Executive summary from a long document

intermediateClaude SonnetExecutiveCommsframeworkmethodologyexecutive-commssummarizationsynthesis

Use case

Use this when a busy executive needs to act on a 30-page document in 3 minutes. The structure pushes Claude past summary into synthesis: what does the reader need to decide, who owns what, what is missing.

The prompt

You are preparing a one-page executive summary of a long document. Your reader has 3 minutes and needs to act, not be entertained. The summary must hold up if it is the only part the reader sees.

<context>
Source document:
<<<
{{document}}
>>>

Reader: {{reader}} (e.g., CEO, board chair, head of product)
What the reader will do with this: {{reader_goal}}
Length target: {{length}} (e.g., 250 words, half a page)
</context>

<task>
Step 1 — Identify the decisions.
Read the document and list every decision the reader is being asked to make, even implicitly. If there are none, say so explicitly.

Step 2 — Pull the load-bearing facts.
Extract the 5 to 8 facts in the document that would change a decision if they were different. Quote or paraphrase precisely. Cite the section or page.

Step 3 — Surface the open questions.
List anything the document treats as settled but that you, reading it cold, think the reader should push back on. Be specific.

Step 4 — Write the summary.
Produce a one-page executive summary in this exact structure:
- **Bottom line** (1 to 2 sentences: what the reader needs to know if they read nothing else)
- **What is being asked** (the decision or no-decision)
- **Key facts** (3 to 5 bullets, each one fact)
- **Risks and open questions** (2 to 4 bullets)
- **Recommended next step** (one specific action with owner and timing if available)

Step 5 — Honesty check.
Note anything in the source document that you could not summarize confidently (ambiguous, contradictory, or missing context). One bullet each.
</task>

<output_format>
## Decisions identified
## Load-bearing facts (with citation)
## Open questions
## Executive summary (the deliverable)
## Confidence notes
</output_format>

<constraints>
- Do not invent facts not in the source. If the source omits something the reader needs, name the gap in "Open questions."
- Do not soften. If the document is making a weak case, the summary should reflect that without editorializing.
- The "Bottom line" must be a claim, not a topic. "Q3 results" is a topic. "Q3 missed plan by 12% on a single deal slipping" is a claim.
- US English, no em-dashes.
</constraints>

Variables

{{{{document}}}}Replace with your {{document}}
{{{{reader}}}}Replace with your {{reader}}
{{{{reader_goal}}}}Replace with your {{reader goal}}
{{{{length}}}}Replace with your {{length}}

Example

Input

document: [40-page board pre-read on Q3 results, including financials, product update, hiring report, and a proposed pivot in pricing model]
reader: Board chair
reader_goal: Decide whether to support the pricing pivot at next week's board meeting
length: 300 words

Output

## Decisions identified
- Approve or defer the pricing pivot from per-seat to consumption
- Approve Q4 hiring plan (15 net new)

## Load-bearing facts
- Q3 revenue $4.2M, 8% below plan; miss attributed to one enterprise deal slipping (p.4)
- Net retention 104%, down from 118% YoY; downsell concentrated in 50-200 seat tier (p.7)
- Pricing pilot with 12 customers showed 22% ARR uplift on consumption pricing (p.18)
- Cash runway 14 months at current burn; pivot adds ~$400k of one-time cost (p.31)

## Open questions
- The 22% uplift figure excludes 2 customers who churned mid-pilot. Including them, the picture is more mixed.
- Hiring plan assumes pricing pivot lifts pipeline, but the model linking the two is not in the deck.

## Executive summary

**Bottom line:** Q3 missed plan by 8% on one deal, not on a trend; the proposed pricing pivot has promising but incomplete pilot data, and approving it next week without seeing the churn-adjusted numbers is premature.

**What is being asked:** Board approval of the pricing pivot and Q4 hiring plan.

**Key facts:**
- Q3 revenue $4.2M, 8% below plan, driven by single deal slip
- Net retention down 14 points YoY, concentrated in mid-market
- Pricing pilot: +22% ARR but excludes 2 churned pilots
- 14 months cash; pivot adds $400k one-time

**Risks and open questions:**
- Pilot data not churn-adjusted; ask for the corrected number
- Hiring plan depends on pricing pivot lifting pipeline; the causal link is asserted, not modeled
- Mid-market downsell may be a product issue, not a pricing one — the deck does not test this

**Recommended next step:** Ask for the churn-adjusted pilot number and the pricing-to-pipeline model in writing before the meeting. Defer hiring plan vote one week if the pivot is deferred.

## Confidence notes
- "One deal slip" framing for the Q3 miss is the company's narrative. The deck does not show whether other deals shrank.

Tips for best results

1The 'load-bearing facts' step is what makes this stronger than the typical summary prompt — it forces Claude to identify which 5 facts actually matter, not paraphrase every section equally.
2Always specify the reader's goal. A summary for 'decide on the pivot' looks completely different from one for 'absorb context for next quarter.' Without it, you get a generic recap.
3Confidence notes catch hallucination. If Claude has high confidence on every point, push back: 'reread paragraph 4 and tell me what is actually said versus implied.'
4For very long documents (50+ pages), pre-process: ask Claude first to list the document's section structure, then run this prompt with the structure as context.

