Home · Business · Customer Service · Feedback & NPS analysis

YES

As of 13 August 2026, AI can identify themes in your customer feedback.

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

Can you do it?

15 minutesto a draft.

1 hourto something you’d act on.

Cost, all in£0

Skill neededchat-fluent

Who has to check ityou

What the alternative costsThe alternative is manual spreadsheet coding or a purpose-built workflow such as Botpress; no price is supplied here.

If this goes wrong: important complaints are hidden inside broad labels, so you make service decisions from a tidy but misleading picture.

What to actually do

  1. 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.

  2. Use a tool built for this

    Second choice

  3. Do it yourself

    The distant third

    A chat interface, chat-fluent skill, and roughly 1 hour until you can act on the result.

    How to actually do it

    1. Open the feedback export and remove names, email addresses, telephone numbers, order numbers and any other details that identify a customer.
    2. Add the exact survey question or feedback prompt, collection period, product or service, and any legitimate segment or channel information that should affect interpretation.
    3. Give every response a stable ID such as R001, R002 and R003, then paste the cleaned rows into a chatbot with the supplied prompt.
    4. Ask the model to identify themes, subthemes, sentiment and representative response IDs without adding causes or customer characteristics that are not stated.
    5. Compare every quoted example and response ID in the result with the original export, correcting any misquote, duplicate, missing response or unsupported inference.
    6. Combine or split labels that describe the same issue, then ask a colleague who understands the service to challenge the definitions and the comments assigned to each theme.
    7. Save the final theme list with its supporting response IDs and use it as an input to a separate decision about priorities, owners and follow-up actions.

    Prompt

    Identify the main themes in the customer feedback below. Treat each row as one response and do not invent information. First remove or mask names, email addresses, telephone numbers, order numbers and other identifying details from your analysis. Then produce: 1) a table of themes with a short name, plain-English definition, number of responses, percentage of all responses, sentiment, and three representative response IDs; 2) subthemes where they are genuinely distinct; 3) the strongest positive and negative examples for each theme, quoting only the supplied text; 4) contradictory or ambiguous comments; 5) comments that do not fit the proposed themes; and 6) a short list of questions I should answer before acting on the findings. Keep similar themes together, do not treat a single vivid comment as a major theme, and do not infer causes, demographics or intent that the feedback does not state. If the data is incomplete or the response IDs are missing, say so. Feedback context: [what was asked, when it was collected, product or service, and relevant customer segment]. Feedback: [paste the export here]

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

What it gets wrong

  • AI cannot decide whether two differently worded complaints have the same underlying cause without your service context.
  • AI cannot tell you which theme matters most unless you supply a defensible measure of impact, urgency or customer value.
  • AI can over-weight vivid comments and understate rare but serious issues, even when its categories look consistent.
  • AI cannot replace a colleague who knows the history behind changes in your product, policy or customer base.

Even on a YES, the friction has a name: judgement under ambiguity, verification cost and stakes of error.

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.

AxisScore (0–2)
Output2
Inputs2
Verification1
Liability1
Effort delta2
Total8 / 10

The methodology and its thresholds are published in full.

FAQ

Can AI analyse customer feedback?
Yes. It can group written responses into recurring themes, identify subthemes and provide example comments. You still need to check the grouping against the original feedback before using it to make decisions.
How do I use AI to find themes in customer feedback?
Clean identifying details from the export, give each response an ID, and include the exact question and service context. Ask the model for theme definitions, supporting response IDs, exceptions and unsupported assumptions, then compare its output with the source data.
Can AI analyse anonymous survey responses?
Yes, if you can provide the response text and enough context to interpret it. Remove indirect identifiers as well as names and contact details, and do not let the model infer sensitive characteristics that respondents did not provide.
How accurate is AI at identifying feedback themes?
There is no fixed accuracy figure for this task because results depend on the wording, volume, context and consistency of the feedback. Check every theme against the original responses and have someone with service knowledge challenge the categories before acting on them.

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.