> ## Content Index
> Fetch the complete content index at: https://pulsecorp.co/blog/llms.txt
> Use this file to discover other available public pages before exploring further.

# How to explain a drop in NPS, step by step
- URL: https://pulsecorp.co/blog/how-to-explain-a-drop-in-nps/
- Published: 2026-11-14T12:00:00.000Z
- Updated: 2026-11-15T02:56:49.000Z
- Description: How to explain a drop in NPS: check the survey, find which customers moved, test what changed and interview them so you can report the real reasons.
- Author: Khyati Ojha

It's Monday morning at the music streaming service company where you work. The new customer survey score has just come in, and it's lower than last time. Your manager walks over with one question: "Was it the app update?" You don't know yet. So how do you find out?

Working that out is how you explain a drop in NPS.

A lower score can only come from three places. The survey changed, a group of customers changed their minds, or something changed for everyone. Check them in that order, then talk to the customers who moved.

The order matters, because a drop that the survey caused needs no interviews at all.

## The short version

- A survey score can fall because the survey changed, not because customers did.
- Look for the group of customers where the drop sits. That group tells you who to talk to.
- A change that happened at the same time is a lead, not a reason, until customers say so.
- The comment box tells you what people dislike. Interviews tell you why.
- Talk to a few customers whose score held steady, so a common complaint doesn't pass for the cause.
- Report each reason with how many people raised it, which group, and one quote.

## Why a survey score drops in the first place

When a score falls, it's natural to look for something that went wrong. But a score is a reading, a bit like a thermometer. If someone moves the thermometer next to the oven, the number goes up, even though the room is no warmer. A survey works the same way.

NPS comes from one survey question: how likely are you to recommend us? People who answer high are called promoters, people in the middle are passives, and people who answer low are detractors. The score goes down when promoters slip, when passives fall, or when detractors grow.

So a lower score has three possible sources:

1. The survey itself changed, so different people answered, at a different time, or to different words.
2. One group of customers changed their minds, because something went wrong for them in particular.
3. Something changed for everyone, such as a price rise, an outage or an update that every customer touched.

Each source sends you to a different fix. That is why the first job is to rule out the survey.

## First, make sure the drop is real

Before you blame the app update, check whether the survey itself changed. If it did, you may be comparing two different things, and the drop may shrink or disappear when you compare like with like. Four quick checks tell you.

### 1\. Count the answers

Fewer answers make any score swing more. A small round of survey answers can move a lot on its own. So put this quarter's count next to last quarter's before you read anything into the gap.

### 2\. Check who got the survey, and when

Say the survey used to go out by email to every subscriber. This quarter, it went out inside the app, right after people contacted support. Those people had just had a problem, so you'd expect a lower score from them. In that case the drop is about the survey, and the update may be innocent.

### 3\. Check the wording

A new question, new labels on the scale, or a survey moved to the end of a long form can all change how people answer. Lay the two versions side by side and look for any difference.

### 4\. Keep the two kinds of survey apart

A relationship survey asks how customers feel about your company overall. A single-contact survey asks about one moment, such as one support call. If this quarter mixes the two, the score is mixing two different questions.

If all four checks pass, the drop is real, and the next question is who moved.

## Find the customers who moved

A score for everyone hides who changed. So split the answers into segments, which just means groups of customers who share something. For the streaming service, useful groups could be:

- how long people have subscribed
- which plan they use, such as music only, family or student
- which country they live in
- which device they use, and how they signed up
- whether they contacted support this quarter

If the drop sits in one group, that group is where your interviews go. If it's spread evenly across every group, look for something every customer touched.

Next, write a short timeline of what changed at the company this quarter. The app update, any price change, songs that left the catalog, outages, new rules on shared accounts. Put each one next to the survey dates.

A change that lines up with the drop is a lead to test, not a reason. Until customers confirm it, say the drop coincided with the update. Save the word "caused" for after the interviews.

Then read the comment box answers from the people who moved. Most will be short, like "bad app" or "too expensive." They tell you what people are unhappy about, but rarely why. That gap is what the [interviews behind the score](https://pulsecorp.co/blog/surveys-tell-you-what-voice-interviews-tell-you-why-the-question-your-nps-score/) are for. If your team already reads those comments, there's more on [turning detractor comments into reasons](https://pulsecorp.co/blog/replacing-the-nps-verbatim-box-turn-detractor-scores-into-15-real-reasons-this-w/).

