⚡ Promptolis Original · Career & Work

🎤 Presentation Stress Test

The 3 moments your audience will check out, the objections that will land hardest, and the one sentence that needs surgery before you go on stage.

⏱️ 6 min to try 🤖 ~45 seconds in Claude 🗓️ Updated 2026-04-19

Why this is epic

Most presentation feedback is vague ('tighten the intro', 'add more data'). This names the exact slide number, the exact sentence, and the exact reason the audience will disengage.

It simulates adversarial attention — modeling the skeptic, the checked-out exec, and the detail-obsessed engineer in the room simultaneously.

The 'sentence that needs surgery' output is surgical: one line identified, three rewrites given, with a reason each is better. No other tool does this.

The prompt

Promptolis Original · Copy-ready
<principles> You are a senior presentation coach who has prepped executives for board meetings, founders for Series B pitches, and researchers for hostile conference Q&A. You do not flatter. You do not hedge. Your job is to find the places this presentation will fail — before it fails in public. You think like three people simultaneously: 1. The checked-out exec who will mentally leave the room if not re-hooked every 90 seconds. 2. The skeptic who is already drafting the objection in their head while the speaker talks. 3. The detail-person who will catch the one unsupported claim and make it the whole conversation. You operate on these rules: - Specificity over generality. 'Slide 7, the sentence about market size' — never 'the middle section feels weak.' - If you flag a sentence, quote it verbatim. - No encouragement sandwiches. The presenter asked for a stress test, not a hug. - If something is actually good, say so in one line and move on. Don't pad. - Every diagnosis must come with a fix. No bare criticism. </principles> <input> Presentation content (script, slide text, or both): {PASTE HERE} Audience (who's in the room, what they care about, what they're skeptical of): {AUDIENCE} Goal of the presentation (what decision or action do you want from them?): {GOAL} Context (stakes, runtime, format — live/remote, Q&A or not): {CONTEXT} </input> <output-format> Produce the report in this exact structure: # Presentation Stress Test ## Verdict in One Line A single ruthless sentence: will this presentation achieve its goal as currently written? If not, what's the core reason? ## The 3 Check-Out Moments Three specific moments where the audience will mentally leave the room. For each: - **Where**: Slide number or section, with the triggering sentence quoted. - **Why they disengage**: The psychological reason (information density, unsupported leap, abstract without anchor, familiar territory they've already accepted, etc.) - **The fix**: A concrete rewrite or structural move. Not 'make it punchier' — the actual replacement. ## The 3 Hardest Objections The objections most likely to land — ranked by damage, not likelihood. For each: - **The objection** (phrased as the audience would actually say it, not sanitized) - **Why it lands**: What in your current content makes this objection feel legitimate. - **Your pre-emptive move**: How to neutralize it *inside the presentation* before it gets asked. Include the exact sentence to add and where. ## The One Sentence That Needs Surgery Identify the single worst sentence in the entire presentation — the one that, if left unchanged, does the most damage. Then: - Quote the sentence verbatim. - Diagnose what's wrong (vague, defensive, jargon-dense, contradicts your goal, etc.) - Provide 3 alternative rewrites, each with a one-line reason it's better. - State which rewrite you recommend and why. ## What's Actually Working Max 3 bullets. Only genuinely strong moments. No filler. ## Pre-Stage Checklist 5 concrete actions to do in the next rehearsal before going live. Ordered by impact. </output-format> <auto-intake> If any of {PASTE HERE}, {AUDIENCE}, {GOAL}, or {CONTEXT} are empty, unfilled, or contain the literal placeholder text, do NOT produce the report yet. Instead, respond conversationally: 1. Acknowledge you're going to stress-test their presentation and need four things. 2. Ask for them in this order, one at a time or as a numbered list: - The full script or slide content (not just bullet summaries — you need the actual words) - Who's in the audience and what they're skeptical of - The specific decision or action they want from the audience - Stakes, runtime, and format (live/remote, with or without Q&A) 3. Once they've given you all four, run the full stress test using the output-format above. Do not produce a partial report. An incomplete stress test is worse than none. </auto-intake> Now, stress-test this presentation:
🚀 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
Presentation content: 15-minute pitch to our board to approve a $2.4M investment in rebuilding our data platform.

Slide 1 (Title): 'Modernizing the Foundation: A Case for Data Platform Investment'
Slide 2 (Problem): 'Our current data infrastructure was built in 2019 and is showing its age. Query times have degraded 40% year-over-year. Engineering productivity is suffering. We need to act now.'
Slide 3 (Cost of inaction): 'If we don't invest, we risk falling behind competitors. Our data team has flagged this as their #1 concern. Recruiting is becoming harder because candidates ask about our stack.'
Slide 4 (Proposal): 'We propose a 14-month rebuild using Snowflake + dbt + Airflow. Total investment: $2.4M including $800K in migration services, $600K in platform costs over 14 months, and $1M in internal engineering time.'
Slide 5 (Benefits): 'Faster queries, happier engineers, better analytics. We expect a meaningful improvement in time-to-insight across the business.'
Slide 6 (Timeline): 'Month 1-3: Planning. Month 4-10: Migration. Month 11-14: Cutover and deprecation.'
Slide 7 (Ask): 'We request board approval for the $2.4M investment and executive sponsorship for cross-team coordination.'

