Why this session exists
AI made drafting free, which means everyone's deck now has words in it. What AI did not make free is the thing partners are actually paid to check: does the first line answer the question, does every claim carry a number, does every fact end in a so-what, and would the client recognize their own company in it? Those are measurable properties of text - so this course measures them, live, on whatever is on the page. The pyramid gives you the theory; the scorer gives you the reps.
Minto's pyramid: the answer goes first 8 min live
Barbara Minto - widely described as McKinsey's first female MBA consultant, inventor of MECE by her own cheerful account, and author of The Pyramid Principle - codified the structure consulting still runs on: lead with the governing thought (the answer, stated as a recommendation), support it with a small set of mutually exclusive, collectively exhaustive arguments, and let every heading do work. Readers who get the answer first evaluate everything after it; readers kept waiting just get anxious.
LiveAction titles: headings that do work4 min▶
The pyramid's most visible descendant in consulting decks is the action title: a heading that states the takeaway rather than the topic. "Retention analysis" is a topic title - it tells the reader where they are. "Enterprise seats retain at 94% vs 71%, so converted accounts compound" is an action title - it tells them what the page proves. A partner flipping the deck in the elevator reads only the titles; if the titles alone tell the story, the storyline works.
- Topic title: "Market pricing" - the reader must do the thinking.
- Action title: "Rivals price at $38-45, leaving room to enter at $30" - the thinking is done and signed.
AI, asked naively, produces topic titles - they are safer, because they claim nothing. Asked properly (draft the takeaway of this exhibit as one sentence with the number in it), it produces raw material for action titles you then verify and sharpen. The scorer's answer-first and so-what measures reward exactly this.
Self-studyMECE without the mysticism2 min▶
MECE - mutually exclusive, collectively exhaustive - just means your support arguments do not overlap and do not leave holes. The working test: could a skeptic say "arguments 1 and 2 are the same point twice" (not mutually exclusive) or "you never addressed X" (not collectively exhaustive)? On the Cadence pyramid: demand, economics, timing - three different failure modes for the recommendation, each answered once. Perfection is not required; visible discipline is. The scorer does not measure MECE directly (no regex can), which is a useful reminder that the scorer is a floor, not a ceiling - part 2 says more.
Six measures, one honest caveat 7 min live
The lab below reads your executive summary the way a tired partner does at 11pm - pattern-matching for the signals of rigor. Six measures, 100 points. They are real checks computed on the real text, and they are also, deliberately, not the whole of quality - the caveat comes with the table.
LiveThe scoring table4 min▶
| Measure | Points | What it checks on the page |
|---|---|---|
| Answer-first opening | 20 | line one contains a recommendation verb AND a number - the governing thought, present and quantified |
| Quantified support | 20 | share of bullets carrying a number - claims with evidence attached |
| So-what density | 15 | share of bullets ending in a consequence ("which means", "so", "worth") |
| Weasel-word discipline | 15 | each hit from the weasel list ("holistic", "robust", "leverage synergies", "in today's...") costs 3 |
| Client anchors | 15 | the client's actual specifics present - Cadence, seats, churn, pilots, procurement |
| Length discipline | 15 | 90-160 words - an executive summary, not an executive essay |
The caveat, stated plainly: these are proxies. A summary can score 100 and still recommend the wrong thing - the scorer checks craft signals, not truth (weeks 2-3 were about truth, and session b9's audit re-checks it). What the proxies catch superbly is the most common real-world failure: fluent text that commits to nothing, quantifies nothing, and could be about any company. You will meet that text in its purest form next session.
Self-studyWhy "so" is worth 15 points2 min▶
A bullet that states a fact - "enterprise seats retain at 94% vs 71%" - hands the partner homework: so what? The consequence clause finishes the thought: "...so every converted account compounds instead of churning." Minto called the discipline vertical logic (each level answers the question the level above raises); modern consulting slang calls it "so-what-ing" a page. It is the cheapest upgrade in professional writing, and the ladder's third rung below is worth 15 points precisely because most drafts skip it.
The storyline lab ▶ ★ 20 min · everyone climbs the ladder
Five buttons, five drafts of the same Cadence executive summary. Every press swaps the text and re-measures it live - the score is always computed from what is actually on the page. Climb in order; the point is feeling WHICH move buys which points.
0 · The intern's draft: 19. Read it - it sounds like strategy. Now read the table: no answer in line one, zero numbers, four weasels. The scorer prices what is on the page, and almost nothing is.
1 · Answer first: 38. One sentence changed. The governing thought lands with a verb and real numbers - $30 per seat, Q1, 210 accounts. Minto's rule alone nearly doubles the score, and the bullets are still fluff.
2 · Quantify every claim: 78. The biggest single jump on the ladder - 40 points. Every bullet now carries evidence (62%, 94% vs 71%, $38-45), and notice the weasel count collapsed without a dedicated pass: vague words die when numbers take their seats.
3 · So-what every bullet: 93. Each fact now ends in its consequence - "which means demand is already organized", "so every converted account compounds". The partner's homework, done in advance.
4 · Anchor to the client: 100. Cadence named, the pilot-to-procurement mechanics visible, the bottleneck called. Compare drafts 0 and 4 side by side: same topic, same author, different planet. Every point of the climb was a learnable, repeatable move.
Now break it on purpose. The ladder is scripted but the scoring is not: mentally delete the numbers from draft 4 and predict the score; the table updates live in session 6's lab where you get free text. Prediction-then-check is how the measures become instinct.
Homework 📝 ◐ 25 minutes
- Rewrite a real summary you have shipped using the ladder: answer first, quantify, so-what, anchor. Keep before and after - session b10's lab has a paste box that will grade both.
- Collect three action titles from any deck you can access, and rewrite one topic title into an action title with a number in it.
- Write the Cadence governing thought from memory - one sentence, verb plus numbers. If you cannot, reread the case facts in b1's cheat sheet; the scorer will keep asking.
Three questions before you go 🎯 ◐ 90 seconds
1 · Minto's governing thought means the document opens with...
Answer first: a reader who stops after line one still leaves with the recommendation. Context and method exist to serve the answer, not to delay it - that inversion is the whole pyramid.
2 · The biggest single jump on the lab's ladder (40 points) came from...
Quantify-every-claim took 38 to 78 - and the weasels collapsed as a side effect, because vague words die when numbers take their seats. Evidence attached beats adjectives deleted.
3 · The honest caveat about the scorer is...
The measures are real pattern checks and superb at catching fluent emptiness - but truth lives in the research and analysis (b3, b4) and gets re-verified in b9's audit. The scorer is a floor, not a ceiling.
What this session covered from the official universe
Certificates, videos, and graded assessments stay with the official providers - this course teaches the working core with a live scorer.