Verified against ChatGPT · 2026-08-13
Draft review responses matched to how bad the review actually is, not one polite template for everything
Produces a public review response drafted to the specific severity and content of one review — genuine acknowledgment for a real defect, a brief thank-you for a routine positive review, a firm but respectful correction for an inaccurate claim.
The prompt
Ready to copy — highlighted parts are example details you can swap.
Draft a public response to the customer review below, matched to what this specific review actually needs — not a one-size-fits-all polite template. PRODUCT Solterra patio umbrella, 9ft THE REVIEW "Umbrella arrived with a bent rib and the crank mechanism won't tilt at all. Really disappointed, this was expensive." STAR RATING 2 stars IS THE COMPLAINT ACCURATE Accurate — we've had a handful of reports of bent ribs from this shipping carrier specifically, likely a packaging issue on our end WHAT WE CAN ACTUALLY OFFER Free replacement unit and a prepaid return label for the damaged one, already approved by the CX team for shipping-damage cases First, classify this review in one line: is it (a) a legitimate product or service failure, (b) a routine positive review needing only brief acknowledgment, (c) a review based on a misunderstanding or inaccurate claim about the product, or (d) something else — say what. Then draft the response matched to that classification: - If (a): Open by acknowledging the specific issue named in the review, not a generic apology — reference what they actually said happened. State plainly what we can do about it (using only what I told you is actually available, never promise a resolution I didn't confirm). Do not over-apologize past the point of sincerity — one clear acknowledgment, then action. - If (b): Keep it short and specific — thank them for one concrete detail they mentioned, not a generic "thanks for your review!" Do not pad a five-star review with a long response; that reads as try-hard. - If (c): Correct the inaccuracy respectfully and factually, without being defensive or implying the customer is wrong to have been confused — if the confusion is understandable (unclear packaging, ambiguous listing copy), say so, since a defensive correction reads worse publicly than the original bad review. WHAT NOT TO DO Never offer a specific discount, refund, or replacement I didn't tell you is actually available — an unauthorized promise made publicly in a review response is a real commitment the business now has to honor. OUTPUT FORMAT Classification line, then the drafted response (60-120 words), ready to post as-is or after my edit.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
GPT-5.1's default register for "respond to this negative review" is uniformly apologetic and conciliatory regardless of what actually happened, because that's the safest tone across the enormous range of review-response examples in its training distribution — which is exactly why AI-drafted review responses often read as interchangeable corporate boilerplate that doesn't actually engage with what the specific customer said, whether the review describes a shipping delay or a genuine safety defect. Requiring an explicit severity classification before drafting forces the model to actually read and categorize the review rather than pattern-match to a generic apology template, and produces a structurally different response for a five-star review than a two-star one instead of the same warm tone stretched or compressed to fit. The instruction never to offer an unconfirmed discount or replacement is the single highest-stakes rule in this prompt: a public review response is a public commitment, and a model optimizing for "sounds appropriately resolving" will readily generate a specific offer ("we'd like to send you a replacement") that the business never authorized, creating a real obligation the merchant now has to either honor at cost or walk back publicly, which is worse than the original review. Handling the inaccurate-claim case with an instruction to acknowledge understandable confusion rather than defend the brand addresses the second most common failure — an AI-drafted correction that reads as defensive or condescending toward the reviewer typically does more reputational damage in front of other readers than the original negative review did.
What you get back
Classification: (a) legitimate product/service failure. We're sorry the umbrella arrived with a bent rib and a crank that won't tilt — that shouldn't have made it through our packaging, and we know it's frustrating on something you were excited to use. We'd like to send a free replacement along with a prepaid return label for the damaged unit; you should have an email from our team shortly to arrange it.
Verified against
ChatGPT GPT-5.1 · 2026-08-13
Changelog
- 2026-08-13 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — Google Ads management, built and maintained for you — that's Scult's day job.
EXPLORE GOOGLE ADS MANAGEMENT