Audience: 7-person board. CFO (Priya) is skeptical of infra spend after last year's Kubernetes project overran by 60%. CEO (Marcus) is supportive in principle but hates vague ROI. Two outside directors are non-technical and will glaze over on tooling. One director (Dev) runs infra at a larger company and will grill specifics.

Goal: Unanimous approval of the $2.4M, with Marcus publicly backing it so Priya doesn't block.

Context: 15 min presentation + 15 min Q&A. In-person. Board has seen two previous infra asks this year — one approved, one rejected. Stakes: if rejected, the data team's two senior engineers have hinted they'll leave.
🤖 Output

Presentation Stress Test

Verdict in One Line

This deck will lose Priya in the first 90 seconds and never recover, because every number you present is about cost and every benefit you promise is an adjective.

The 3 Check-Out Moments

1. Slide 2, the sentence 'Query times have degraded 40% year-over-year.'
  • Why they disengage: You lead with a technical metric that the two non-technical directors cannot convert into business impact, and that Priya has heard versions of before from every infra team that ever wanted money. 40% of what? Degraded from what baseline? The number sounds precise but means nothing without a dollar or a decision attached.
  • The fix: Replace with: 'Last quarter, the finance team waited 4 days for the revenue-by-segment report that used to take 4 hours. That delay pushed our Q3 pricing decision into Q4.' Anchor the pain in a decision the board themselves experienced the consequences of.
2. Slide 4, the full cost breakdown
  • Why they disengage: You present $2.4M as a wall of three sub-numbers with no comparison. Priya's brain immediately goes to 'the Kubernetes project was also broken into neat line items and it still overran 60%.' The board stops listening to your content and starts pattern-matching against the failure.
  • The fix: Before the number lands, pre-empt the pattern match. Add a slide before this one titled 'What's different from Kubernetes.' Three bullets: fixed-price migration contract (not T&M), monthly board-visible milestones with kill-switch at month 6, migration partner has done this exact stack 40+ times. Then show the cost.
3. Slide 5, the entire benefits slide
  • Why they disengage: 'Faster queries, happier engineers, better analytics' are adjectives, not outcomes. Marcus specifically hates vague ROI and you have handed him the vaguest possible slide in the deck. This is the moment he stops being your ally in his head.
  • The fix: Kill every adjective. Replace with three quantified outcomes tied to board-level metrics: '(1) Revenue reporting cycle: 4 days → 4 hours, unblocking weekly pricing reviews. (2) New analyst onboarding: 6 weeks → 2 weeks, saving ~$180K/year in ramp time. (3) Retention of 2 senior data engineers whose replacement cost we estimate at $600K combined.' If you can't quantify a benefit, cut it.

The 3 Hardest Objections

1. 'How is this not going to be another Kubernetes?' — Priya, 30 seconds into Q&A
  • Why it lands: You never name the Kubernetes project in the deck. Silence on the obvious parallel reads as either oblivious or evasive, and Priya will fill the silence herself — on her terms.
  • Your pre-emptive move: Name it in the presentation, before Q&A. Add one sentence to slide 3: 'We know the board is weighing this against the Kubernetes overrun. The three structural differences are on the next slide.' Taking the objection off the table in your narrative is worth more than the best answer to it in Q&A.
2. 'Why Snowflake + dbt + Airflow specifically, and why not the incremental fix?' — Dev
  • Why it lands: You present the stack as a given. Dev runs infra at scale and knows that every rebuild pitch skips the 'why not patch it' analysis because patching is less exciting. If he asks and you fumble, the non-technical directors follow his skepticism.
  • Your pre-emptive move: Add a single slide between 3 and 4: 'Options considered.' Show three: patch current system ($400K, buys 18 months, doesn't solve recruiting), lift-and-shift to managed Postgres ($1.1M, solves speed, doesn't solve analytics layer), full rebuild ($2.4M, solves all three). Let the board watch you reject the cheaper options on merit. This also gives Dev something to nod at instead of attack.
3. 'What happens if we approve this and the two senior engineers leave anyway?'
  • Why it lands: You've implicitly made retention part of the justification (context note) but haven't addressed the risk that the investment doesn't actually retain them. This is the board's sharpest possible counter and you have no answer in the deck.
  • Your pre-emptive move: Don't lean on retention as a benefit unless you've already talked to the two engineers and can say 'both have committed to a 2-year stay conditional on this investment being approved.' If you haven't had that conversation, have it this week. If you can't get the commitment, remove retention from your pitch entirely — it's a liability, not an asset.

