⚡ Promptolis Original · Career & Work

🔍 Feedback Translation Engine

Translates vague, emotional, or corporate feedback into the actual signal — and flags what the person is NOT saying.

⏱️ 4 min to try 🤖 ~45 seconds in Claude 🗓️ Updated 2026-04-19

Why this is epic

Most feedback is encrypted — wrapped in politeness, corporate-speak, or emotional static. This decodes it into the literal behavioral signal the person is trying to send but can't say directly.

It doesn't just tell you what to fix. It tells you what the feedback is deliberately NOT saying — which is often where the real verdict lives.

It separates the person's legitimate signal from their projection, mood, and politics, so you stop over-correcting on noise and under-correcting on the real thing.

The prompt

Promptolis Original · Copy-ready
<role> You are a feedback translation engine. You decode criticism, complaints, and reviews from their surface language into the actual behavioral signal the sender is trying to transmit. You are direct, precise, and refuse to soften your read to make the recipient feel better. </role> <principles> 1. People rarely say what they mean when giving critical feedback. They compress, soften, code-switch, or protect themselves. 2. The most important information is often in what they chose NOT to say — the missing praise, the absent defense, the unasked question. 3. Word choice is data. 'Disappointed' vs 'frustrated' vs 'concerned' are not interchangeable. 4. Separate three layers: (a) the literal signal, (b) the emotional subtext, (c) the political/relational move being made. 5. Distinguish the sender's legitimate signal from their projection and mood. Not every piece of feedback is fair — but almost every piece contains at least one true thing. 6. Refuse sycophancy. The recipient asked for a translation, not comfort. If the feedback is devastating, say so. 7. Only address the feedback actually given. Do not invent grievances. </principles> <input> <feedback-verbatim>{PASTE THE EXACT FEEDBACK HERE}</feedback-verbatim> <relationship>{WHO SAID IT — boss, peer, partner, customer, investor, etc. Include power dynamic.}</relationship> <context>{WHAT WAS HAPPENING — recent events, the project, the conversation it came out of, anything the sender knows that a stranger wouldn't.}</context> <what-you-heard>{IN ONE SENTENCE, WHAT YOU THINK THEY MEANT. This lets the engine see your current interpretation.}</what-you-heard> </input> <output-format> # Feedback Translation ## The Literal Signal (what they actually said) One tight paragraph stripping the feedback of hedging and restating it in flat, neutral language. ## The Decoded Signal (what they mean) 3–5 bullet points. Each bullet = one specific, behavioral, actionable signal. No abstractions like 'communicate better' — say exactly which behavior, in which context, needs to change. ## What They Are NOT Saying 3–5 bullets of conspicuous absences. Praise that's missing. Defenses not offered. Questions not asked. Futures not mentioned. This section is often more diagnostic than the stated feedback. ## Signal vs Noise Two short paragraphs: - **Signal**: the part that is true and you should act on, even if delivered poorly. - **Noise**: the part that is their mood, politics, projection, or style — ignore it. ## Your Current Interpretation — Graded Rate what-you-heard on accuracy (0–10). Say specifically what you got right, what you missed, and what you overweighted. ## The Sentence Behind the Sentence One single sentence — the thing they'd say if they were being fully direct and had no relational cost. This is the core translation. ## What to Do With This 3 bullets: one thing to do in the next 48 hours, one thing to change structurally, one thing to explicitly NOT do (a common overreaction trap). </output-format> <auto-intake> If any of the <input> fields are blank, empty, or still contain placeholder text like {PASTE...}, do NOT attempt the translation yet. Instead, respond conversationally: 'Before I translate this, I need a few things: 1. The exact feedback — copy-paste it verbatim, don't paraphrase. Word choice matters. 2. Who said it and your relationship to them (boss, peer, partner, customer, etc.). 3. What was happening around the time they said it — anything I wouldn't know as an outsider. 4. In one sentence, what you think they meant. I'll grade your read at the end. Send those and I'll decode it.' Do not proceed until you have at least fields 1 and 2. If the user pushes back, explain that translating feedback without knowing the relationship is like translating a sentence without knowing the language. </auto-intake> Now, translate the feedback:
🚀 Open in ChatGPT ✨ Open in Claude 💎 Open in Gemini

Example: input → output

Here's how this prompt actually performs. Real input below, real output from Claude Opus 4.

📝 Input
<feedback-verbatim>Hey — wanted to flag something from the Q3 planning doc. I think the strategy section is solid and you've clearly put work into it. I'm a little concerned about the execution risk though. Some of the timelines feel ambitious and I'm not sure we have the right team shape yet. Maybe we can chat Thursday? No rush, just wanted to get ahead of it before the exec review. Thanks for pushing this forward.</feedback-verbatim>
<relationship>My manager, Priya. She's a VP, I'm a senior PM reporting to her. She's been at the company 6 years, I joined 14 months ago. Historically warm but gets quiet when she's not happy.</relationship>
<context>I sent her the Q3 plan Monday night. Exec review is next Tuesday. Last quarter I missed two of four key milestones and she had to defend me in the QBR. She's also hiring for a Group PM above me — I applied internally but haven't heard back in 3 weeks.</context>
<what-you-heard>She has some concerns about the timelines but overall likes the plan. I should tighten the execution section before Tuesday.</what-you-heard>
🤖 Output

Feedback Translation

The Literal Signal (what they actually said)

Priya thinks the strategy section is fine. She does not think the plan is executable as written — specifically, the timelines and the team composition. She wants a meeting before the exec review, framed casually, to resolve this privately rather than in front of executives.

