Skip to content

Consulting Knowledge

Updated: 2026-06-04

Durable consulting frameworks for external client work. Written for AI retrieval — dense and structured.


Delivery hierarchy (read first)

Every client engagement follows this order. Do not skip steps or invert the stack.

Priority Layer What it is Where it lives Gate
1 Key message One conclusion the client must leave with — a decision, judgment, or "so what" brief.mdinsights.md Core message Written and signed off before storyline
2 Storyline Logical argument that proves the key message insights.md Storyline Every beat must trace to the key message
3 Fact base Sourced data that supports each storyline beat research.md, sources.md Every headline number maps to a beat + source
4 Deck Slides that execute the storyline deck-outline.md → render Each slide = one beat; fact-based

Rules

  • One key message per project — if you need "and also," split or sharpen
  • Storyline before deck — no deck-outline until Core message + Storyline are approved
  • Research in service of storyline — gather facts to fill gaps in the argument, not topic dumps
  • No orphan facts — if a data point does not support the storyline, it goes in the appendix

Workflow order

brief.md (decision + audience)
  → insights.md: Core message [sign-off]
  → insights.md: Storyline [sign-off]
  → research.md + sources.md (facts for beats only)
  → deck-outline.md → render

Core frameworks

Problem structuring

  • Define the question precisely before researching — a bad question produces useless research
  • Use MECE to break down issues without gaps or overlaps
  • Always ask "so what?" — insights only matter if they lead to a decision or action

Communication (Pyramid Principle)

  • Lead with the key message — conclusion first, then support it
  • Structure: Key message → High-level supporting points → Facts / evidence
  • Each level must logically prove the level above it
  • Be concise: if a message takes more than 2 lines, it contains more than one idea

SCR narrative

  • Situation: what is true today (non-controversial facts)
  • Complication: why the situation creates a problem or opportunity
  • Resolution: what the client should do about it

Slide writing principles

Golden rules

  1. One message per slide — two insights = two slides
  2. Action titles, not topic titles — "Market is shrinking 3% YoY" not "Market Overview"
  3. Title ≤ 2 lines — if longer, the message isn't sharp enough
  4. Body proves the title — every element supports the slide message (vertical logic)
  5. Titles tell the story — reading only titles top to bottom reveals the full argument (horizontal logic)
  6. "So what?" rule — every chart and bullet must answer what it means for the client

Standard slide types

slide_type Purpose Implication required?
cover Title, date, recommendation preview No
exec_summary One-page SCR or key-message stack No
agenda Section signposting No
context Situation or Complication (SCR) Optional
finding One insight that changes the decision Yes
data Quantitative proof Yes
framework Prioritization or strategic structure (2×2, pyramid) Yes
process Workflow, journey, operating model Yes
comparison Options, vendors, scenarios Yes
roadmap Phasing, milestones Yes
recommendation What to do next; economics Yes
summary Wrap-up, next steps No
appendix Methodology, raw data No

Design rules

  • 3 colors max — one primary, one accent, one neutral
  • No 3D charts — always flat
  • Label the conclusion, not all the data
  • Always cite sources with dates at the bottom
  • Add a 1-line implication box on data/finding slides

Project folder structure

Each client project lives in wiki/consulting/[project-name]/:

brief.md        ← audience, decision, scope
research.md     ← sourced data, mapped to storyline beats
insights.md     ← Core message + Storyline (sign-off required)
deck-outline.md ← slide-by-slide outline
sources.md      ← all source URLs + access dates

Mistakes to avoid

  • Researching before locking the key message — produces unfocused content
  • Using topic titles on slides ("Market Overview") instead of action titles
  • Fact dumps — data for its own sake with no "so what"
  • Building the deck before the storyline is approved
  • Missing source dates on financial or market data