Learnings 25
Supported 18
Weakened 2
Broken 5
Insufficient 0
all document kinds / all orientations

Canon library

The 25 theses are the product. The stories are exhibits.

Each learning is a visual evidence card: claim, live metric, scope, tier, examples and canon links. Broad theses can travel; annotation leads stay marked as leads until enrichment makes them robust.

Global canon candidates

Claims that can generalize across the corpus

Only theses whose support is not tied to one document kind and is not merely an annotation slice. These are closest to corpus canon today.
4/4 live / 4 broad / 0 annotated
#19 Live 100% scope

Charts are punctuation, not the sentence

Across the current metadata layer, only a minority of slides carry charts, so charts punctuate arguments rather than replace them.

slides with charts24.4%Broad corpus base
319,974 metadata slides / 100% of 319,975 rendered slides in cohort 3 examples

Use charts where they settle a comparison. Use prose, structure or examples everywhere else.

Proof, use and examples
Human lesson

A deck is not "data-driven" because every slide has a chart; it is data-driven when charts arrive at the decisive moment.

Move to try

Use charts where they settle a comparison. Use prose, structure or examples everywhere else.

Evidence contract chart_slide_share

Share of slides in the current metadata layer with at least one chart component.

319,974 metadata slides out of 319,975 rendered slides in cohort (100% coverage); the denominator is the current slide-metadata layer for all slides with metadata.

Test: <= 32%. Refreshed 7/17/2026, 4:02:16 AM.

Chart slides78,138Metadata slides319,974Chart components122,056Docs hit8,655
#20 Live 100% scope

Bars and lines win because bosses compare.

Most executive decisions need magnitude or movement before they need visual novelty.

bar/line chart components74.8%Broad corpus base
122,056 chart components 3 examples

Default to bar for magnitude and line for time unless the argument demands another grammar.

Proof, use and examples
Human lesson

Most executive arguments need comparison before decoration.

Move to try

Default to bar for magnitude and line for time unless the argument demands another grammar.

Evidence contract chart_subkind_share

Share of chart components whose subtype is a bar or line.

122,056 chart components; this is within detected chart components, not all slides.

Test: >= 65%. Refreshed 7/17/2026, 4:02:16 AM.

Bar/line charts91,337Chart components122,056Docs hit8,162
#21 Live 100% scope

Dense is not guilty. Confusing is.

The corpus has many dense slides; the crime is not density, it is making the point hard to find.

dense slides32%Broad corpus base
319,974 slides with density metadata / 100% of 319,975 rendered slides in cohort 3 examples

Ask whether the density creates comparison, sequence or proof. If not, split it.

Proof, use and examples
Human lesson

Reviewers should preserve useful density and attack only the slides that make the point hard to find.

Move to try

Ask whether the density creates comparison, sequence or proof. If not, split it.

Evidence contract density_share

Share of slides in the current metadata layer tagged dense.

319,974 slides with density metadata out of 319,975 rendered slides in cohort (100% coverage); the denominator is the current slide-metadata layer for all slides with density metadata.

Test: >= 25%. Refreshed 7/17/2026, 4:02:16 AM.

Dense slides102,246Density metadata slides319,974Docs hit9,866
#22 Live 100% scope

Callouts are the reader rail

Callouts appear often enough to be treated as a core reading aid, not a decorative box.

slides with callouts30.2%Broad corpus base
319,974 metadata slides / 100% of 319,975 rendered slides in cohort 3 examples

Use callouts for the conclusion or exception, not to repeat the title.

Proof, use and examples
Human lesson

A callout gives the skimmer a safe path through a busy slide.

Move to try

Use callouts for the conclusion or exception, not to repeat the title.

Evidence contract component_slide_share

Share of slides in the current metadata layer with at least one callout component.

319,974 metadata slides out of 319,975 rendered slides in cohort (100% coverage); the denominator is the current slide-metadata layer for all slides with metadata.

Test: >= 20%. Refreshed 7/17/2026, 4:02:16 AM.

Slides hit96,741Metadata slides319,974Components117,153Docs hit8,431

By kind / consulting decks

What appears specific to consulting-style decks

Useful and often punchy, but not universal unless later enrichment shows the same behavior outside consulting decks.
3/3 live / 2 broad / 0 annotated
#01 Live 97.2% scope

The opener gets love. The ending gets abandoned.

