Appendix E of The Defensible Decision is the operational companion to Chapter 2. Chapter 2 covers the six-field cognitive minimum: the discipline of asking the right questions before any chart is built. Appendix E covers what the discipline turns into when the work has to ship: the eight-section template that gets signed off, the technique that produces a first-draft BRD from a meeting that already happened, and worked walkthroughs against five reference datasets.
This page hosts the assets the appendix references: the template in copy-paste markdown form, the AIRS completed exemplar, four starter BRDs intentionally underspecified for practice, and the three prompts that drive the transcript-driven elicitation loop.
Every artifact is available as both Markdown and Word (.docx). Markdown for copy-paste into Copilot or authoring in a text editor; Word for the review workflow analysts actually use to sign off on a BRD — edit in Word, route through stakeholders, get signatures.
The eight-section template
BRD Template (markdown)
The eight-section schema reproduced verbatim from Appendix E § Section 1. Each section's bracketed prompt explains what the section is for and also serves as the natural [PLACEHOLDER] marker when the elicitation loop encounters a gap.
Reference BRDs — five exemplars, four starters
The companion site ships nine reference BRDs across five completed exemplar domains and four starter datasets. The exemplars are the finished state to compare your work against; the starters are the exercise surface, each paired with a same-domain completed exemplar for direct comparison after you complete Sections 4–8.
AIRS Longitudinal — Completed Exemplar
A fully authored research-grade BRD with every section filled in to the level a real principal investigator would sign off on. The finished state a BRD reaches when the discipline has been applied end to end. Includes an optional Section 9 (Research Ethics) appropriate to the domain.
Read (Markdown) Download WordCloudRevenue — Completed Exemplar
Commercial-operations exemplar. VP of Cloud Sales Operations, Q4 portfolio investment decision, hard leadership-offsite deadline. Pairs with the Chapter 2 activity.
Read (Markdown) Download WordM365Marketing — Completed Exemplar
Marketing-operations exemplar. Director of Product Marketing (M365), weekly Monday-morning campaign reallocation, cross-platform attribution reconciliation across four paid channels. The hard-deadline analogue is Monday 8 AM dashboard availability.
Read (Markdown) Download WordSupportInsights — Completed Exemplar
Support-operations exemplar. Director of Customer Support, three weekly decisions (staffing, escalation thresholds, KB priorities), the largest primary-user stakeholder base of the five exemplars (five primary users to displace three legacy reports). Illustrates BRD-driven cross-team consensus.
Read (Markdown) Download WordEnterpriseGovernance — Completed Exemplar
Governance-program exemplar. Chief Data Officer, annual roadmap + quarterly investment review, twelve business units with structurally divergent stakeholder interests. Includes the optional Section 9 (Governance Maturity Model) with eight capabilities and a five-level scale. The most challenging exemplar for the disagreement-check validity test.
Read (Markdown) Download WordCloudRevenue — Starter
Sections 1–3 filled in. Sections 4–8 underspecified — the exercise. Stakeholder: VP of Cloud Sales Operations. Decision the chart informs: Q4 portfolio investment cuts. Compare against the CloudRevenue completed exemplar.
Read (Markdown) Download WordM365Marketing — Starter
Sections 1–3 filled in. Sections 4–8 underspecified. Stakeholder: Director of Product Marketing. Decision: weekly campaign-spend reallocation across twelve concurrent campaigns. Compare against the M365Marketing completed exemplar.
Read (Markdown) Download WordSupportInsights — Starter
Sections 1–3 filled in. Sections 4–8 underspecified. Stakeholder: Director of Customer Support. Decision: weekly staffing, escalation thresholds, and knowledge-base priorities. The "ticket" definition is itself one of the open problems. Compare against the SupportInsights completed exemplar.
Read (Markdown) Download WordEnterpriseGovernance — Starter
Sections 1–3 filled in. Sections 4–8 underspecified. Includes an optional Section 9 (Governance Maturity Model). Stakeholder: Chief Data Officer. Decision: annual governance-capability roadmap. The highest cross-stakeholder coordination cost of the four starters. Compare against the EnterpriseGovernance completed exemplar.
Read (Markdown) Download WordPick a starter whose domain is closest to your working context, complete it using the techniques in Appendix E and the prompts below, and compare your version against the same-domain completed exemplar (direct comparison) or the AIRS exemplar (cross-domain shape check). The point is not to match either exemplar field for field; it is to feel which sections are easy, which sections force a stakeholder conversation, and which sections you are tempted to skip because they are uncomfortable. The discomfort is the data.
The transcript-driven elicitation prompts
The three prompts the appendix references — one for the common case, two for the situations that need a separate first pass.
Template-Fill Prompt (one-pass)
The single prompt from Appendix E § Section 2 that does both jobs — extract the facts the transcript supports, then fill the eight-section template using only those facts, marking every gap explicitly. Engineered to keep the LLM from smoothing over gaps with plausible-sounding inference.
Read (Markdown) Download WordExtraction-Only Prompt
First-pass extraction for the three situations where the one-pass prompt is the wrong tool: transcripts longer than the LLM's working context, high-stakes BRDs where you want to audit extraction before it shapes the draft, and per-source passes when reconciling conflicting transcripts.
Read (Markdown) Download WordContradiction-Log Prompt
The third prompt — for when multiple transcripts, emails, or charters disagree. Surfaces every cross-source contradiction as a structured log so a human can resolve before the BRD is drafted. The discipline that prevents consensus that no one actually agrees to.
Read (Markdown) Download WordForthcoming sections (future editions)
Sections 1 and 2 are complete in this edition. The following sections are planned for future editions of Appendix E.
- Section 3 — Templates that survive review. Phrasing patterns for each BRD field; the difference between a BRD and a BRD-shaped artifact.
- Section 4 — Anti-patterns. The most common ways BRDs fail in practice, each named so they can be diagnosed when you see them.
- Section 5 — The stakeholder conversation. Scripts for opening the BRD discussion, navigating disagreement, and getting sign-off.
- Section 6 — The BRD and the AI prompt. How a well-authored BRD becomes the scaffold for the prompts that direct Copilot in Chapters 8 through 11.
- Section 7 — Completing the starter BRDs. Worked walkthroughs against each of the four starter dataset BRDs.
Comments and corrections welcome via the errata page.