⏳ Read Time: Less than 4 min.

I was in Vienna earlier this week, delivering The AI Advantage to the operations team of a financial services client,

I watched how fast AI now produces a confident claim. Ask it, and back comes: “This initiative will cut processing time by forty percent.” Fluent. Specific. Beautifully formatted.

And, to a committee, completely unconvincing — because nobody in the room has any reason to trust a round number that appeared out of a prompt.

That gap — between a claim and a claim people believe — is what we're covering today.
__________________________________________________________________

✅ The Proof Sentence — the one line that turns a claim into something a committee believes
✅ The four parts every believable proof sentence needs — and what breaks when you skip one
✅ Why precise numbers persuade and round numbers invite doubt
✅ AI tip: the prompt that rewrites a vague claim into a checkable sentence
✅ Your action challenge: write one Proof Sentence before your next meeting

Most people think proof means more: another testimonial, a longer case study, a slide of logos. The opposite is true. A wall of glowing quotes reads as marketing, and committees discount it on sight. What they believe is a single, specific line they could check if they wanted to.

I call it the Proof Sentence: one sentence, specific and verifiable, tied directly to the decision you’re asking for. Not a success-story slide. Not an appendix. One line a sceptical person could test.

Weak proof: “This approach has been very successful and widely praised.”

Strong proof: “When we ran it on the reconciliation team’s own backlog for two weeks, it cleared 1,900 items with zero exceptions.”

Same claim underneath. Only the second one gets believed — because it names who, what, how much, and when.

Want the structure that puts proof where a committee looks hardest — the results and evidence slides? My Executive Slide System (£39) gives you the templates for it: 26 board-ready templates, 93 AI prompts and 16 scenario playbooks.

→ Get the Executive Slide System (£39)

💡 Tip — Before your next meeting, find the one round number in your deck ("about 40%", "significantly faster") and replace it with the exact figure and timeframe behind it. Precision is the cheapest credibility you can buy.

🎭 Power Move — After you say your proof sentence out loud, stop. Don’t rush to the next slide. A specific, checkable claim earns a beat of silence — let the room register that you could defend it.

📖 Read — Made to Stick by Chip Heath & Dan Heath (Random House). Their "Concrete" and "Credible" principles are the Proof Sentence in book form — why specific, testable claims lodge in memory while vague ones slide off.

The Four Parts of a Line People Believe

A Proof Sentence has four parts, and skipping any one is what makes proof collapse under pressure:

1. Specific actor — whose result this actually was. Miss it and the claim sounds hypothetical.

2. Specific situation — the real conditions, not a demo or a pilot in ideal circumstances.

3. Measured outcome — a number you can defend, not a rounded one you reached for. "1,900 items", not "most of them".

4. Timeframe — over what period. Miss it and the result sounds cherry-picked from a good week.

Include all four and the sentence stops sounding like a promise and starts sounding like a fact — because it is one. Leave one out, and that’s the exact gap a sharp committee member will put their finger on.

The Claim-to-Proof Prompt

Run this on any claim in your deck before a high-stakes meeting. Works with ChatGPT, Claude, or Copilot.

“Here is a claim from my presentation: [paste the claim]. Rewrite it as one specific, checkable sentence a sceptical committee would believe — name the actor, the exact number, and the timeframe. If I haven’t given you enough detail to do that, tell me precisely what I need to find first.”

What you get: either a sentence you can defend, or a short list of exactly what evidence is missing — which is the most useful thing to learn the night before, not in the room.

Don’t ask Copilot for a statistic — ask it for the evidence you already have. In the file or channel where your data lives, try:

“Here are our reconciliation logs for the last fortnight. Give me one sentence stating what actually happened — the exact count and the timeframe — that I could defend to a committee.”

What it does: turns Copilot from a generator of plausible numbers into a compressor of real ones. The proof was in your own data the whole time; you’re just making it sayable.

✅ This week's challenge: — three steps before Friday:

WRITE IT — Take the single claim your next recommendation most depends on and write its Proof Sentence: actor, situation, number, timeframe.

TEST IT — Read it back and ask "how do you know?" If your sentence is already the answer, it’s ready. If not, you’ve found the gap to close

Community question: What’s the one claim in your work you’ve never been able to prove cleanly? Reply and tell me. I read every response.

One action. One week. One step closer to being believed the moment you speak.

P.S.
Thanks for reading this week’s edition. Next Thursday we’re on the Recommendation Lock — why the most "complete" AI deck, the one with a chart on every slide, is so often the one a committee quietly waves through and forgets. It’s a thirty-second habit you do before AI builds a single slide, and it changes what the whole deck is for. Don’t miss it.

See you then!

Keep building your edge—one conversation at a time.

If you have more feedback or just want to get in touch, respond to this email, and we’ll get back to you!

Thanks for reading!

Until next week!

Mary Beth