Consultants know how to land the first punch; too often they leave the decision slide to fend for itself.

opening-closing gap12.8 ptsFocused sample
2,073 reviewed docs / 97.2% of 2,133 docs in cohort 3 examples

Write the final three slides before polishing the opener: decision, action, owner.

Proof, use and examples
Human lesson

The last slides are where agreement should turn into action; weak closings make the deck feel smart but unfinished.

Move to try

Write the final three slides before polishing the opener: decision, action, owner.

Evidence contract review_score_gap

Average Storymakers review opening score minus closing score.

2,073 reviewed docs out of 2,133 docs in cohort (97.2% coverage); the score is only as broad as the reviewed slice.

Test: >= 8pts. Refreshed 7/17/2026, 4:02:16 AM.

Opening avg57 ptsClosing avg44.2 ptsGap12.8 ptsNarrative avg58.9 pts
#06 Live 100% scope

Consultants prove before they persuade.

The consulting deck usually earns action through structured analysis, not through plot alone.

slides that analyze data21.6%Broad corpus base
88,048 metadata slides 3 examples

Before the recommendation, force one slide to decompose the problem and one slide to quantify the gap.

Proof, use and examples
Human lesson

A consulting deck that jumps to recommendation too early feels asserted, not proven.

Move to try

Before the recommendation, force one slide to decompose the problem and one slide to quantify the gap.

Evidence contract slide_function_share

Share of consulting slide metadata tagged analyze_data.

88,048 metadata slides out of 88,048 rendered slides in cohort; the denominator is the current slide-metadata layer for consulting slides with metadata.

Test: >= 20%. Refreshed 7/17/2026, 4:02:16 AM.

Slides hit19,048Metadata slides88,048Docs hit1,810
#10 Live 100% scope

Dense is allowed. Lost is not.

A dense slide can work when it creates comparison; it fails when the reader has to hunt for the point.

overcrowded slides15.2%Broad corpus base
88,048 slides with density metadata 3 examples

Keep dense slides when they create comparison; split them when the reader must hunt for the point.

Proof, use and examples
Human lesson

This lets reviewers critique overloaded slides without pretending every dense analytical slide is bad.

Move to try

Keep dense slides when they create comparison; split them when the reader must hunt for the point.

Evidence contract density_share

Share of consulting slides tagged overcrowded.

88,048 slides with density metadata out of 88,048 rendered slides in cohort; the denominator is the current slide-metadata layer for consulting slides with density metadata.

Test: <= 20%. Refreshed 7/17/2026, 4:02:16 AM.

Overcrowded slides13,383Density metadata slides88,048Docs hit1,722

By kind / activist and research decks

What appears specific to contrarian persuasion

These explain campaign pressure mechanics. They are real for activist/research material, but should not be taught as universal deck law.
2/8 live / 7 broad / 0 annotated
#11 Broken 100% scope

Contrarian decks need a culprit.

A gap becomes pressure when the deck names the actor, incentive or governance failure behind it.

slides naming a villain3.4%Broad corpus base
33,589 metadata slides 3 examples

State the failure mode and the accountable actor in the same sentence.

Proof, use and examples
Human lesson

A gap without an accountable actor is a spreadsheet; a named villain creates pressure.

Move to try

State the failure mode and the accountable actor in the same sentence.

Evidence contract slide_function_share

Share of activist and research slide metadata tagged name_villain.

33,589 metadata slides out of 33,589 rendered slides in cohort; the denominator is the current slide-metadata layer for activist + research slides with metadata.

Test: >= 8%. Refreshed 7/17/2026, 4:02:16 AM.

Slides hit1,137Metadata slides33,589Docs hit300
#13 Broken 100% scope

Peer gaps turn failure into choice.

The peer set says: others already solved this, so underperformance is no longer abstract.

peer-gap slides0.5%Broad corpus base
27,239 metadata slides 3 examples

Show the target, the peer set and the avoidable delta in one frame.

Proof, use and examples
Human lesson

A peer gap converts "could improve" into "others already do this."

Move to try

Show the target, the peer set and the avoidable delta in one frame.

