How to use Pendo NPS: setup, segments, and what to do next

You turned on Pendo NPS, the score came back a 34, and nobody in the room can say whether that is good, bad, or a rounding error. That is the normal first reaction, and it is the point where most teams stop. The score becomes a tile on a dashboard, the follow-up comments go unread, and by the third quarter the survey is still running for an audience nobody has looked at since launch.

I have seen the other version too. A fintech client I started working with this fall was heading into a Pendo renewal with no clear owner for the tool and two internal teams that did not talk to each other. The first thing we shipped to earn the renewal was not a dashboard. It was an NPS survey and two short questionnaires, built in their QA environment, with a named owner for each one. The survey was the fastest way to put something in front of leadership that Pendo had produced. Here is how to set one up so it does that for you.

Check your contract before you check the settings

Pendo’s own setup article is clear on this: NPS is a standalone feature on free subscriptions only. On a paid subscription, NPS and every other survey type come through Sentiment, which has to be included in your contract. If you open Pendo and cannot find NPS where the Set up an NPS survey article says it should be, that is usually why, and no amount of clicking fixes it. Ask your Pendo account team what your contract covers before you plan a survey program around it.

While you are there, confirm who the admin is. NPS surveys are created by people with the right role, and the person who signed the contract is often not that person.

Set up Pendo NPS so the answers mean something

The mechanics take an afternoon. The choices inside them decide whether you get a signal or a number. Four of them matter more than the rest.

Pick an audience that has used the product

The default segment is everyone. Pendo recommends visitors who have been in the app for at least 30 days, and I would go further for B2B products: people who have reached the moment where your product does its job. Someone who signed up Tuesday and never finished onboarding is guessing, and their guess drags your score around. Build the segment from usage, not from account age alone.

Keep the question, write the follow-up

Leave the NPS question alone. Its whole value is that it is the same question every product asks, so your 34 can be compared with your own 38 last year. Spend your effort on the follow-up. Pendo lets you use one universal follow-up or a different one for detractors, passives, and promoters. Use the split. Ask detractors what got in their way and promoters what they would miss. Those two answers are the material the rest of this post depends on.

Set the cadence once and leave it

Pendo can repeat the survey every 30, 60, 90, 180, or 365 days, or run it once. Quarterly is the common choice for B2B SaaS, and the right choice is whichever one you will not change. Every time the cadence moves, the trend line breaks and you lose the ability to say the score fell after a release or a price change. Turn on even distribution so the survey spreads across the window instead of hitting every eligible visitor at sign-in on day one.

Decide about email

In-app only is the simpler setup. In-app plus email sends a follow-up to visitors who did not see the survey in the app within 14 days, which helps response rates for products people do not open every week. The fintech client took the email option because their users only log in when they have a transaction to make. If you take it, you need the recipient email in your metadata and a sender name people recognize.

Read the comments before the score

The score is a subtraction: promoters (9 to 10) minus detractors (0 to 6), as percentages. It moves for reasons that have nothing to do with your product, including who happened to be eligible that quarter. The comments do not have that problem. Read every detractor comment the week the survey closes. With a few hundred responses a quarter you will see themes; with 15 you are reading anecdotes, and you should say so to whoever asked for the report.

Three things to look for in the first pass:

• The same complaint from detractors in different accounts. One account with a bad month is noise. Four accounts naming the same screen is a backlog item.
• Promoters and detractors describing the same feature. If both groups mention your reporting module, it works for one kind of user and not another, and your segments will tell you which.
• Timing. Lay the response dates next to your release calendar and your support incident log before you explain a dip.

If you have hundreds of comments, I wrote up how we sort open text survey responses into themes without reading them one at a time. The method does not change for NPS.

Give every theme an owner and a decision

A trend with nobody attached to it becomes a slide. The fix is boring: one person owns the survey, and one person (often the same one) owns the follow-up. Their job is not to fix everything. It is to make sure every theme gets one of three decisions, written down where the team can see it.

Copy that table into whatever tool your team already uses. The last column is the one teams skip, and it is the one that moves the next score. A detractor who hears back is a different person at the next survey than one who did not.

This is also where the governance basics for a small Pendo team earn their keep. If the survey owner leaves and nobody knows the cadence, the segment, or where the decisions were written, the program resets to zero.

Put the score next to the behavior

NPS is one signal. The reason to run it inside Pendo rather than in a separate tool is that the responses sit next to the usage data. A detractor who uses the product daily is telling you something different from a detractor who logged in twice. Build one segment of detractors and look at their feature usage, then do the same for promoters. That comparison is usually the first real insight the program produces, and it is the thing to show leadership instead of the score.

Pair it with support ticket themes. If your detractors are also your loudest tickets, that overlap goes to the top of the list. Our Pendo analytics and reporting work usually starts with exactly that join, because it turns a survey into a priority list in one meeting.

Questions people ask

What is a good Pendo NPS score?

There is no universal good number. B2B SaaS products commonly land in the 30s and 40s, but the useful comparison is your own last two or three readings. A 15 to 20 point swing in either direction is worth investigating; a 3 point move with a few dozen responses is noise.

How often should I run an NPS survey in Pendo?

Quarterly works for most B2B products: long enough for changes to show up, short enough to tie a move to a release. Pendo supports 30, 60, 90, 180, and 365 day recurrence. Pick one cadence and keep it, because changing it breaks the trend line.

Do I need Pendo Sentiment to run NPS?

On paid subscriptions, yes. Pendo’s setup documentation states that NPS and other survey types come through Sentiment, which must be in your contract, while free subscriptions get NPS as a standalone feature. Check your contract before you plan the program.

If your survey is running and nobody has read the comments yet, that is the place to start, and it takes one afternoon. If you would rather have someone who has run this program before set it up with you, a free Pendo strategy call is the easiest way to find out what the first quarter should look like.