As of 13 August 2026, AI can explain why your NPS has changed.

This still needs a person who signs their name to it.

Can you do it?

15 minutesto a draft.

30 minutesto something you’d act on.

Cost, all in£0

Skill neededchat-fluent

Who has to check ita colleague

What the alternative costsA human analyst or customer-insight team is the alternative; no price is stated in the supplied tool data.

If this goes wrong: you mistake correlation for a cause and spend time or money fixing the wrong part of the customer experience.

What to actually do

Hand it to a person
The route this page recommends
A person who owns the outcome does this end to end, worth it when the failure is dear.
Use a tool built for this
Second choice

Do it yourself

The distant third

A chat interface, chat-fluent skill, and roughly 30 minutes until you can act on the result.

How to actually do it

Open the survey or customer-feedback export and record the NPS, response count, promoter, passive and detractor counts or percentages for the earlier and later periods.
Copy the exact survey question, scoring method, survey dates, channel, audience and any segment fields into a working document.
Export or paste the open-ended responses, keeping the period and customer segment attached to each response where possible.
Gather a dated list of changes between the two periods, including product releases, pricing, service levels, policies, staffing, customer mix and survey-method changes.
Paste the collected information into the prompt and ask the chatbot to separate evidence-supported explanations from hypotheses and unknowns.
Check every reported figure against the survey export, trace each claimed driver to the cited comments or segment results, then ask a colleague to challenge the leading explanations before sharing the briefing or changing customer-service priorities.

Prompt

Explain why our NPS changed between [EARLIER PERIOD] and [LATER PERIOD]. Use only the information supplied below and do not invent causes, figures or customer views.

NPS results:
- Earlier period: [NPS, number of responses, promoter count or percentage, passive count or percentage, detractor count or percentage]
- Later period: [NPS, number of responses, promoter count or percentage, passive count or percentage, detractor count or percentage]
- Survey question and scoring method: [PASTE]
- Survey dates, channel and audience: [PASTE]
- Results by useful segment, such as product, region, customer type or journey stage: [PASTE]
- Open-ended feedback, with the period and segment for each comment where available: [PASTE]
- Relevant changes between the periods, such as pricing, product releases, service levels, policies, staffing or survey method: [PASTE]

First check the NPS arithmetic and identify any change in response volume or sample composition. Then compare the periods and group the feedback into recurring themes. Produce:
1. The strongest evidence-supported explanations, ranked by strength.
2. The figures, segment differences and representative comments supporting each explanation.
3. Alternative explanations and confounding factors, including survey-method changes and response bias.
4. A clear distinction between observed evidence, plausible hypotheses and claims that cannot be established from these data.
5. Three follow-up checks that would help test the leading explanations.
6. A short briefing for a customer-service or leadership meeting.

Do not claim that a factor caused the NPS change unless the supplied evidence establishes that. Flag small, missing or incomparable samples and say when the data is insufficient.

Open it prefilled in ChatGPT or Claude, or copy it into Gemini, which takes no prefill link.

What it gets wrong

AI cannot establish causation from a change in NPS and observational feedback alone.
AI cannot know which operational changes were material unless you provide that context.
AI can over-weight vivid comments or frequent words when the feedback is not representative of the whole responding population.
AI cannot decide whether a statistically uncertain movement is important for your commercial or customer-service priorities.
AI cannot replace a properly designed survey, sampling plan or causal analysis when the decision carries substantial cost.

Even on a YES, the friction has a name: judgement under ambiguity, context depth and verification cost.

How we scored this

Five axes, each scored nought to two by hand: ten means AI carries the task cleanly, and the thresholds that turn a total into YES, PARTLY or NO are published in the methodology. Each axis name links to its definition.

Axis	Score (0–2)
Output	2
Inputs	2
Verification	1
Liability	2
Effort delta	2
Total	9 / 10

The methodology and its thresholds are published in full.

FAQ

Can ChatGPT tell me why my NPS dropped?: It can compare the periods, group customer comments and identify plausible drivers from the data you provide. It cannot prove that any one factor caused the drop, so treat the result as an evidence-ranked explanation rather than a finding of fact.
What data do I need to give AI to analyse NPS?: Give it the NPS and response counts for both periods, the survey wording and method, segment results, open-ended comments and relevant changes to your product or service. Include response volume and customer mix because a change in who answered can look like a change in satisfaction.
Can AI analyse NPS comments?: Yes. It can group comments into themes, compare themes between periods and link them to segments when those labels are present. Check the grouping against the original comments because it can miss context, sarcasm or multiple issues in one response.
Can I trust an AI explanation of an NPS change?: You can use it as a structured first analysis, not as proof of the cause. Check every figure and quoted comment against the source data, ask a colleague to challenge the interpretation and run follow-up checks before making a costly change.

Nearby answers

All Feedback & NPS analysis rulings

Assessed by gpt-5.6-luna (gpt-5.6-luna) on 2026-08-13, second-checked by an independent model. Wrong somewhere? Email [email protected] and it gets re-checked.

The newsletter

AI news, new answers and product picks, straight to your inbox.