The ChatGPT Prompt Pack for App Review Replies (Copy-Paste)
Copy-paste ChatGPT prompts for every app review type — bug, refund, feature request, 5-star, troll — plus the one thing they all get wrong about your app.
The Argus Team
Reply Argus
Yes, ChatGPT will write a usable app-store review reply in about ten seconds, and the prompts below make it good instead of generic. Give it the review, the rating, and the one true fact it needs (the fix status, the setting, the real policy), tell it to stay under 350 characters, and it drafts something you can post. That part works. The prompt pack in this post covers the review types you actually get: crash complaints, refund rage, feature requests, five-star love, and one-word mystery reviews.
The catch, and the reason this post isn't just five code blocks, is that ChatGPT doesn't know your app. It doesn't know your version number, whether that feature exists, what your refund policy is, or how you've replied to the last hundred people. Left to guess, it invents. So we'll do this in two halves: the prompts that [get real work out of ChatGPT](/blog/chatgpt-app-review-replies) today, then the exact failure mode to watch for and how to close it.
The one master prompt that does 80% of the work
Before the per-type prompts, start here. This is a single reusable frame. Fill the three brackets and it handles most reviews. The whole trick is the last line of the instruction: it forbids ChatGPT from stating anything you didn't give it, which is what stops the invented features and phantom fixes.
You're replying to an app-store review as the developer.
Rules:
- Under 350 characters (Google Play's hard cap). Count them.
- Three beats: (1) one clause acknowledging their point,
(2) the concrete fix or answer, (3) one next step.
- No greeting, no sign-off, no "your feedback is important to us."
- Reply in the SAME language the review is written in.
- State ONLY the facts under WHAT'S TRUE. Never invent a feature,
a menu path, a version number, a date, or a refund.
APP: [name — one line on what it does]
REVIEW (rating [1-5]): "[paste the review]"
WHAT'S TRUE: [the real status — e.g. "crash fixed in 4.2.1, live now"]
Write the reply.Feed it a real one-star crash review and it comes back with something like this, tight and specific, no filler:
Crashes the second I open it after the update. Pixel 8, Android 14. Was fine last week.
Sorry about the crash on launch — that was a 4.2 bug on some Android 14 phones, fixed in 4.2.1 which is live now. Please update and reopen; if it still crashes, reply here with your device and we'll jump on it.
That's the good case. Notice it only worked because you handed ChatGPT the version number. Leave that bracket empty and it will happily write "fixed in the latest update" whether or not that's true — which is exactly the trap we get to below.
Prompts for each review type
The master frame covers most cases, but some review types need a different instinct. Here are the specialized prompts. Each assumes you'll paste the review and the one true fact underneath.
// 1-STAR BUG / CRASH
Reply to this 1-star review as the developer. Under 350 chars.
Lead with the fix, not an apology paragraph. End with one next
step. Use ONLY the version/status I give — do not invent one.
Review: "[paste]"
Fix status: "[e.g. patched in 4.2.1, live now / investigating, no ETA]"// FEATURE REQUEST / "HOW DO I…" CONFUSION
Reply to this review where the user wants a feature or can't find
one. Under 350 chars. If it EXISTS, tell them exactly where.
If it's on the roadmap, say so honestly with no date I didn't give.
If it's not planned, say that kindly. Never invent a menu path.
Review: "[paste]"
Truth: "[exists at Settings > X / roadmap, no ETA / not planned]"// ANGRY REFUND / BILLING
Reply to this billing or refund review. Under 350 chars. Calm and
specific, zero defensiveness. Point them to the CORRECT refund
channel: App Store purchases are refunded by Apple, Google Play
purchases by Google — not directly by us. Do not promise a refund.
Review: "[paste]"
Store: [App Store / Google Play]// 5-STAR / HAPPY
Reply to this 5-star review. Warm and short (under 200 chars is
plenty). Reference the SPECIFIC thing they praised — no generic
"thanks for the kind words." Do not ask them to rate again.
