# NPS Analysis: Find the Themes Behind Your Detractors

Explain your NPS score with its period, trend, and respondent split, then find the themes and quotes that drive detractors. Run it in Deck or with your agent.

- URL: https://getdeck.io/playbooks/nps-breakdown
- Author: Matt Teixeira
- Published: 2026-11-01
- Updated: 2026-11-01
- Stage: Decide

## How Do You Analyze NPS Results?

Start with the score and the period it covers, then the trend and the split of promoters, passives, and detractors. Next, find the themes that drive detractors and read their verbatims. Check what keeps passives from promoting and what promoters value. A score without the themes behind it does not tell you what to fix.

## Why an NPS Score Alone Does Not Tell You What to Fix

The score dropped and leadership wants to know why. The survey tool shows the number and a wall of comments. Someone skims a few, picks a likely cause, and the explanation becomes whatever was read last.

## How Deck Links NPS to the Themes Behind It

Deck detects NPS responses in your feedback, sorts respondents into promoters, passives, and detractors, and links each response to your product themes. It shows the score with its period, the trend, the top themes for each group, and a weekly synthesis with recommendations. You can check whether a detractor theme is also large outside NPS, which makes it a stronger signal.

Deck features used:

- [NPS](https://docs.getdeck.io/docs/features/nps-score)
- [Themes](https://getdeck.io/product/themes)

## How to Run an NPS Analysis in Deck

### 1. Make Sure NPS Analysis Is On

Deck turns NPS analysis on when it first finds NPS feedback, on the Starter plan or higher. An admin can also turn it on in Settings → Organization → NPS Score Analysis.

### 2. Read the Score and Its Period

Open Visualize → NPS Score. The Summary shows the overall score and trend, the response distribution, and the NPS over time chart. Each point on the chart is a rolling 30-day score, so always name the period.

### 3. Find What Drives Detractors

Open the Detractors view for respondents who scored 0 to 6. Read the top themes and concerns, then the individual responses behind each one. These themes are your main drivers.

### 4. Check Passives and Promoters

Open the Passives view to see what keeps people at 7 or 8. Open the Promoters view to see what people who score 9 or 10 value, so you know what to keep.

### 5. Check Each Driver Outside NPS

Open each top detractor theme in Visualize → Themes. A theme that is also large in support tickets, interviews, or Slack is a stronger signal than one that only shows up in NPS.

### 6. Read the Recommendations and Write It Up

Read the Summary's insights on converting passives and winning back detractors, then fill in the template below. Note that the synthesis runs weekly, so a new response can appear in the counts before it appears in the synthesis.

## Break Down Your NPS with Your AI Agent

You need:

- Deck MCP connected to your agent
- Customer feedback already in Deck
- Deck Starter plan or higher

Your agent starts with the `explore_nps` tool.

Paste this prompt into your agent:

```text
Use Deck MCP. Fetch and follow https://docs.getdeck.io/docs/agents/workflows/nps-breakdown.md
```

Full agent workflow: https://docs.getdeck.io/docs/agents/workflows/nps-breakdown.md

## NPS Analysis Template

### NPS Analysis Write-Up

```text
## NPS breakdown: [period]

Score: [Deck score] ([rolling 30 days / latest synthesis]). All-time: [Deck score]
Trend: [Deck change], [direction]. Drivers (only if Deck gives them): [driver]

| Group | Count | Share |
|---|---|---|
| Promoters (9-10) | [Deck count] | [Deck share] |
| Passives (7-8) | [Deck count] | [Deck share] |
| Detractors (0-6) | [Deck count] | [Deck share] |

Detractor drivers:
1. [Theme] — [Deck count]. "[quote]" "[quote]" ([Deck link])
2. [Theme] — [Deck count]. "[quote]" "[quote]" ([Deck link])

Passives need: [theme], [theme]
What to keep: [promoter theme], [promoter theme]
Recommendations: [Deck's recommendations, short form]

Caveats:
- [Synthesis timing, org-wide parts, small themes]
```

## Frequently Asked Questions

### Which plan do we need?

NPS analysis needs the Starter plan or higher. On the Free plan, your agent offers the most negative themes instead.

### Where does Deck find NPS responses?

From PostHog surveys, Slack messages in synced channels, CSV uploads, survey tools such as Typeform, SurveyMonkey, Qualtrics, Hotjar, Sprig, and Fillout, and API imports.

### Can it tell us why our NPS dropped?

Deck names drivers only when the score went up. For a drop, the explanation comes from the detractor themes and the weekly respondent shares. The agent does not invent a cause.

### Why is a new response missing from the synthesis?

Deck builds the NPS synthesis once a week, usually ready on Monday. A new response can be in the counts before it is in the synthesis.

### Can we break NPS down by segment?

Yes. Use the Segment filter on the NPS Score page. Some parts stay org-wide, such as Deck's recommendations, and the write-up should say which.

### Can my AI agent run this?

Yes. With Deck MCP connected, paste the prompt on this page. The agent reads the score, trend, and each respondent group, then writes the breakdown with quotes and Deck links. It only reads, and listed responses have no respondent name or email.

## Related Playbooks

- [Weekly Customer Feedback Report: What Changed This Week](https://getdeck.io/playbooks/weekly-signal-brief.md): Write a weekly brief of new feedback, growing themes, negative signal, and NPS movement, using Deck's own numbers and quotes. Run it in Deck or with your agent.
- [Customer Segment Comparison: What Each Group Asks For](https://getdeck.io/playbooks/segment-comparison.md): Compare what two customer segments ask for, such as enterprise and SMB, with shared themes, leading themes, and quotes. Run it in Deck or with your agent.
- [Prioritize Your Product Backlog with Customer Evidence](https://getdeck.io/playbooks/prioritize-backlog.md): Pick the top opportunities from a backlog ranked on customer evidence, with a reason, a risk, and quotes for each pick. Run it in Deck or with your AI agent.