Render One Analysis for Four Audiences Without Retyping It Four Times

Workflow area: Stakeholder Communication · Tier: Foundation · Hands-on time: ~25 minutes

Analysts lose entire days to reformatting: the steering committee wants half a page, the delivery team wants tables and IDs, the vendor needs unambiguous scope language, and end users need to know what changes for them on Monday. Writing four documents from scratch invites the worst failure in stakeholder communication — version drift, where the exec deck says 4 days and the vendor letter says 5 because they were edited separately. The fix is a single source-of-truth pack rendered into audience-specific outputs, with a verification pass per rendering. One truth, four voices.

When to use — and when not to

• Use it at every analysis milestonefindings ready, options decided, scope agreed — where three or more distinct audiences need the same content at different altitudes.
• Use it when version drift has already bittenif two documents have disagreed in public, moving to render-from-one-source is the structural fix, not more careful editing.
• Use it for recurring reportinga monthly program update rendered for steering, delivery, and business audiences from one status pack keeps the three consistent by construction.
• Skip it for genuinely different contentif the exec message and the team message must differ in substance (a reorg, a vendor dispute), that is judgment work, not rendering; write them by hand.
• Skip the vendor rendering without contract reviewanything that could be read as a scope commitment goes past whoever owns the commercial relationship before it leaves the building.

The recipe

1. 1.

   Build the source-of-truth pack. One document containing: the findings with citations (from pb-01), the numbers with sources and evidence grades (from pb-09), decisions with owners, open questions, and recommended next steps. Every fact appears here exactly once. If it is not in the pack, no rendering may contain it.

2. 2.

   Define the audience matrix. For each audience, one line each: what they must decide or do after reading, what they already know, what altitude (page count, tables vs prose), and their sensitivities. Two minutes of thinking that transforms output quality.

3. 3.

   Render all four in one pass, bound to the pack.

   Copy

   You are rendering ONE analysis for four audiences. Use ONLY the
   source pack below. Every number must appear exactly as in the
   pack, with its source tag. Invent nothing; omit freely.
   
   AUDIENCE MATRIX:
   1. EXEC SUMMARY — audience: steering committee; decision: approve
     option B and its budget; half a page; leads with the "so what";
     numbers as ranges with confidence, no jargon.
   2. DELIVERY DETAIL — audience: dev team + QA; action: refine into
     backlog; full detail, tables, requirement IDs, open questions
     flagged as OPEN with owners.
   3. VENDOR SCOPE LETTER — audience: implementation vendor;
     action: bounded estimate; formal tone; explicit in-scope /
     out-of-scope lists; no internal politics, costs, or stakeholder
     names; unresolved items listed as "to be confirmed", never
     silently resolved.
   4. END-USER FAQ — audience: AP clerks affected by the change;
     action: know what changes for them and where to ask questions;
     plain language, Q&A format, honest about what is not decided.
   
   For EACH rendering, end with a SO-WHAT line: the single sentence
   this audience should remember. Make it specific to their stake —
   do not reuse the same sentence across audiences.
   
   SOURCE PACK:
   [PASTE SOURCE-OF-TRUTH PACK]
   

4. 4.

   Check a sample of good output for the failure patterns.

   Copy

   EXEC SUMMARY (extract)
   Duplicate vendor records cost an estimated $38k–$71k annually in
   reconciliation labor [S: sizing model v2, base $53k]. Option B
   (detection at entry) removes ~80% of new duplicates for a one-time
   build of ... SO WHAT: approving option B this month means the
   Q4 close runs without the reconciliation backlog.
   
   END-USER FAQ (extract)
   Q: Will this change how I enter vendors?
   A: Yes — one new step. If the system finds a possible duplicate,
   you'll see the existing record and choose "use existing" or
   "create anyway with a note". Nothing else in your screen changes.
   

5. 5.

   Run the two-pass verification. Pass one, numbers: every figure in every rendering, checked against the pack — this is where "up to 80%" quietly becomes "80%", and ranges collapse into their upper bounds. Pass two, audience safety: read the vendor letter as the vendor's lawyer (any accidental commitments? internal information leaked?), read the FAQ as a nervous clerk (does it overpromise? does it respect that their workaround knowledge matters?), read the exec summary as the one skeptical director (can every claim survive "says who?").

6. 6.

   Rewrite the four so-what lines yourself. The model's so-whats are competent; yours know that the CFO cares about the Q4 close and the AP team fears headcount subtext. This is the highest-leverage two minutes in the playbook — the so-what is the only line most readers retain.

7. 7.

   Version the pack, not the renderings. When facts change, update the pack, re-render, re-verify. Never patch a rendering directly — that is how drift returns.

Make it yours

• Claude Projects as the standing renderer: pin the audience matrix and house style per project; each milestone then needs only the updated pack. The matrix improves as you learn each audience's reactions.
• Notion AI for wiki-native teams: keep the source pack as the parent page and renderings as child pages, regenerated in place — the hierarchy makes "which is the source of truth" self-documenting.
• Gamma-class tools for the deck step: feed the verified exec summary (not the raw pack) into deck generators to get slides — verify content before it acquires design polish, because polished wrong numbers are the most persuasive kind.
• Team scale: standardize the audience matrix across analysts so every project's steering update has the same shape — reviewers stop re-learning formats and start reading content.

Quality check

• Every number in every rendering matches the pack, with its sourceone full pass per rendering; drift starts with a single unsourced figure.
• Qualifiers survived translationranges stayed ranges, "estimated" stayed attached, and no "up to" was dropped in simplification.
• The vendor letter contains zero internal informationno stakeholder names, cost baselines, or political context; and its scope lists were reviewed by the commercial owner.
• The FAQ is honest about undecided itemsend users forgive "not decided yet" and never forgive discovering the FAQ was spin.
• Each so-what is audience-specific and human-writtenfour different sentences, each naming that audience's actual stake.
• The pack is the only edited artifactno rendering has been hand-patched since its last generation.