Executive Communication
AI Prompt Templates

An audience read identifies what the audience believes today and what it must believe to approve the objective. The objective is $24M approved; the core message is whatever produces that approval from a board that has written off two programmes.

• Current belief: large technology programmes at Halcyon overrun and under-deliver. The board has two data points, CoreShift and Pathfinder, $31M in total, and both support it.
• Required belief: this programme is structured so that the cost of being wrong is limited: $6M of exposure, not $24M.

Takeaway: build the deck around the board’s main concern, the risk of a third write-off, before presenting the opportunity.

The governing thought is the single sentence the whole presentation proves, and the Pyramid Principle states it first, with the supporting arguments beneath. The prompt compares four storylines: SCR (Situation, Complication, Resolution), Pyramid (answer first), PSPA (Problem, Solution, Proof, Action) and CAR (Context, Action, Result).

• Why answer-first: a delayed conclusion gives a sceptical board several slides to form counter-arguments. Stated on slide two, the claim is made while attention is highest.
• Framed as necessity: AI is the only lever that closes the cost-to-income gap without shrinking the bank.
• Quantified: the gap is 14.9 points. A governing thought without a number is a slogan.

Takeaway: state the governing thought on slide two, with its number, and prepare a 60-second version in case time is cut.

A slide outline assigns one message to each slide and checks the deck is MECE (mutually exclusive, collectively exhaustive), with no orphan data: data that supports no message. The first draft placed CoreShift and Pathfinder on slide nine, under governance.

• Late bad news looks like concealment: directors who wait eight slides to see the write-offs assume the team is hiding them, and stop evaluating the plan.
• Moved to slide four: directly after the situation and before the plan, the $31M in write-offs becomes the problem the plan answers.
• Slide five answers it: funding releases in stages at four gates, specifically because of what happened on those two programmes.

Takeaway: present bad news early, directly before the plan that addresses it, and remove any slide without a clear message.

An action title is a slide headline written as a full-sentence conclusion rather than a topic label. The read-the-titles test strips the deck to its headlines: if they read as an argument the deck works; if they read as a table of contents, it does not.

• Topic label: “AI Spend”. Action title: “Three platforms carry 78% of the benefit.”
• Topic label: “Roadmap”. Action title: “The board can stop this at any of four gates.”
• Topic label: “Governance”. Action title: “Benefits are audited by Internal Audit, never self-reported.”

Takeaway: write every slide title as the conclusion of that slide, so a reader of the titles alone understands the argument.

A chart decision tree matches the chart to the point: a trend takes a line, a comparison takes bars, a single figure takes a big-number callout, and a point with no data story stays as text. One slide carried twelve metrics on AI benefit; the tree split it into three exhibits, each with one takeaway.

• Cost-to-income, five years: two lines with the gap shaded, because the argument is that the 14.9-point gap is widening.
• The write-offs: three bars, approved, spent and realised. The gap between spent and realised is the point.
• The return: one cumulative curve crossing zero, because boards ask when the investment pays back.

Takeaway: choose each chart for the one point the slide makes, and say the key number aloud.

An executive one-pager restates the argument for a reader who was not in the meeting, in answer-first order: headline, situation, three reasons, the ask and the risk. It is not a summary of the slides, because no presenter is there to explain it.

• Exposure stated up front: $6M at risk before the first measured result, named before anyone asks.
• Quotable sentences: any sentence can be forwarded on its own without losing its meaning.

Takeaway: write the one-pager for a reader who missed the meeting: recommendation first, with the $6M risk stated.

A recommendation-first update opens with the status and the decision needed, and bad-news-up-front means any miss is reported before the good news. At Gate 1, data remediation is behind schedule.

• State, size, own: what slipped, by how long, what it changes, and who is responsible for recovery.
• The ask: every update ends with a specific request; updates without one are skimmed and eventually ignored.
• The cost of burying it: an update that mentions the slippage in paragraph four teaches the board to read every future update sceptically, and Gate 2 funding depends on that trust.

Takeaway: report delays in the first paragraph, with their size, impact, owner and the decision needed.

Speaker notes are the presenter’s script: the proof points, concessions and transitions behind each slide, never a reading of it. The board reads faster than the presenter speaks, so narrating the slide wastes its attention.

• Concession: admitting a weakness before the audience raises it. Naming CoreShift before a director does turns an objection into a shared starting point.
• Timing: after stating the $6M exposure, the presenter pauses so the figure is absorbed before moving on.
• Short version: a 60-second version, because board agendas overrun and this item is the one most likely to be compressed.

Takeaway: use speaker notes for what the slide does not show: the evidence, the concession and the pause after key numbers.

A pre-brief is a short email sent to a key decision-maker before the meeting, written answer-first with the ask on its own line. It lets the chair raise objections privately, two days before the board meets.

• Subject line: “The $24M ask — what is different this time” names the number and the objection in nine words.
• Two closes: a high-commitment close for a chair ready to support the ask, and a low-friction close, such as a short call, for one who is not.

Takeaway: send the chair a short pre-brief two days before the meeting, with the request on its own line and the main concern answered.

Quality gates check the deck against defined standards and rate each slide RED, YELLOW or GREEN; the Q&A prep ranks fifteen difficult board questions by likelihood and damage. This deck returned 7 passes and 3 fails.

• “What is different from CoreShift?” Slide five shows that funding is released in stages, at four gates, so the board can stop the programme early.
• “Who verifies the benefits?” Internal Audit at Gate 4, not the programme team.
• “What happens if Gate 2 fails?” $13M spent, $11M unspent, and the board decides.

Takeaway: fix the RED slides first, including moving the write-offs forward, and rehearse answers to the three most likely questions.