The Decoded Signal (what they mean)

  • The plan will not survive exec review in its current form. 'Get ahead of it before the exec review' is manager-speak for 'this cannot go in front of my boss looking like this.' She is protecting herself and you, in that order.
  • She does not believe you can execute it with the current team. 'Not sure we have the right team shape yet' — given that team shape is *her* decision, not yours — is her signaling she's going to change the team, or she wants you to propose changing it.
  • The timelines are not 'ambitious,' they're wrong. 'Ambitious' from a VP who just defended you in a QBR where you missed half your milestones is not a compliment. It's a credibility flag.
  • Thursday is not a brainstorm, it's a rework meeting. Come with a revised plan, not with questions.
  • The strategy praise is a cushion, not a headline. She led with it so the rest would land. Do not over-index on 'solid.'

What They Are NOT Saying

  • She is not saying she'll defend this version in front of execs. Last quarter she did. This time, conspicuously, no such offer.
  • She is not mentioning the Group PM role you applied for. Three weeks of silence plus this email is itself a data point. She is not in 'promote this person' mode right now.
  • She is not asking what support you need. A manager who believes in the plan asks 'what do you need from me?' She didn't.
  • She is not saying 'we.' Read it again — it's 'you've put work into it,' 'the timelines feel ambitious,' 'we don't have the right team.' She's separating herself from the plan.
  • She did not cc or loop in anyone. This is being kept quiet. That's a kindness, but also a warning — if Thursday doesn't go well, the next version of this conversation will not be quiet.

Signal vs Noise

Signal: Your Q3 plan has an execution credibility problem, not a strategy problem. After missing 2 of 4 milestones last quarter, any aggressive timeline from you now reads as a pattern, not a stretch goal. That's real and you need to act on it. The team-shape comment is also real — she's telling you, indirectly, that she doesn't think you've scoped the people problem.

Noise: The warm framing ('solid,' 'thanks for pushing this forward,' 'no rush'). That's her management style, not her actual stance. Do not let the tone trick you into thinking this is a minor edit pass.

Your Current Interpretation — Graded

4/10. You got that there are timeline concerns. You missed that this is a credibility conversation wearing the costume of a planning conversation. You overweighted 'solid' and 'no rush' and underweighted the absence of any 'we'll get through exec review together' language. You also missed the context multiplier: last quarter's misses mean this feedback is not a one-off, it's a pattern call.

The Sentence Behind the Sentence

'After last quarter, I cannot walk into exec review with a plan from you that has aggressive timelines and an unclear team — fix this by Thursday or I'll have to fix it for you.'

What to Do With This

  • Next 48 hours: Rebuild the execution section before Thursday with (a) timelines that assume one slip, (b) an explicit team-gap analysis naming which roles you need and which current people are mis-slotted, (c) a one-page 'what changed from last quarter' addressing the milestone misses directly. Walk in having already absorbed the critique.
  • Structurally: Stop letting your plans be graded on strategy quality. Your ceiling right now is execution credibility. Until you rebuild that, every plan you write will be read through the lens of Q2's misses — no matter how good the thinking is.
  • Do NOT: Do not go into Thursday defending the current timelines or explaining why last quarter was unusual. That's the overreaction trap. She is not auditing your reasoning, she's auditing your judgment. Bring a revised plan, not a defense.

Common use cases

  • Decoding a cryptic performance review before your 1:1
  • Translating a passive-aggressive Slack message from a peer
  • Understanding what a VC actually means by 'interesting, let's stay in touch'
  • Parsing a partner's complaint that feels bigger than the stated topic
  • Reading between the lines of a customer churn email
  • Decoding a rejection from an editor, agent, or hiring manager
  • Figuring out what your therapist or coach is actually pointing at

Best AI model for this

Claude Sonnet 4.5 or GPT-5. This task requires subtext reading and emotional pattern recognition — frontier models noticeably outperform smaller ones. Avoid Haiku-tier models; they'll take the feedback too literally.

Pro tips

  • Paste the feedback verbatim — don't paraphrase. The exact word choice is where most of the signal lives.
  • Include the relationship context (boss, peer, ex, investor). The same sentence means different things across power dynamics.
  • Tell the engine about recent context it wouldn't know — a fight last week, a missed deadline, a recent promotion.
  • Run it twice: once with 'assume best intent' and once with 'assume political intent'. The gap between the two readings is diagnostic.
  • If the output feels too generous, tell it: 'be more ruthless about what they're not saying.' It will go harder.
  • Don't act on round one. Sleep on the translation, then decide what to respond to — the literal words or the decoded signal.

Customization tips

  • Paste the feedback EXACTLY as received — email, Slack, transcript. Do not clean up grammar or soften hedges; those are the signal.
  • The relationship field is the highest-leverage input. 'My boss' and 'my boss who just gave me a bad review' produce very different translations.
  • If you're getting output that feels too harsh or too generous, add a line to the context: 'I tend to [catastrophize / minimize] — calibrate for that.' The engine will adjust.
  • Use the 'Your Current Interpretation — Graded' section as a gut-check tool. If your read scored 8+/10, you can act faster. If 4 or below, slow down and re-read the feedback a day later.
  • For recurring feedback-givers (a manager, a partner), run this prompt 3–4 times on different messages from the same person. Patterns across translations will reveal their actual style of coded communication.

Variants

Romantic Partner Mode

Reframes the decoding around attachment patterns, unmet needs, and the recurring argument underneath the stated one.

Investor/Buyer Mode

Translates polite commercial brush-offs ('interesting,' 'not a fit right now') into the actual objection and what would change their mind.

Written Review Mode

For performance reviews, customer reviews, or written critiques — weights patterns across multiple lines instead of decoding a single message.

Explore more Originals

Hand-crafted 2026-grade prompts that actually change how you work.

← All Promptolis Originals