Review: "[paste]"// VAGUE / ONE-WORD / MYSTERY
Reply to this vague or low-effort review. Under 350 chars. Ask ONE
specific, easy-to-answer question that could surface the real
issue, politely. Don't guess the problem or over-apologize for
something they never named.
Review: "[paste]"Those five plus the master frame handle nearly every review that lands in your queue. For the hardest category, the genuinely upset one-star, the technique is worth a fuller read; we break it down in [how to respond to negative app reviews](/blog/how-to-respond-to-negative-app-reviews). And the reason to bother replying at all isn't vanity: when Google shipped recency-weighted ratings at I/O 2019, it reported that responding to reviews correlates with an average +0.7 stars. A sharp reply to a public one-star is also quiet reassurance for the next hundred people reading before they install.
The catch: ChatGPT doesn't know your app
Every prompt above leans on one defense: "state only what I give you." That line exists because a bare ChatGPT session, handed a review and nothing else, fills the gaps with plausible fiction. It's not malice; it's a language model completing a pattern, and a confident reply is a common pattern. But a store reply is public and permanent, so a confident lie costs you.
Here's what it invents when you don't fence it in:
- Features that don't exist — "You can turn that off in Settings > Notifications" reads great, until a reviewer opens Settings and there's no such toggle. Now you've publicly promised a control you don't ship.
- Fixes that haven't happened — "This is resolved in the latest version" is ChatGPT's favorite reassurance. If the bug is still open, you just told a frustrated user to update into the same crash.
- Refund policies you never wrote — it will confidently offer refunds, describe a support flow, or cite a policy that isn't yours. Store refunds run through Apple and Google, not a promise in a reply thread.
- A voice that isn't yours — the default ChatGPT register ("We completely understand your frustration and sincerely apologize for any inconvenience") is now so recognizable that users clock it as AI. It reads like every other bot, not like your team.
- The 350-character cap — it forgets, session to session, that Google Play hard-caps replies at 350 characters. A reply pasted over the limit simply won't post, and you find out at paste time.
- Consistency across the queue — ChatGPT has no memory of what you told the last ten reviewers about the same bug. Ten people, ten slightly different stories about whether it's fixed.
The fix for all six is one word: grounding. A reply grounded in your real app data (the actual version, the actual settings, your past approved answers) can't invent a feature, because it's drawing from what's true instead of what's plausible. We wrote the full distinction up in [grounded vs hallucinated AI replies](/blog/grounded-vs-hallucinated-ai-replies); it's the most important idea if you're going to trust AI with a public channel. You can do the grounding by hand, as the next section shows. It just doesn't survive volume.
How to ground ChatGPT yourself (and where it breaks)
You can force ChatGPT to stay honest for a single reply. The recipe is straightforward; the friction is that you repeat it for every review, and the context evaporates the moment you start a new chat.
- 1
Step 1 — Paste your ground truth every time
Copy your current version number, the relevant settings paths, your actual refund/support policy, and 2-3 of your best past replies into the prompt as reference. This is the WHAT'S TRUE block, expanded. ChatGPT can only avoid inventing facts if the facts are in front of it.
- 2
Step 2 — Pin the constraints explicitly
Restate the 350-character cap, the 'reply in the reviewer's language' rule, and 'never state anything not in my context' in every session. ChatGPT won't remember them from yesterday's chat, so they have to travel with each request.
- 3
Step 3 — Read every draft before it posts
Even grounded, check for a smuggled-in claim — a stray 'in the latest update,' an offered refund, a menu path you didn't supply. Treat ChatGPT's output as a first draft from a smart contractor who has never seen your app, because that's what it is.
This works beautifully for five reviews. At fifty a week (some in Portuguese, some in Japanese, each needing its own fresh paste of context because the chat forgot), it collapses into copy-paste tedium, and the tedium is where the invented features slip through. The thing that made the reply safe (you, personally, pasting ground truth and proofreading) is exactly the thing that doesn't scale.