![Decision tree for an NPS drop: if the drop is in the survey, fix the survey; if it is in one group, interview that group plus a few who held; if it is everywhere, look at what every customer touched](https://pulsecorp.co/blog/content/images/2026/11/how-to-explain-a-drop-in-nps-v8f8t3.webp)

## How the streaming team would find the real reason

Back to your Monday morning. Say the survey checks pass, and the drop sits mostly with long-time family-plan subscribers. The app update went out two weeks before the survey. The comment box says "bad app" and "annoying now." You have a lead, but not a reason yet.

So who should you interview? The people who moved. Invite family-plan subscribers who now score lower, and who agreed to be contacted when they took the survey. Also invite a few family-plan subscribers whose score held steady, so you have something to compare against.

Invite them without pressure. Tell them plainly that you want to learn what happened, and never ask anyone to change their score or offer a reward for a better one.

In each interview, you ask about real recent moments. When did they last use the app? When did their opinion change? What did they do next? What would win back their old score? A real moment, like "last Tuesday my kid's playlist showed up on my profile," is far easier to act on than "it's worse now."

You also hold back your own guess. A leading question suggests the answer, such as "Did the new update make things worse?" A neutral question lets them choose, such as "Tell me about the last time you opened the app." If the update comes up on its own, that tells you far more.

Now say both groups grumble about the new layout. Then the layout is a common complaint, but it probably isn't why some scores fell. And say only the people who moved keep describing the same trouble with shared family profiles. That is the reason you were looking for.

You'd report it to your manager in one line: what customers said, how many raised it, which group it explains, and one quote that shows it. A hunch that only two people mentioned stays labelled as a lead.

Then you agree how you'll know a fix worked. This quarter's score is your baseline, the number you measure change from. In the next survey round, ask the same question and compare the family-plan group against that baseline.

If the drop comes from the survey, fix the survey. If it sits in one group, talk to that group. If it's everywhere, look at what every customer touched.

![Three cards showing who to interview after an NPS drop: new detractors from the group where the drop sits, promoters who slipped, and a few customers whose score held](https://pulsecorp.co/blog/content/images/2026/11/how-to-explain-a-drop-in-nps-x95ty9.webp)

## Where Pulse fits

The slowest part of explaining a drop is usually the interviews. You have to invite the customers who moved, ask each one the same good questions, and turn their answers into reasons before your manager stops waiting. That is the part [Pulse](https://pulsecorp.co/product/ai-interviews/) is built for.

You upload the list of customers who moved and agreed to be contacted, and Pulse runs AI-moderated voice interviews with them. Everyone gets the same clear questions from one guide. When an answer is vague, the AI moderator asks a second question, building on what the person said earlier. So "it's worse now" turns into a reason you can use.

[Trackers](https://pulsecorp.co/product/trackers/#:~:text=In%20Pulse%2C%20trackers%20repeat,interviews%20can%20explore%20it.) repeat the same questions each quarter, so next quarter's result sits next to this one. Answers are grouped into themes with the quotes behind them, and each quote links to its full interview. Your part stays the same: asking the right question, reading the surprises, and making the call.

To try it on your own customers, Pulse offers a free trial and a free study, and you can [book a demo](https://pulsecorp.co/demo/) to see how a study runs.

## Questions readers ask

### 1\. Is a small drop in NPS worth looking into?

It depends on whether the drop is bigger than your score's normal ups and downs. Look at how much the score moved over the last few rounds when nothing changed. If this drop sits inside that range, wait for the next round. If it's bigger, or sits in one group of customers, look into it now.

### 2\. How many interviews do you need to explain a drop?

There is no fixed number. You need enough to hear the main reasons repeat within each group that moved, plus a few customers whose score held steady to compare against. A drop in one group needs fewer interviews than a drop spread across several plans, countries or devices.

### 3\. Should you contact customers who gave you a low score?

Yes, if they agreed to be contacted when they took the survey. Reach out soon, while the experience is fresh, and say plainly that you want to understand what happened. Never offer a reward for a better score, because that changes the very answer you're trying to learn from.

### 4\. What does it mean if your NPS is low?

A low score means detractors outnumber promoters by a lot, but it doesn't say why. The level alone is hard to read without knowing how your own score usually moves. A change in your own score, explained by your own customers, tells you more than the level does.