Evidence contract slide_type_share

Share of activist slide metadata tagged peer_gap.

27,239 metadata slides out of 27,239 rendered slides in cohort; the denominator is the current slide-metadata layer for activist slides with metadata.

Test: >= 6%. Refreshed 7/17/2026, 4:02:16 AM.

Slides hit142Metadata slides27,239Docs hit91
#14 Broken 100% scope

Make management testify against itself.

A CEO quote is strongest when the next slide turns the company's words into evidence against it.

CEO quote slides0.8%Broad corpus base
27,239 metadata slides 3 examples

Use quotes only when the next visual proves or breaks the promise.

Proof, use and examples
Human lesson

A quote lowers the burden of proof: the target supplies the promise, the deck supplies the gap.

Move to try

Use quotes only when the next visual proves or breaks the promise.

Evidence contract slide_type_share

Share of activist slide metadata tagged ceo_quote.

27,239 metadata slides out of 27,239 rendered slides in cohort; the denominator is the current slide-metadata layer for activist slides with metadata.

Test: >= 8%. Refreshed 7/17/2026, 4:02:16 AM.

Slides hit219Metadata slides27,239Docs hit89
#15 Broken 100% scope

A critique without an ask is just mood.

The activist deck becomes actionable when the outrage resolves into a concrete demand.

demand-list slides0.3%Broad corpus base
27,239 metadata slides 3 examples

Make the demand slide impossible to confuse with another evidence slide.

Proof, use and examples
Human lesson

Without an ask, the audience can agree with the diagnosis and still do nothing.

Move to try

Make the demand slide impossible to confuse with another evidence slide.

Evidence contract slide_type_share

Share of activist slide metadata tagged demand_list.

27,239 metadata slides out of 27,239 rendered slides in cohort; the denominator is the current slide-metadata layer for activist slides with metadata.

Test: >= 2%. Refreshed 7/17/2026, 4:02:16 AM.

Slides hit88Metadata slides27,239Docs hit59
#17 Broken 100% scope

Footnotes are armor.

In contrarian work, sourcing is not housekeeping; it is how the accusation survives scrutiny.

slides carrying footnotes0%Broad corpus base
33,589 metadata slides 0 examples

Source the slide's most damaging sentence, not only the raw data table.

Proof, use and examples
Human lesson

The footnote says "this claim can survive discovery" before the audience checks it.

Move to try

Source the slide's most damaging sentence, not only the raw data table.

Evidence contract component_slide_share

Share of activist and research slides with at least one footnote component.

33,589 metadata slides out of 33,589 rendered slides in cohort; the denominator is the current slide-metadata layer for activist + research slides with metadata.

Test: >= 8%. Refreshed 7/17/2026, 4:02:16 AM.

Slides hit0Metadata slides33,589Components0Docs hit0
#16 Weak 100% scope

The payoff is where outrage becomes money.

The valuation reveal is the moment a governance or operating critique gets priced.

valuation-reveal slides0.7%Broad corpus base
33,589 metadata slides 3 examples

Delay the valuation bridge until the operational or governance failure has been made visible.

Proof, use and examples
Human lesson

The audience may debate governance, but upside/downside gives the story a price.

Move to try

Delay the valuation bridge until the operational or governance failure has been made visible.

Evidence contract slide_type_share

Share of activist and research slide metadata tagged valuation_reveal.

33,589 metadata slides out of 33,589 rendered slides in cohort; the denominator is the current slide-metadata layer for activist + research slides with metadata.

Test: >= 1%. Refreshed 7/17/2026, 4:02:16 AM.

Slides hit219Metadata slides33,589Docs hit132
#12 Live 100% scope

The kill shot is contradiction.

The contrarian move is to put management's promise beside reality and let the mismatch do the damage.

slides exposing contradiction40.5%Broad corpus base
33,589 metadata slides 3 examples

Pair the public claim with the counter-metric, then write the gap as the title.

Proof, use and examples
Human lesson

Contradiction is more persuasive than complaint because the audience can see the standard being violated.

Move to try

Pair the public claim with the counter-metric, then write the gap as the title.

Evidence contract slide_function_share

Share of activist and research slide metadata tagged expose_contradiction.

