NPS for B2B SaaS: The Complete Guide

NPS is a one-question loyalty metric introduced by Frederick Reichheld of Bain & Company in a December 2003 Harvard Business Review article titled "The One Number You Need to Grow".

The question is simple: "How likely is it that you would recommend [company or product] to a friend or colleague?" Customers answer on a scale from 0 (not at all likely) to 10 (extremely likely). Reichheld's research compared survey answers with what customers actually did, such as repeat purchases and referrals, and found that this question was the best single predictor in most of the industries studied.

Two decades later, NPS is one of the most widely used customer metrics in software. It is also one of the most criticised.

A frequent critique targets not the question itself but the idea that one number is enough to run a business. That critique matters even more in B2B, where a handful of people decide whether an account renews.

The practical takeaway: treat NPS as a listening system, not a scoreboard. The score tells you where to look. The comments, and what you do next, are where the value is.

NPS = % of promoters − % of detractors. Every answer falls into one of three groups:

Passives are not part of the subtraction, but they stay in the total. That is why moving a customer from 8 to 9 raises your score, and moving one from 7 to 6 lowers it.

Worked example. You receive 200 answers this quarter: 90 promoters, 70 passives and 40 detractors.

• Promoters: 90 ÷ 200 = 45%
• Detractors: 40 ÷ 200 = 20%
• NPS = 45 − 20 = +25

The score always falls between −100 (every respondent is a detractor) and +100 (every respondent is a promoter). It is not a percentage, so write it as "+25", not "25%".

Two habits make the number trustworthy. First, always show the number of responses next to the score: "+25 (200 responses)" means something; "+100 (2 responses)" does not. Second, compare the score with your own previous periods before comparing it with anyone else's.

In B2C, one customer is one person. In B2B SaaS, one customer is an account: a company with several people who use, administer, pay for and renew your product. That changes how you read NPS in three ways.

1. The people who answer are not all equal. An end user, the admin who configures your tool and the executive who signs the renewal can give three different scores. The renewal decision usually sits with very few of them.

2. Samples are small. A B2B SaaS company with 300 accounts may get 150 answers a quarter. At that size, a few responses move the score by several points. An account-level score built on two or three answers is a signal to investigate, not a statistic.

3. One detractor can outweigh the average. Take an account with six admins: five answer 9 or 10, one answers 3. The account's NPS is +67, which looks excellent. But if that one detractor is the person who owns the budget, the account is at risk.

So in B2B, track NPS at two levels:

A simple rule works well in practice: flag an account as "at risk" as soon as it has one detractor, whatever its overall score, and keep the flag until someone has followed up. Your customer success team then works from a short list of accounts, not from an average.

Published figures vary widely, and they say as much about who was asked as about the software itself.

Note what was measured: users rating the tool they work in every day. Customers rating the vendor relationship as a whole can give quite different scores. In B2B SaaS you have both groups, often inside the same account.

Treat these figures as a rough frame, not a target. Benchmarks differ in sample, survey channel and the kind of respondents included, so two serious studies can publish scores ten points apart for the same "industry".

Three better questions than "is my score good?":

1. Is it going up or down compared with my last three periods?
2. Which accounts and roles pull it down?
3. What do detractors say, in their own words, and does the same reason come up again and again?

If you must pick a working reference, a score above your previous period and above +30 is a reasonable first goal for a B2B SaaS product. The trend matters more than the absolute value.

Use both, but for different people, and never both on the same person in the same period.

In B2B, the email channel is often underrated. The admin who has not logged in for two months will never see your in-app widget, yet that silence is exactly what you need to understand before the renewal.

Four rules make both channels work:

• Put the 0–10 scale in the email itself, so a single click starts the answer.
• Send from your own domain, with the sender a customer recognises. Security-minded buyers ignore emails from unknown domains.
• Never show the in-app survey on the first login or in the middle of a critical task. Wait until an account has used the product for a few weeks.
• Apply one contact limit across channels. Someone who answered by email should not see the widget the next day.

Many companies protect their inboxes with email security gateways. Some of these tools check the links in incoming emails before or when the recipient clicks, to detect phishing. Microsoft Defender for Office 365, for example, rewrites and checks links through its Safe Links feature.

This creates a quiet problem for one-click NPS emails. If each score from 0 to 10 is a link that records the vote, an automated check that opens the links can record answers nobody gave. Your B2B customers, especially in regulated or security-conscious sectors, are exactly the ones most likely to run these tools.

