---
name: "convention-deck-narrative"
type: convention
description: Defines each deck narrative structure — when to use it, section order and job, slide-count split, opening and closing instruction, proof placement, the detection signal that identifies it in an existing deck, the evaluation signals that judge whether it is carried to the end, and its common failure. Holds the purpose-to-structure selection table. Read by skill-deck-narrative before drafting or reviewing any deck outline, so structure is built and judged against a consistent, named definition rather than improvised per run. Baseline is a 12-slide deck, with scaling noted per structure. Sits beside convention-deck, which holds the visual system the outline's archetypes come from. Surfaced by APP-563; grown as structures are reused.
license: CC BY-NC-SA 4.0
---

# Deck narrative structures

The definitions `skill-deck-narrative` reads. Each entry describes one narrative structure in enough detail to *build* against and to *evaluate* against — the two jobs the skill does with it. Owned by Pixel; read by `skill-deck-narrative`, and through it by Reach and Sonar. `convention-deck` holds the visual system an outline's archetypes come from; this convention holds the argument's shape.

The split is deliberate and follows `core-operating-model`'s classify-by-nature test: a structure is a **definition** — it is read, not run — so it sits at Layer 5 and is consumed at Layer 0, while the single behaviour that selects, applies, and checks it sits at Layer 1 as `skill-deck-narrative`. This is the same relationship as `convention-user-modelling` → `skill-user-modelling`, and it earns the same benefit: a structure can be added, sharpened, or retired without touching the behaviour, and the behaviour stays one loop rather than growing a branch per structure (APP-563).

## How to read an entry

Every structure carries the same eight fields, in this order:

* **Use when** — the purpose this structure serves. The selection table below is the index of these.
* **Sections and job** — the section order, what each section is *for*, and its slide split.
* **Opening instruction** — what the deck must do at the top. Most structures are won or lost here.
* **Closing instruction** — what the last slide must be.
* **Proof placement** — where evidence goes, which is a structural choice, not a leftover.
* **Detection signal** — how to recognise this structure in a deck nobody labelled. Used by Pass 3 step 1.
* **Evaluation signals** — how to judge whether it is carried through. Used by Pass 3 step 3.
* **Common failure** — the way this structure actually breaks in practice.

Detection and evaluation are separate fields because Pass 3 asks two separate questions — *which structure is this?* and *is it done properly?* — and a deck can answer the first well and the second badly. Collapsing them into one field is what produces a review that says "this isn't SCQA" when the honest finding is "this is SCQA with no Question slide".

## Slide counts — the baseline and how it scales

**The baseline is a 12-slide deck**, and the per-structure splits below are written against it. The baseline is a **proportional anchor, not a target length**: it exists so the splits express a *ratio* between sections, which is the part that carries meaning. A structure whose Agitate section is 2–3 of 12 is telling you Agitate is roughly a fifth of the deck; at 24 slides it is 4–6, not still 2–3.

Scale by holding the ratio and the section order, never by dropping a section — a dropped section is exactly what each structure's common failure describes. Two exceptions, noted in their entries: **Three-Act** is defined in percentages because its natural length is a talk rather than a deck, and **ABT** has no deck-level count at all because it operates per section.

Where a purpose has a natural default length that differs from 12 (an exec update is usually shorter; a keynote longer), the structure's entry says so. The baseline stays 12 because the alternative — a different anchor per purpose — makes the ratios incomparable across structures for no gain (drain decision, APP-563).