33,589 metadata slides out of 33,589 rendered slides in cohort; the denominator is the current slide-metadata layer for activist + research slides with metadata.

Test: >= 22%. Refreshed 7/17/2026, 4:02:16 AM.

Slides hit13,590Metadata slides33,589Docs hit565
#18 Live 100% scope

Good contrarian decks prosecute; they do not rant.

The useful campaign deck has a charge, evidence, motive, damages and an ask.

activist docs with thesis summary74.4%Focused sample
399 activist docs 3 examples

Before designing slides, write the one-liner, three reasons, proof KPIs and ask.

Proof, use and examples

Annotation leads

Promising patterns that need broader coverage

High hit rates inside annotated docs are not enough for canon. Keep them as leads until enrichment expands the denominator.
9/10 live / 0 broad / 10 annotated
#09 Weak 92.8% scope

Facts need a so-what ladder.

A finding only becomes useful when the deck climbs from what happened to what must change.

loop-annotated docs using the loop11%Annotated slice
1,979 loop-annotated docs / 92.8% of 2,133 docs in cohort 3 examples

Under every key chart write three lines: what happened, why it matters, what we do.

Proof, use and examples
Human lesson

Without the cascade, the deck has data but no managerial consequence.

Move to try

Under every key chart write three lines: what happened, why it matters, what we do.

Evidence contract loop_doc_share

Share of consulting documents with any loop annotation that also use So What Cascade.

1,979 loop-annotated docs out of 2,133 docs in cohort (92.8% coverage); this is an annotation slice, not whole-corpus prevalence.

Test: >= 12%. Refreshed 7/17/2026, 4:02:16 AM.

Docs hit217Annotated docs1,979Loop ranges278Avg confidence68.6%
#02 Live 98.1% scope

The title is the argument.

The strongest consulting slides do not label the topic; they spend the title line on the point.

docs with action-title evidence37.8%Annotated slice
2,092 tool-annotated docs / 98.1% of 2,133 docs in cohort 3 examples

Rewrite every slide title as a complete sentence with a verb and a consequence.

Proof, use and examples
Human lesson

Readers skim titles first; a topical title wastes the only line everyone will read.

Move to try

Rewrite every slide title as a complete sentence with a verb and a consequence.

Evidence contract tool_doc_share

Share of consulting documents with any tool annotation that also carry Action Titles evidence.

2,092 tool-annotated docs out of 2,133 docs in cohort (98.1% coverage); this is an annotation slice, not whole-corpus prevalence.

Test: >= 20%. Refreshed 7/17/2026, 4:02:16 AM.

Docs hit790Annotated docs2,092Match rows11,749Avg confidence78.3%
#03 Live 98.1% scope

Vague language is where weak thinking hides.

When a deck names actors, numbers and consequences, the idea has fewer places to hide.

docs with concrete-language evidence19%Annotated slice
2,092 tool-annotated docs / 98.1% of 2,133 docs in cohort 3 examples

Replace one abstract noun per slide with a customer, cost, process, competitor or behavior.

Proof, use and examples
Human lesson

Abstract copy hides weak thinking. Concrete copy exposes whether the idea can survive contact with a slide.

Move to try

Replace one abstract noun per slide with a customer, cost, process, competitor or behavior.

Evidence contract tool_doc_share

Share of consulting documents with any tool annotation that also carry Concrete Language evidence.

2,092 tool-annotated docs out of 2,133 docs in cohort (98.1% coverage); this is an annotation slice, not whole-corpus prevalence.

Test: >= 18%. Refreshed 7/17/2026, 4:02:16 AM.

Docs hit397Annotated docs2,092Match rows2,407Avg confidence76.3%
#04 Live 98.1% scope

Charts need a narrator.

A chart becomes persuasive when the slide tells the reader which movement, gap or exception matters.

docs with annotation evidence26.5%Annotated slice
2,092 tool-annotated docs / 98.1% of 2,133 docs in cohort 3 examples

Add one annotation that names the inflection, gap or threshold the reader should not miss.

Proof, use and examples
Human lesson

A chart without annotation asks the audience to do the presenter's reasoning in real time.

Move to try

Add one annotation that names the inflection, gap or threshold the reader should not miss.

Evidence contract tool_doc_share

Share of consulting documents with any tool annotation that also carry Annotation evidence.

