LOCAL

Positive review responses at scale

Use when you have a backlog of good reviews and need replies that do not all read identically.

positive-review-responses.md

Download .md

Write owner responses to positive reviews for a local business.

Business: {{BUSINESS_NAME}}
Person signing the replies: {{SIGNER_NAME}}
Tone: {{TONE}}
Reviews, one per line: {{REVIEWS}}

For each review, write one reply.

Rules:
- 25 to 55 words. Shorter is fine where the review is short.
- Reference something specific the reviewer actually mentioned. If the review is a bare star rating with no text, write a short generic thank you and say so.
- Do not begin two replies with the same word across the whole batch. Vary the sentence shape.
- Use the customer's first name only where the review shows a first name.
- Mention the service by name where the reviewer named it. Do not add keywords or the city name where the review did not raise them - forced keyword stuffing in replies is obvious to readers.
- No offers, no discount codes, no requests for more reviews.
- No em dashes. No exclamation marks beyond one per reply at most. No "we're thrilled". No "it's not just X, it's Y".

Output as a numbered list: the review first in italics-free plain text, then the reply beneath it.

Fill in before running

Replace each placeholder with your own detail. The more specific you are, the less the model invents.

{{BUSINESS_NAME}}
{{SIGNER_NAME}}
{{TONE}}
{{REVIEWS}}

Getting a better result

Paste ten or twenty reviews at once so the model can actively vary openings across the batch.
Set the signer name to a real person - replies signed by a named owner read very differently from "The Team".
Read the batch through before posting and change any reply that mirrors the wording of another.

Questions about this prompt

When is this worth it rather than just replying as reviews come in?

When you have a backlog. Replying to one review needs no help. The value here is that the model sees the whole batch at once and can vary openings and sentence shapes across it, which is exactly what you cannot do reliably writing replies one at a time over a fortnight.

What should I paste in?

Ten or twenty reviews, one per line, into {{REVIEWS}}, a real person in {{SIGNER_NAME}}, and a {{TONE}} you can sustain. Include the bare star ratings with no text: the prompt writes a short generic thank you for those and tells you it did so, rather than inventing detail the reviewer never gave.

What comes back, and how should I read it?

A numbered list with each review followed by its reply, 25 to 55 words, each referencing something the reviewer actually mentioned. Read it as a batch rather than reply by reply, because the thing you are checking is whether any two share an opening word or a sentence shape. Change the ones that do.

What is the mistake here?

Adding the city name or the service keyword back into replies where the reviewer never raised them. It is obvious to readers and does nothing for you. The prompt bans it, so the mistake is yours at the editing stage. Signing as The Team rather than a named person costs you the rest.