## Selection table — purpose to structure

The canonical index. `skill-deck-narrative` Pass 1 reads this table; it is held here rather than in the skill so that adding a structure is a one-file change, and so the selection rationale sits next to the definition it selects.

| Deck purpose | Structure |
| -- | -- |
| Pitch / persuade | **PAS** or **Hero's Journey** |
| Exec update | **Inverted Pyramid** |
| Strategy / consulting brief | **SCQA** |
| Vision / brand | **Big Idea + Rule of Three**, or **Before/After/Bridge** |
| Findings / audit / retro | **What / So What / Now What**, or **Chapter structure** |
| Decision brief | **Options → Trade-offs → Recommendation** |
| Long-form talk (20+ min) | **Three-Act Structure** |
| Section transitions within any deck | **ABT** |

Where a row offers two, the entries' *Use when* fields separate them; pick on the audience's relationship to the subject, not on preference. Where a deck genuinely serves two purposes, it is two decks or one deck with a stated primary — a structure applied to a purpose it does not serve is the mismatch Pass 3 is built to flag.

---

## 1. PAS — Problem, Agitate, Solve

**Use when:** pitching, persuading, selling a change.

**Sections and job:** Problem (name the pain, 1–2 slides) → Agitate (the cost of inaction, made concrete and specific — numbers, quotes, consequences, 2–3 slides) → Solve (the offer, and proof it works, 5–6 slides) → Ask (1 slide).

**Opening instruction:** open on the problem stated as *the audience's* problem, in their language, not the presenter's. No title or agenda slide first.

**Closing instruction:** end on a single-sentence ask. No recap slide.

**Proof placement:** inside Solve, after the mechanism is explained and before the ask. Proof before mechanism is a claim the audience cannot yet interpret.

**Detection signal:** the deck opens on a pain rather than a company or an agenda, and the offer does not appear until roughly halfway.

**Evaluation signals:** is the problem stated before any credentials or agenda slide; is the agitation section present and using specifics rather than a generic "this is a growing problem"; is the ask a single stated action rather than an implied one.

**Common failure:** skips Agitate and jumps Problem → Solve. It kills urgency, and it is both the weakest and the most common break — the presenter knows the cost of inaction so well that they forget the audience does not.

## 2. SCQA — Situation, Complication, Question, Answer

**Use when:** strategy or consulting brief, exec-facing analysis.

**Sections and job:** Situation (shared, uncontroversial context, 1–2 slides) → Complication (what changed, or the tension, 1–2 slides) → Question (the one question this deck answers, 1 slide, often implicit rather than literal) → Answer (the recommendation, structured as 3–5 supporting points, 6–8 slides).

**Opening instruction:** the Situation must be genuinely uncontroversial. If the audience would push back on slide 1, it is already Complication.

**Closing instruction:** the Answer restated as the recommendation in one sentence, then next steps.

**Proof placement:** distributed across the Answer — one proof point per supporting argument, never batched at the end. Batched proof reads as an appendix the audience has already stopped reading.

**Detection signal:** an opening section the audience would nod along to, followed by a stated turn, and a recommendation broken into a small number of named supports.

**Evaluation signals:** does Situation actually stay neutral or smuggle the argument in early; is there one Question or several competing ones (several dilutes); does the Answer map 1:1 to the Complication.

**Common failure:** no real Question — the deck jumps Situation → Answer and reads as assertion rather than analysis.

## 3. Before / After / Bridge (BAB)

**Use when:** vision, transformation, and roadmap decks.

**Sections and job:** Before (the current state, felt as a limitation, 2–3 slides) → After (the future state, vivid and specific, 2–3 slides) → Bridge (how we get there — the plan, 5–6 slides).

**Opening instruction:** Before must be described from the audience's lived experience, not from a market-stats slide.

**Closing instruction:** end on the first concrete step of the Bridge, not a restatement of After.

**Proof placement:** inside Bridge, as evidence that each step is achievable.

**Detection signal:** two contrasting states presented in sequence before any plan appears.

**Evaluation signals:** is After vivid and specific, or generic aspiration; does the Bridge have actual sequenced steps, or does it amount to "and then we execute".

**Common failure:** After is compelling but Bridge is one hand-wavy slide — vision without a plan, which spends credibility rather than building it.