2,092 tool-annotated docs out of 2,133 docs in cohort (98.1% coverage); this is an annotation slice, not whole-corpus prevalence.

Test: >= 20%. Refreshed 7/17/2026, 4:02:16 AM.

Docs hit554Annotated docs2,092Match rows2,839Avg confidence75.8%
#05 Live 98.1% scope

The chart type is the argument type.

Pick a chart and you pick the comparison the audience will naturally make.

docs with chart-selection evidence48.6%Annotated slice
2,092 tool-annotated docs / 98.1% of 2,133 docs in cohort 3 examples

Name the comparison first, then pick the chart: magnitude, trend, composition, bridge or distribution.

Proof, use and examples
Human lesson

A wrong chart does not just look bad; it asks the reader to compare the wrong thing.

Move to try

Name the comparison first, then pick the chart: magnitude, trend, composition, bridge or distribution.

Evidence contract tool_doc_share

Share of consulting documents with any tool annotation that also carry Chart Selection Guide evidence.

2,092 tool-annotated docs out of 2,133 docs in cohort (98.1% coverage); this is an annotation slice, not whole-corpus prevalence.

Test: >= 12%. Refreshed 7/17/2026, 4:02:16 AM.

Docs hit1,017Annotated docs2,092Match rows4,550Avg confidence37%
#07 Live 95.5% scope

The default consulting story is controlled descent.

The recurring shape is context, complication, approach, proof, then a handoff into action.

arc-annotated docs using the arc52.6%Annotated slice
2,038 arc-annotated docs / 95.5% of 2,133 docs in cohort 3 examples

Audit your deck against five beats: context, complication, approach, evidence, action.

Proof, use and examples
Human lesson

This makes the canon teachable: the structure is observed, not invented in a workshop.

Move to try

Audit your deck against five beats: context, complication, approach, evidence, action.

Evidence contract arc_doc_share

Share of consulting documents with any arc annotation assigned The Consultant's Gambit.

2,038 arc-annotated docs out of 2,133 docs in cohort (95.5% coverage); this is an annotation slice, not whole-corpus prevalence.

Test: >= 30%. Refreshed 7/17/2026, 4:02:16 AM.

Docs hit1,071Annotated docs2,038Arc assignments1,071Avg confidence31.7%
#08 Live 92.8% scope

A pattern beats an anecdote.

One example can be dismissed; three examples become a management problem.

loop-annotated docs using the loop21.1%Annotated slice
1,979 loop-annotated docs / 92.8% of 2,133 docs in cohort 3 examples

Turn three isolated examples into one pattern slide before writing the recommendation.

Proof, use and examples
Human lesson

One example is anecdote; a pattern is permission to act.

Move to try

Turn three isolated examples into one pattern slide before writing the recommendation.

Evidence contract loop_doc_share

Share of consulting documents with any loop annotation that also use Pattern Hunter.

1,979 loop-annotated docs out of 2,133 docs in cohort (92.8% coverage); this is an annotation slice, not whole-corpus prevalence.

Test: >= 20%. Refreshed 7/17/2026, 4:02:16 AM.

Docs hit418Annotated docs1,979Loop ranges643Avg confidence61.8%
#23 Live 92.8% scope

MECE only matters when it changes the decision.

Decomposition becomes canon when each bucket changes what the audience should do.

loop-annotated docs using the loop16.8%Annotated slice
1,979 loop-annotated docs / 92.8% of 2,133 docs in cohort 3 examples

Use MECE only when each bucket changes the decision, not when it merely tidies the slide.

Proof, use and examples
Human lesson

It proves that decomposition is useful when it carries the story forward.

Move to try

Use MECE only when each bucket changes the decision, not when it merely tidies the slide.

Evidence contract loop_doc_share

Share of consulting documents with any loop annotation that also use MECE Breakdown.

1,979 loop-annotated docs out of 2,133 docs in cohort (92.8% coverage); this is an annotation slice, not whole-corpus prevalence.

Test: >= 10%. Refreshed 7/17/2026, 4:02:16 AM.

Docs hit333Annotated docs1,979Loop ranges356Avg confidence44.5%
#24 Live 92.8% scope

Zoom-in makes belief.