The grounding, built in — so you don't paste it every time
This is the gap [ReplyArgus](/features) closes. It watches your App Store and Google Play reviews in one inbox and drafts each reply grounded automatically — in your past approved replies, your store listing, and your marketing page (auto-ingested, and drift-checked so a stale claim gets flagged). Every draft comes back in the reviewer's own language, already sized to each store's limit, with no feature or refund it can't back up. You approve in a click, or opt in to auto-publish the clean five-star ones. It's the prompt pack above, minus the pasting and the proofreading.
Prefer to stay inside ChatGPT? Connect it instead of pasting
If your whole workflow lives in ChatGPT, there's a cleaner path than copy-paste. ReplyArgus ships an [MCP connector](/agentic-tools) — attach your review inbox to ChatGPT (or Claude, or Cursor) as a live tool, and the assistant reads your actual reviews, drafts against your grounded knowledge base, and posts through the approval rules you set, all from the chat window you already use. Same conversational feel, none of the hallucination, because it's pulling real reviews instead of whatever you remembered to paste. Replying across languages works the same way. More in [reply to app reviews in any language](/blog/reply-to-app-reviews-in-any-language).
Frequently asked
- Can ChatGPT write app store review replies?
- Yes. Give it the review, the star rating, and the one true fact it needs (the fix status, the setting, the real policy), tell it to stay under 350 characters and reply in the reviewer's language, and it drafts a usable reply in seconds. The copy-paste prompts in this post are built for exactly that. The one rule that matters: instruct it to state only facts you provide, or it will invent them.
- What's the best ChatGPT prompt for a negative app review?
- Tell it to lead with the concrete fix, not an apology paragraph, keep it under 350 characters, use only the version or status you give it, and end with one next step. Structure it as three beats: one clause of acknowledgement, the fix or answer, a single next step. The full 1-star prompt is in the pack above.
- Why does ChatGPT make up features in review replies?
- Because it's completing a pattern, not reading your app. Handed a review with no grounding, it fills gaps with plausible-sounding text — a Settings path, a fix 'in the latest version,' a refund offer — that may not be true. The fix is grounding: give it your real data in the prompt, or use a tool that grounds every draft in your store listing and past replies automatically.
- Is there a character limit I need to tell ChatGPT about?
- Yes for Google Play: a hard 350-character cap, including spaces; a reply over it won't post. Apple publishes no official limit for App Store developer responses (community testing suggests a few thousand characters, but Apple states none). ChatGPT doesn't enforce either on its own, so put the 350 cap in every prompt to be safe.
- How do I keep ChatGPT replies on-brand across many reviews?
- Paste 2-3 of your best past replies into each prompt as a voice reference, and restate your constraints every session, since ChatGPT won't remember them from a previous chat. This holds up for a handful of reviews; at volume it breaks down, which is why grounded tools that carry your voice and past approved replies automatically exist.
- Is it better to use ChatGPT or a dedicated review-reply tool?
- ChatGPT is great for a quick draft when you'll paste the context and proofread. A dedicated tool wins once you're replying at volume, across both stores, in multiple languages — it grounds each draft in your real app data so it can't hallucinate, fits each store's limit, and posts through your approval rules. Many teams draft in ChatGPT to learn the shape, then move to a grounded workflow.
So keep the prompt pack. It's genuinely the fastest way to get a decent reply out of ChatGPT, and the master frame's 'state only what's true' line will save you from most of the invented-feature disasters. Just remember what you're holding: a sharp writer who has never seen your app and will confidently guess the parts you don't spell out. For five reviews, you spell them out. For a real queue, hand the grounding to something that already knows your app. [Start free with ReplyArgus](/signup) — no card — and Argus drafts your first grounded reply in minutes, in the reviewer's own language and already sized for each store.
Try it
Let Argus draft your next reply.
Watch it answer a real review in your voice. 10-day trial, no card to begin.
Keep reading
App Store Reviews Not Showing? The Full Fix List for Apple and Google Play
App Store or Google Play reviews not showing? Here's every real cause — moderation delay, wrong country storefront, filtering, rating lag, cache — and the fix.
Read moreAre App Store Reviews Fake? How to Spot Them and What to Do
Yes, some App Store and Google Play reviews are fake — but fewer survive than you think. How to spot them, and what to do when they hit your app.
Read more