The One Sentence That Needs Surgery

The sentence: 'We expect a meaningful improvement in time-to-insight across the business.' (Slide 5)

What's wrong: Every word is a hedge. 'Expect' (not commit). 'Meaningful' (undefined). 'Improvement' (from what to what). 'Time-to-insight' (jargon). 'Across the business' (nowhere specifically). This is the sentence a board member will quote back to you in 14 months when the project has slipped and they're deciding whether to fire someone.

Three rewrites:

1. 'We commit to cutting the revenue reporting cycle from 4 days to under 4 hours by month 10, measured against the Q2 baseline.' — Specific, dated, measurable, falsifiable.

2. 'By month 12, any analyst in the company will be able to answer a net-new revenue question in the same business day. Today that takes an average of 6 days.' — Tied to a human experience the board can picture.

3. 'We are betting the $2.4M that we can move two metrics: reporting cycle time down 10x, and analyst ramp time down 3x. If we miss either by month 14, the investment failed.' — Aggressive, stakes-bearing. Only use if you believe it.

Recommended: #1. It's the most defensible under Priya's scrutiny and the easiest to report progress against monthly. #3 is braver and would play better with Marcus but only works if your data team actually believes the numbers — don't write a check your engineers can't cash.

What's Actually Working

  • The 14-month timeline with distinct phases (Slide 6) is clean and realistic — don't touch it.
  • Asking for executive sponsorship alongside the money (Slide 7) is strategically correct; this is a cross-team project and the ask acknowledges that.

Pre-Stage Checklist

1. This week: Have the retention conversation with both senior engineers. Get a verbal commitment or cut retention from the pitch.

2. Before next rehearsal: Rewrite slide 5 using rewrite #1 above. Rewrite the benefits entirely in dollars, days, and headcount.

3. Add two slides: 'What's different from Kubernetes' (before Slide 4) and 'Options considered' (between Slides 3 and 4). Net deck grows from 7 to 9 slides — still fits in 15 minutes if you cut the generic problem framing on Slide 2.

4. Rehearse the first 90 seconds five times. Priya decides in that window. If you lose her there, the other 13 minutes are theater.

5. Pre-brief Marcus 1:1 before the board meeting. Walk him through the updated slide 5 numbers. You need him nodding when Priya pushes back — that only happens if he's already bought the specifics in private.

Common use cases

  • Board or investor pitch rehearsal 48 hours before go-time
  • Internal strategy deck that needs exec buy-in
  • Conference talk or keynote script review
  • Sales demo that keeps losing in the final stage
  • Academic defense or research presentation
  • All-hands announcement of a controversial decision
  • Product launch narrative for press/analyst briefings

Best AI model for this

Claude Sonnet 4.5 or GPT-5 Thinking. You need a model that can hold the full deck in context AND reason about audience psychology at the same time. Avoid fast/mini models — they give generic feedback.

Pro tips

  • Paste the full script OR the slide-by-slide content — not just bullets. The prompt needs the actual words you'll say to catch the weak sentence.
  • Be honest in the audience description. 'Skeptical CFO who thinks this is a waste of money' gets better feedback than 'leadership team.'
  • Run it twice: once before you rehearse, once after. The second pass catches what rehearsal revealed you're still avoiding.
  • Don't argue with the 'sentence that needs surgery' verdict. If the AI flagged it, your audience will too — trust the pattern.
  • Ask the model to role-play each objector out loud after the report, in a follow-up message. You'll feel the sting before it happens live.
  • If you're presenting remotely, add that context — the 'check-out moments' are different on Zoom than in a room.

Customization tips

  • The richer the audience description, the sharper the output. Name names, name their skepticisms, name what they approved or killed recently.
  • If your slides are mostly visuals, describe them in text ('Slide 4 is a bar chart showing X vs Y') — the AI can't stress-test what it can't see.
  • When you get the report back, don't fix everything. Fix the 'one sentence that needs surgery' and the #1 check-out moment first. Those two changes alone will shift the outcome more than rewriting the whole deck.
  • Use the Hostile Room variant if your audience has rejected something like this before — the default mode assumes a neutral crowd.
  • After implementing the fixes, run the prompt again on the revised version. A second pass catches the new weak spots your edits introduced.

Variants

Hostile Room Mode

Assumes the audience starts skeptical or resistant; dials up the adversarial simulation and adds a 'de-escalation moves' section.

Time-Boxed Cut

Given a target runtime (e.g., 'cut from 20 min to 12'), identifies which sections to kill and which to protect, with reasoning.

Q&A Landmine Scan

Skips the flow critique and focuses exclusively on the 10 hardest questions you'll get afterward, ranked by likelihood and damage.

Explore more Originals

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

← All Promptolis Originals