Periodic Report Synthesizer

Read a stack of periodic reports from different authors and produce one cross-source synthesis memo: what everyone agrees on, where they disagree, who changed their stance or priorities, what's new, and who is contradicting their own prior report. The reader is a leader who has 20 reports in their inbox and one hour.

Why this skill exists

Every month or quarter, leaders receive a pile of reports and most get skimmed at best — regional managers' updates, vendor business reviews, business-unit or portfolio company board updates, grantee progress reports, sell-side notes on a category, newsletters from a dozen industry sources. The real signal isn't in any single report — it's across them: three regions independently reporting the same customer objection, one vendor quietly reversing what they promised six months ago, a risk everyone mentions in passing and nobody escalates. Reading for that signal manually takes a full day. This skill compresses it into a memo and preserves the receipts (quotes and page references) so every claim can be checked against the source.

Inputs

Required:

• Two or more periodic reports covering the same or overlapping period (PDFs, pasted text, or files in a folder). One report is not a synthesis — if only one is provided, suggest the executive-briefing skill instead.

Optional (each one materially improves the output):

• Prior-period reports from the same authors — enables the "shift vs. prior report" and "contradictions" sections
• The user's context: which authors matter most, what they own or are accountable for, how the reports feed a decision
• Specific questions ("what is everyone saying about hiring?" / "who mentions the pricing change?")
• Author labels — role, region, strategy, function — if not obvious from the reports

Missing-input behavior: If reports lack dates or author names, ask once for a quick mapping. Otherwise proceed — label any assumption explicitly (e.g., "Assumed Report 3 is Q1 2026 based on references to March data"). Never guess an author's role or mandate silently; mark it "role inferred from report content."

How it works

1. Inventory the stack. List every report: author, unit or organization, period covered, length, whether a prior report is available. Flag any report that is marketing material or a template with nothing filled in rather than a true periodic update.
2. Extract per report, before comparing. For each: views on the environment (market, customers, competitors, regulation, budget climate), key stance and priority statements, new initiatives, products, hires, or positions mentioned, initiatives dropped, paused, or cut, results commentary (only as stated — never compute or infer results), notable tone (defensive, confident, apologetic), and anything unusual (personnel notes, scope or terms changes, budget or headcount comments).
3. Build the theme map. Cluster views across authors: where do 2+ authors converge? Note whether convergence is independent observation or everyone citing the same data point, memo, or news story.
4. Find the divergences. Where do authors take opposite sides of the same question? These are the most valuable findings — quote both sides.
5. Diff against prior reports (if provided). For each author: what changed in stance, priorities, tone, or argument? Flag explicit reversals and silent reversals (a previously emphasized theme, initiative, or risk that simply disappeared).
6. Flag contradictions. An author saying something inconsistent with their own prior report, with their stated mandate, or with another author's account of the same event gets quoted side by side with both dates.
7. Write the memo using the output format below, with every claim attributed to a specific author and report.

Output format

# Periodic Report Synthesis — [Period]
Reports reviewed: [N] | Authors: [list] | Prior reports available: [Y/N per author]

## 1. One-paragraph read
[The 3–4 sentences you'd say to the CEO in the hallway.]

## 2. Shared themes (convergence)
- **[Theme]** — raised by [Author A, B, C]. [1–2 sentences each on how they express it.]
  > "[Short supporting quote]" — [Author], [period report], p.[X]
- Consensus note: [is this theme consensus across the stack, and is it independent or one shared source?]

## 3. Divergent views
- **[Question]**: [Author A] argues [X]; [Author B] argues the opposite.
  > A: "[quote]"  |  B: "[quote]"

## 4. Shifts vs. prior reports
| Author | Then ([prior period]) | Now ([current period]) | Shift |
|---|---|---|---|

## 5. Notable new / dropped initiatives, priorities, and positions
[Per author, only as explicitly disclosed in the report.]

## 6. Contradictions and silent reversals
- **[Author]**: said "[quote]" ([prior date]) vs. "[quote]" ([current date]). [Why it matters; possible benign explanation.]

## 7. Tone and red-flag watch
[Defensiveness, blame-shifting, metric or baseline switching, moved goalposts, personnel/budget/scope mentions.]

## 8. Questions for follow-up conversations
[3–7 specific questions, each tied to a finding above and addressed to a named author.]

## Not covered in these reports
[Topics the user might expect that no author addressed.]

Adjust length to the stack: 5 reports → ~2 pages; 20 reports → 4–5 pages with the table doing more work.

Guardrails

• Quote, don't paraphrase, for anything contentious. Every contradiction, reversal, or red flag must carry a verbatim quote with author name and report date. If you can't find the quote, drop the claim.
• Never invent or compute results figures. Report numbers only as stated in a report, with the as-of date. Do not annualize, net, re-baseline, or compare figures the authors didn't compare themselves.
• Separate fact from inference. "Region A paused the rollout" (stated) vs. "Region A appears to have lost confidence in the rollout" (inference — label it as such).
• Read against the author's incentive. A vendor's business review, a business unit's board update, and a fund manager's letter are all written by someone with a stake in how they're read. Note where an author's framing serves the author, without assuming bad faith.
• No advice. The memo informs follow-up conversations and decisions; it does not recommend budget moves, vendor changes, allocations, or personnel actions.
• Confidentiality. Periodic reports are often confidential — board updates, vendor reviews, investor communications, grantee reports. Remind the user not to circulate the synthesis outside the group entitled to see the underlying reports.
• A missing theme is data, not license to speculate. If no author discussed something, say so — don't fill the gap with your own view.

Quality checks

Before delivering, verify:

• Every author in the stack appears at least once in the memo
• Every quote has an author name and report period attached
• No results figure appears that isn't verbatim from a report
• Each contradiction shows both quotes and both dates
• Inferences are labeled as inferences
• The follow-up questions map back to specific findings and name who to ask

Example prompts

"Here are the monthly reports from our eight regional sales directors, plus last month's. Synthesize: what themes are shared, where they disagree, who shifted their priorities, and whether anyone is contradicting what they told us last month. We're especially interested in what they're saying about the pricing change."

"Attached are quarterly business reviews from our five biggest vendors. Tell me what they're all saying about their roadmaps, where their accounts of our joint projects don't match each other, and who's quietly walked back a commitment from the last QBR."

"Twelve grantee progress reports for the year, plus their mid-year reports. What outcomes are they converging on, where do their explanations for delays diverge, and which grantees dropped a workstream without saying so?"

"Q1 letters from our six fund managers, plus their Q4 letters. What themes are shared, where they disagree, who shifted positioning, and whether anyone is contradicting what they told us last quarter — particularly on credit."

Works well with

Use presentation-transcript-extractor first when some of the "reports" are recorded calls or town halls rather than documents. Feed findings into stakeholder-update-generator when the synthesis needs to become your own upward report, and into executive-briefing when leadership needs the one-paragraph read as a standalone.