.
Send personal one-use links by email or SMS to the customers you choose, or share an open link. You decide who gets asked, which is what makes the result a measurement instead of a popup tally.
Every NPS we report carries its margin of error, and segment comparisons carry significance tests. A 4-point move on 200 responses reads as what it is: noise until proven otherwise.
The follow-up verbatims get coded into auditable themes tied back to quotes, so the score arrives with its reasons. A tracked NPS without coded reasons is a thermometer without a diagnosis.
Re-field the identical instrument quarterly or twice a year. The questionnaire stays locked across waves, so the trend line measures your customers and not your edits.
SPSS and CSV with weights, quality flags marked rather than silently dropped, and the raw 0 to 10 distribution included, because sometimes the real story is passives sliding to detractors before the score moves.
Reading the results honestly, margins, crosstabs, and what counts as a real difference, is covered in how to analyze survey results.
Surveys firing after every closed ticket, scores streaming into a live dashboard: that is the Delighted and Retently class, and it is good at the job. We field studies at moments you choose; we do not live inside your product.
Detractor alerts routed to account owners, follow-up tasks in Salesforce, escalation SLAs: that is operations tooling in the AskNicely and CustomerGauge class, priced accordingly, and worth it if your team will actually work the queue.
If the responses are already collected, you do not need software at all. The NPS calculator takes your three counts and returns the score with its margin of error, free, no signup, no email gate.
Pull from your full customer list and decide who gets asked: stratify by plan, tenure, or region so the invited group looks like the customer base. This single step separates a measurement from a popup tally, and it is the step always-on NPS tools skip by construction, because a widget can only sample whoever happens to show up.
Two to three weeks, with one or two reminders to non-responders. A defined window makes waves comparable; a survey that is always open samples different moods in different months and calls the blend a trend. Track the response rate as a first-class number, because it is the best early warning that a wave is going to mislead you.
If enterprise customers answered at twice the rate of starter plans, the raw score is a measurement of your inbox, not your base. Weight the results back to the customer mix, then read the score with its margin of error in view and resist narrating any move smaller than it.
The open-ended follow-up is where the actionable material lives. Code the verbatims into themes, trace each theme back to quotes, and report the score and its reasons as one artifact. A score alone starts arguments; a score with coded reasons starts fixes.
One honest caveat for B2B companies: on a base of 60 accounts, NPS stops being a statistic and becomes a roll call. Survey everyone, report counts next to percentages, and read every verbatim, because at that scale each account is a segment of one. The economics of researching business audiences are mapped in B2B market research.
NPS survey software fields the standard Net Promoter Score question (how likely are you to recommend us, on a 0 to 10 scale), sorts responses into promoters, passives, and detractors, and computes the score: promoter percentage minus detractor percentage. The category splits into always-on tools that prompt customers continuously through in-app widgets and email triggers, and study tools that field NPS to a chosen sample at defined moments. Wavefield is the second kind.
It depends on which job you are hiring for. For always-on transactional NPS wired into a CRM with detractor follow-up workflows, the dedicated class (AskNicely, CustomerGauge, Delighted and similar) is built for exactly that. For relational NPS measured as a proper study, with a designed sample, a margin of error, and coded follow-up reasons, a research platform like Wavefield fits better. The most common mistake is buying the always-on kind and using it twice a year.
Mostly in the form of free tiers with response caps, and the arithmetic itself is free because it is subtraction: Wavefield's NPS calculator computes your score and its margin of error from three counts, no signup. Fielding is where real costs live. A Wavefield NPS study is $99 per project and $99 per repeat wave, with no per-response or per-seat charges.
Relational NPS asks about the overall relationship on a periodic cycle, quarterly or twice a year, and reads as a loyalty trend. Transactional NPS fires right after a specific event, like a support ticket closing, and reads as process feedback for that touchpoint. Both are legitimate; mixing them in one trend line is not, because the blend moves whenever your ticket volume does. Wavefield is built for the relational kind.
For relational NPS, quarterly is the common ceiling and twice a year is enough for most companies; more often mostly resamples the same opinions and burns goodwill. Hold the sample design and question wording constant between waves, and survey any individual customer at most a couple of times a year. Frequency is the easiest lever to break a trend with, because rising fatigue shows up disguised as falling loyalty.
More than most teams think, because NPS is a difference between two percentages and inherits the noise of both. As working ranges: around 250 to 400 responses puts the margin of error near ±10 points, and reaching ±5 typically takes 900 to 1,500. Below a few hundred responses, report the margin alongside the score and judge direction across several waves instead of reacting to one reading.
Related: NPS calculator · customer feedback software · customer satisfaction survey software · survey analysis software
Brief the agent and field a real NPS study: the score, its margin, the coded reasons behind it. From $99 per study, $99 per wave, no subscription.
Start an NPS study