## 4. Hero's Journey

**Use when:** brand story, customer-as-hero pitch, keynote.

**Sections and job:** Ordinary world (1 slide) → Call to adventure, the problem (1–2) → Struggle (2–3) → Guide appears — the product or company as *mentor*, not hero (1–2) → Transformation (2–3) → New normal (1–2).

**Opening instruction:** open inside the hero's world — the audience's or the customer's — not the presenter's company.

**Closing instruction:** end in the hero's new normal. The presenter and the product stay offstage in the close.

**Proof placement:** in Transformation, as the hero's *own* evidence — a case study, a testimonial, a before-and-after metric.

**Detection signal:** a named protagonist who is not the presenting company, carried across most slides.

**Evaluation signals:** who is the subject of most slides? If it is the company or the product throughout, this is not a Hero's Journey — it is a product deck wearing a story structure.

**Common failure:** the company positions itself as the hero rather than the guide. The single most common misapplication, and the one that makes an otherwise well-built story deck read as vanity.

## 5. ABT — And, But, Therefore

**Use when:** a short narrative spine inside any deck, or single-section transitions. Good for update sections within a larger deck.

**Sections and job:** And (setup — what is true, 1 beat) → But (the complication, 1 beat) → Therefore (the resolution or action, 1 beat). Applied per section, not to the whole deck.

**Opening instruction:** not applicable at deck level — this is a transition discipline applied at section boundaries.

**Closing instruction:** not applicable — see above.

**Proof placement:** not applicable.

**Detection signal:** section-to-section transitions that introduce tension before resolving it, inside a deck whose overall shape is some other structure.

**Evaluation signals:** do section transitions read as "and… and… and…" — flat, no tension — or do they use a genuine "but" to introduce complication before the "therefore".

**Common failure:** used as the whole-deck structure when it is a sentence-level and section-transition tool. Flag for reclassification if it is applied at deck level; the fix is to pick a real deck structure from the table and keep ABT for the transitions inside it.

## 6. Three-Act Structure

**Use when:** a longer keynote or talk (20+ minutes). Not a standard business deck.

**Sections and job:** Act 1 — Setup (establish the world and the stakes, ~25%) → Act 2 — Confrontation (complications escalate, ~50%) → Act 3 — Resolution (~25%). Stated in percentages rather than slides: at this length the unit is time, not slide count.

**Opening instruction:** Act 1 ends on an inciting incident or question, not a summary.

**Closing instruction:** Act 3 resolves the Act 1 question explicitly. The callback is required, not optional.

**Proof placement:** distributed through Act 2, escalating in stakes.

**Detection signal:** a quarter-half-quarter shape with an unresolved question posed early and answered at the end.

**Evaluation signals:** is there an explicit callback from the close to the open; does Act 2 escalate, or does it merely list.

**Common failure:** applied to a 12-slide business deck, where it is structurally the wrong scale — the acts collapse to two or three slides each and the escalation has nowhere to happen. Flag as a mismatch and reselect from the table.

## 7. Big Idea + Rule of Three

**Use when:** vision and brand decks; any deck that needs one memorable takeaway.

**Sections and job:** the Big Idea stated (1 slide, one sentence) → three supporting pillars (1–2 slides each) → synthesis back to the Big Idea (1 slide).

**Opening instruction:** the Big Idea is *stated* as a single sentence within the first two slides — not built up to. Withholding it is a different structure.

**Closing instruction:** restate the Big Idea verbatim, or near-verbatim, as the last line.

**Proof placement:** one proof point per pillar.

**Detection signal:** a single claim asserted early and returned to at the close, with exactly three named supports between.

**Evaluation signals:** can the deck's core message be stated in one sentence? If not, there is no Big Idea — only a topic. Are there exactly three pillars, or a sprawl of five-plus (which dilutes retention)?

**Common failure:** the "big idea" is actually a topic ("our platform strategy") rather than a claim ("we win by owning the workflow, not the tool"). The test is whether a reasonable person could disagree with it; if not, it is a topic.