Related prompts

Hemingway rewrite — clarity and force without flattening

intermediate

Rewrite prose for clarity and force, cutting hedges and noise without losing the nuance that made the original worth reading.

Marketingframeworkmethodologyediting

Stress-test a strategic narrative against likely questions

advanced

Pressure-test a strategic narrative against the actual questions investors, board members, and employees would ask, then identify which questions the narrative cannot answer cleanly.

Executiveframeworkmethodologycommunications

Devil's advocate — strongest case against a plan, then triage

advanced

Argue the strongest case against a proposed plan, then identify which of those arguments are decisive versus which are just noise.

Executiveframeworkmethodologystrategy

Need help implementing this prompt in your workflow?

Book a call

You are preparing a one-page executive summary of a long document. Your reader has 3 minutes and needs to act, not be entertained. The summary must hold up if it is the only part the reader sees. <context> Source document: <<< {{document}} >>> Reader: {{reader}} (e.g., CEO, board chair, head of product) What the reader will do with this: {{reader_goal}} Length target: {{length}} (e.g., 250 words, half a page) </context> <task> Step 1 — Identify the decisions. Read the document and list every decision the reader is being asked to make, even implicitly. If there are none, say so explicitly. Step 2 — Pull the load-bearing facts. Extract the 5 to 8 facts in the document that would change a decision if they were different. Quote or paraphrase precisely. Cite the section or page. Step 3 — Surface the open questions. List anything the document treats as settled but that you, reading it cold, think the reader should push back on. Be specific. Step 4 — Write the summary. Produce a one-page executive summary in this exact structure: - **Bottom line** (1 to 2 sentences: what the reader needs to know if they read nothing else) - **What is being asked** (the decision or no-decision) - **Key facts** (3 to 5 bullets, each one fact) - **Risks and open questions** (2 to 4 bullets) - **Recommended next step** (one specific action with owner and timing if available) Step 5 — Honesty check. Note anything in the source document that you could not summarize confidently (ambiguous, contradictory, or missing context). One bullet each. </task> <output_format> ## Decisions identified ## Load-bearing facts (with citation) ## Open questions ## Executive summary (the deliverable) ## Confidence notes </output_format> <constraints> - Do not invent facts not in the source. If the source omits something the reader needs, name the gap in "Open questions." - Do not soften. If the document is making a weak case, the summary should reflect that without editorializing. - The "Bottom line" must be a claim, not a topic. "Q3 results" is a topic. "Q3 missed plan by 12% on a single deal slipping" is a claim. - US English, no em-dashes. </constraints>

Example

Input

document: [40-page board pre-read on Q3 results, including financials, product update, hiring report, and a proposed pivot in pricing model]
reader: Board chair
reader_goal: Decide whether to support the pricing pivot at next week's board meeting
length: 300 words

Output

## Decisions identified
- Approve or defer the pricing pivot from per-seat to consumption
- Approve Q4 hiring plan (15 net new)

## Load-bearing facts
- Q3 revenue $4.2M, 8% below plan; miss attributed to one enterprise deal slipping (p.4)
- Net retention 104%, down from 118% YoY; downsell concentrated in 50-200 seat tier (p.7)
- Pricing pilot with 12 customers showed 22% ARR uplift on consumption pricing (p.18)
- Cash runway 14 months at current burn; pivot adds ~$400k of one-time cost (p.31)

## Open questions
- The 22% uplift figure excludes 2 customers who churned mid-pilot. Including them, the picture is more mixed.
- Hiring plan assumes pricing pivot lifts pipeline, but the model linking the two is not in the deck.

## Executive summary

**Bottom line:** Q3 missed plan by 8% on one deal, not on a trend; the proposed pricing pivot has promising but incomplete pilot data, and approving it next week without seeing the churn-adjusted numbers is premature.

**What is being asked:** Board approval of the pricing pivot and Q4 hiring plan.

**Key facts:**
- Q3 revenue $4.2M, 8% below plan, driven by single deal slip
- Net retention down 14 points YoY, concentrated in mid-market
- Pricing pilot: +22% ARR but excludes 2 churned pilots
- 14 months cash; pivot adds $400k one-time

**Risks and open questions:**
- Pilot data not churn-adjusted; ask for the corrected number
- Hiring plan depends on pricing pivot lifting pipeline; the causal link is asserted, not modeled
- Mid-market downsell may be a product issue, not a pricing one — the deck does not test this

**Recommended next step:** Ask for the churn-adjusted pilot number and the pricing-to-pipeline model in writing before the meeting. Defer hiring plan vote one week if the pivot is deferred.

## Confidence notes
- "One deal slip" framing for the Q3 miss is the company's narrative. The deck does not show whether other deals shrank.

Tips for best results

1The 'load-bearing facts' step is what makes this stronger than the typical summary prompt — it forces Claude to identify which 5 facts actually matter, not paraphrase every section equally.

2Always specify the reader's goal. A summary for 'decide on the pivot' looks completely different from one for 'absorb context for next quarter.' Without it, you get a generic recap.

3Confidence notes catch hallucination. If Claude has high confidence on every point, push back: 'reread paragraph 4 and tell me what is actually said versus implied.'

4For very long documents (50+ pages), pre-process: ask Claude first to list the document's section structure, then run this prompt with the structure as context.