Your slides are ready. Can you answer these three objections?

Tomorrow you will ask your team to approve a proposal. The slides look finished. Then someone asks: “How will we know this worked?” If your answer is “AI will improve productivity,” the decision still needs work. Use the example below to prepare three short answers, then copy the worksheet for your own proposal.

1. Start with the five-slide proposal

• 1. Problem: the fictional support team has to rewrite internal how-to notes.
• 2. Idea: AI will improve productivity.
• 3. Plan: try AI-assisted drafts for six weeks.
• 4. Evaluation: ask the team whether it helped.
• 5. Decision: approve the pilot.

The proposal names an activity but leaves the approval scope, baseline, owner and failure rule unclear. “The team liked it” would not tell us whether a usable note took less total work.

2. Repair the decision slide

In this fictional scenario, Eva, the support lead, proposes one repeatable process: drafting internal how-to notes from approved public product instructions. She has not measured the baseline yet. The pilot starts only after that measurement and team capacity are confirmed. Copy the comparison below into a document and adapt it; do not carry its sample numbers into a real proposal as evidence.

3. Three objections, three repaired answers

3.1 Why this process?

Weak answer: “AI is useful for many things, so we should start somewhere.”

Objection: “Why documentation rather than the other work we could improve?”

Repaired answer: “I propose internal how-to notes because they have repeatable inputs and a named reviewer. I do not yet have evidence that they are our biggest bottleneck. Eva will time five comparable manual notes first. If that measurement does not support this choice, we will reconsider the process before starting.”

3.2 How will accepted quality be measured?

Weak answer: “We will count how many drafts AI produces.”

Objection: “A fast draft is not necessarily an accepted note. What counts as good enough?”

Repaired answer: “Eva will check every pilot note for correct steps, working public links and the agreed template before anyone uses it. We will count accepted notes and total hands-on minutes per accepted note, including prompting, checking and rework. We will compare them with five manual baseline notes of similar complexity. The proposed target is a median at least 20% below baseline with no unresolved critical errors in accepted notes. This is a test threshold, not an expected saving.”

3.3 What happens if the trial fails?

Weak answer: “We will adjust the prompts and keep improving.”

Objection: “Who stops the trial, and what happens to the work?”

Repaired answer: “Eva will pause the pilot immediately if confidential data is entered or a critical error passes the acceptance check. At the end of week six, she will stop AI-assisted drafting if the agreed quality and time criteria are not both met. The team will return to the manual process; only human-accepted notes stay in use. Any new trial requires a new decision.”

4. Use AI to question the proposal

Copy the repaired proposal facts into the prompt below. Use only fictional or public material, or material explicitly permitted for the tool by your organization. Do not paste a confidential deck or customer data. If the tool invents a fact, discard that part and verify against your inputs. No particular model performance is promised.

You are a skeptical colleague rehearsing my proposal.
Use only the proposal facts supplied below.
Ask these three questions one at a time and wait for my answer:
1. Why this process?
2. How will accepted quality be measured?
3. What happens if the trial fails?
After each answer, identify any missing evidence. Do not invent measurements, sources, owners or results. Label unsupported assumptions.
Check: specific decision, evidence for assumptions, owner, measurable check.
Ask me to revise the answer before moving on.
Proposal facts: [paste the repaired proposal and mark illustrative values].

Without AI, ask a colleague to read the same three objections, or read them yourself and record your answers. The useful output is a revised answer you can defend, not a fluent transcript.

5. Check whether the answer improved

The original “AI will improve productivity” fails all four checks below. The repaired proposal passes the decision, owner and measurable-check checks. Evidence remains pending until Eva collects the baseline; naming that gap is more honest than awarding the proposal a full score.

• Specific decision: can the listener say exactly what is being approved?
• Supported assumption: is there a verifiable source or measurement, or an explicit evidence gap?
• Owner: who checks quality and decides whether to stop?
• Measurable check: what is measured, compared with what, and when?

6. Copy your rehearsal worksheet

Fill this in for one fictional or public proposal. Copy it into a document, complete the blanks, and revise the decision slide. You do not need to buy a course to use it.

Decision I need: ______
Process and reason for choosing it: ______
Evidence/source, or what is still unknown: ______
Process owner and acceptance reviewer: ______
Baseline sample and measurement date: ______
Acceptance criteria and critical-error definition: ______
Total-work measure, comparison and review date: ______
Pilot scope, duration and capacity limit: ______
Immediate pause trigger and final stop rule: ______
Fallback if we stop: ______

Why this process? My answer: ______
How is accepted quality measured? My answer: ______
What happens if it fails? My answer: ______
One missing fact I will verify before asking for approval: ______

Now close the worksheet and answer aloud without AI: “What exactly are you asking us to approve, and what would make you stop?” If you cannot answer briefly, revise the proposal before polishing another slide.

If you are new to AI, try the free foundation What is AI (and What It Isn’t). It is an introduction to AI, not a presentation coaching course; the worksheet above stands on its own.

7. Source and limits

Canva’s Presentation Paradox survey reports that 74% of respondents get nervous presenting and 53% use AI as a presentation coach. Canva and Pure Spectrum surveyed 4,000 regular presenters in eight countries. These are vendor-reported self-reports, not evidence of local demand or that this exercise reduces anxiety. Source checked October 8, 2026. The worked proposal, thresholds and dialogue above are original illustrative teaching material.