Three safeguards keep your data clean:

1. Confirm on the landing page. The click opens a page with the score preselected; the answer is saved only after a human action, such as pressing "Send".
2. Discard suspicious patterns. A real person clicks one score. If the same email opens several score links within a few seconds, ignore those clicks.
3. Use an invisible trap link. Add a link no human can see. If it is opened, mark that email's clicks in the following seconds as automated.

Before your first campaign, test it: send the survey to a mailbox protected by the same kind of gateway your customers use, and check that no answer is recorded without you.

There are two kinds of NPS, and they answer different questions.

• Relationship NPS asks about the product and the company as a whole. It tracks loyalty over time. In B2B SaaS, surveying each contact about every six months is a reasonable starting rhythm: often enough to see a trend, rare enough not to tire anyone.
• Transactional NPS follows a specific moment: the end of onboarding, a support ticket closed, a major feature released. It tells you how that moment went, not how loyal the account is.

Whatever the rhythm, a few timing rules apply:

• Not too early. Wait until a new account has had time to get value, often 30 to 90 days after go-live depending on your onboarding.
• Not before a renewal negotiation you already know is difficult; you will measure the negotiation, not the product.
• On a working day, in the contact's time zone. An email that lands at 3 a.m. on a Sunday sinks under Monday's inbox.
• One limit per person, across all surveys. If someone answered a transactional survey last week, skip them in this month's relationship survey.

For large bases, spread a campaign over several weeks instead of sending everything on one morning. Your team can then follow up with detractors as answers arrive, instead of facing a wave of alerts at once.

The score tells you how people feel; the follow-up question tells you why. Keep it to one open question, two at most, and make it optional so nobody abandons the survey.

The default works for almost everyone: "What is the main reason for your score?"

You can tailor it to the group the person just joined:

Three mistakes to avoid:

• Leading questions such as "What do you love about us?" They produce flattering comments and little insight.
• Long forms. Every extra field lowers the completion rate. If you need detailed feedback, ask a few volunteers for a call instead.
• Questions you will not act on. Asking about a roadmap item you have no plan to build creates expectations you will disappoint.

An NPS programme earns its keep after the survey, not during it. Customers who answer and hear nothing back stop answering. Here is a simple playbook for B2B teams.

Write the score where your team already works. Most account managers live in their CRM, not in a survey tool. Push each score and comment to the contact and company records, so the information is in front of them before every call.

Be careful with public reviews. It is tempting to ask only promoters for a review on G2, Capterra or Google. Platforms restrict this practice, often called review gating: Google's policies, for example, forbid selectively soliciting positive reviews (Google Maps User Generated Content Policy).

The safe approach is to invite all your customers to leave a review in the same way, and to keep private feedback separate. Referral invitations are a different matter: inviting your promoters to recommend you to a peer is not a public review.

Close the loop publicly too. When a theme comes up again and again and you fix it, tell the customers who raised it. "You told us, we did it" is one of the strongest ways to turn a passive into a promoter.

B2B does not mean outside the GDPR. A business email address such as firstname.lastname@company.com identifies a person, so it is personal data, and so are that person's score and comments (European Commission, what is personal data).

This is not legal advice, but these are the questions European B2B teams usually need to answer before running NPS:

• Why are we allowed to survey these contacts? Identify the legal basis for the processing and document it.
• Have we told them? Your privacy notice should mention customer satisfaction surveys, and each survey email should say who sends it and how to stop receiving them.
• How long do we keep the answers? Set a retention period and apply it.
• Who else processes the data? If a survey tool sends the emails and stores the answers, it acts as your processor: you need a data processing agreement with it, and you should know where the data is hosted.
• Do we show personal data to everyone? Many people in your company need the scores; fewer need the names and email addresses behind them.

For the reference texts and guidance, start with your national data protection authority, such as the CNIL in France, and ask your legal adviser to confirm your setup.

Gradiamo is a new NPS and customer feedback tool for B2B SaaS teams, built in Europe. Join the waitlist to hear when it opens.

• Frederick F. Reichheld, "The One Number You Need to Grow", Harvard Business Review, December 2003
• Business Software UX & NPS Benchmarks (2025), MeasuringU
• Safe Links in Microsoft Defender for Office 365, Microsoft Learn
• Prohibited & restricted content, Maps User Generated Content Policy, Google
• Data protection explained: What is personal data?, European Commission