A big claim becomes credible when the reader can inspect one sharp specimen.

loop-annotated docs using the loop16.3%Annotated slice
1,979 loop-annotated docs / 92.8% of 2,133 docs in cohort 3 examples

After a broad claim, pick one example and make it carry the whole pattern.

Proof, use and examples
Human lesson

A big thesis becomes credible when the reader can inspect one sharp specimen.

Move to try

After a broad claim, pick one example and make it carry the whole pattern.

Evidence contract loop_doc_share

Share of consulting documents with any loop annotation that also use Zoom In.

1,979 loop-annotated docs out of 2,133 docs in cohort (92.8% coverage); this is an annotation slice, not whole-corpus prevalence.

Test: >= 14%. Refreshed 7/17/2026, 4:02:16 AM.

Docs hit323Annotated docs1,979Loop ranges437Avg confidence56%
#25 Live 88.9% scope

Before/after makes change feel real.

The future state lands when the audience can feel what disappears from the old state.

loop-annotated docs using the loop7.9%Annotated slice
9,144 loop-annotated docs / 88.9% of 10,285 docs in cohort 3 examples

Show current pain and future relief as a pair before adding implementation detail.

Proof, use and examples
Human lesson

Transformation stories need contrast; otherwise the future state is just a feature list.

Move to try

Show current pain and future relief as a pair before adding implementation detail.

Evidence contract loop_doc_share

Share of loop-annotated documents with Before/After.

9,144 loop-annotated docs out of 10,285 docs in cohort (88.9% coverage); this is an annotation slice, not whole-corpus prevalence.

Test: >= 6%. Refreshed 7/17/2026, 4:02:16 AM.

Docs hit723Annotated docs9,144Loop ranges774Avg confidence18.2%

Data stories

Exhibits derived from the canon, not a second canon.

These chapters exist to make the evidence memorable. They point back to a thesis, expose scope, and stay honest when a dataset is only an annotated slice.

1. Every reviewed deck, plotted

Each dot is one consulting deck. Its x is how strongly it opens; its y is how well it closes. If decks were balanced, the cloud would hug the diagonal. It does not: the pattern crowds below it.

97.2% scope
Closing scoreOpening scorebalanced arc

Takeaway: reviewed consulting decks open at 57 and close at 44.2 on average. 60.1% finish at least 8 points weaker than they open, so the craft gap is not "make better covers"; it is turn agreement into decision.

12.8 pts opening-closing gap 2,073 reviewed decks canonical thesis

2. The chart vocabulary

When the corpus puts a chart on a slide, what kind? The answer is not decorative variety. It is a narrow grammar of comparison.

84.2% scope

Takeaway: 77.1% of detected chart components are bars or lines. Pies and donuts are 11.7%. The bias is toward chart types where the human eye can compare magnitude or movement directly.

8,655 docs with charts 122,056 chart components canonical thesis

3. The title is not copy. It is the argument.

The writing system says the same thing four ways: the headline chooses the claim, concrete language exposes thinking, so-what ladders convert facts, and callouts carry the reader through dense proof.

97.3% avg scope

Takeaway: writing is where decks either reduce cognitive work or hide weak thinking. The action-title signal is strong, but its scope is still an annotated slice, so it should be taught as a high-conviction lead until enrichment catches up.

4. Contrarian decks prosecute. They do not merely complain.

The best activist stories do not just say management is wrong. They build pressure through a named culprit, a contradiction, third-party proof, quantified damage and an explicit ask.

100% avg scope

Takeaway: the contrarian canon should teach a prosecution sequence, not a tone of voice: name the failing, show the contradiction, benchmark the gap, make the economic consequence visible, then ask for something specific.

5. Narrative is a set of moves, not a mood.

The corpus keeps showing repeatable moves: analysis-first proof, controlled descent, pattern hunting, MECE framing, zoom-in proof and before/after bridges. That is closer to a playbook than a vibe.

93.8% avg scope

Takeaway: Learn should not feel disconnected from Insights. Each Learn concept should point back to the move-map: this is the principle, here is where the corpus does it, here is where the evidence is still thin.

6. Canon is earned by scope

Some theses are already broad corpus candidates, while others are promising annotation leads. The interface should make that distinction instantly visible.