## 8. What / So What / Now What

**Use when:** findings, retro, and audit read-outs.

**Sections and job:** What (the observation and data, 3–4 slides) → So What (why it matters — the implication, 2–3 slides) → Now What (recommended actions, 3–4 slides).

**Opening instruction:** open on the single most important finding, not on a methodology slide. Method is appendix material unless method *is* the finding.

**Closing instruction:** Now What ends on owned, dated actions — never "areas for further exploration".

**Proof placement:** inside What, as the data itself.

**Detection signal:** an observation-then-implication-then-action progression, usually across findings rather than a single argument.

**Evaluation signals:** does So What actually interpret — state an implication — or merely repeat What in different words; are the Now What items assignable and actionable, or vague.

**Common failure:** the So What section is missing — What jumps straight to Now What, so the recommendations feel unearned. The audience is asked to accept an action whose justification was never stated.

**Note for A1 consulting work:** where the deck reports a Sonar engagement, this structure is the natural carrier for the **Finding → Theme → Opportunity → Initiative** ladder (`convention-audit`): Findings and Theme Summaries are the What, the theme's root cause is the So What, and Opportunities and Initiatives are the Now What. Use the ladder's own card vocabulary in the outline rather than re-labelling it; the structure and the ladder are the same shape at different grain (APP-557).

## 9. Inverted Pyramid

**Use when:** exec updates and status reports — a time-pressured audience who may not reach slide 5.

**Sections and job:** headline conclusion (1 slide) → key supporting points, most important first (3–5 slides) → detail and appendix (as needed, often not presented live). Natural length runs shorter than the 12-slide baseline; hold the ordering rule rather than padding to it.

**Opening instruction:** slide 1 is the conclusion, or the decision needed, stated directly. No context-first build-up.

**Closing instruction:** end on the specific decision or action needed from the audience, not on a summary of what was covered.

**Proof placement:** front-loaded — the most important evidence immediately after the headline, decreasing in priority thereafter.

**Detection signal:** the deck's strongest claim is on slide 1, and each subsequent slide matters less than the one before it.

**Evaluation signals:** could the audience get the point from slide 1 alone; is the supporting evidence ordered by importance or chronologically (chronological means it is not actually an inverted pyramid).

**Common failure:** the headline slide is a topic ("Q3 Update") rather than a conclusion ("Q3 revenue on track; hiring is the risk to Q4"). Note this is the same failure as `convention-deck`'s label-title anti-pattern, arriving from the structural side — an exec deck that opens on a label has lost the only advantage the structure offers.

## 10. Options → Trade-offs → Recommendation

**Use when:** decision briefs.

**Sections and job:** decision framing — what is being decided, and by when (1 slide) → Options, 2–4 presented evenhandedly (1 slide each) → Trade-offs, compared directly (1–2 slides, ideally a comparison table) → Recommendation with reasoning (1–2 slides) → what is needed from the audience (1 slide).

**Opening instruction:** state the decision to be made and its deadline before presenting any option.

**Closing instruction:** end on what is needed from the audience — approval, resources, a choice — not a recap of the options.

**Proof placement:** inside Trade-offs, as the comparison criteria and the evidence per option.

**Detection signal:** parallel options presented at equal weight before any preference is expressed.

**Evaluation signals:** are the options presented evenhandedly before the recommendation, or is the "preferred" option obviously stacked; is there an actual trade-off comparison, or just a list of pros.

**Common failure:** the recommendation is revealed before the trade-offs are shown, which makes the options section feel like theatre. The audience notices, and it costs more trust than simply leading with the recommendation would have.

## 11. Chapter structure (data-driven)

**Use when:** research findings, and market or data-heavy decks with no single narrative arc.

**Sections and job:** 3–5 thematic chapters, each: chapter question (1 slide) → data (2–3 slides) → chapter takeaway (1 slide). Close with a roll-up takeaway across all chapters (1–2 slides). Naturally runs longer than the baseline; scale by adding chapters, never by dropping a chapter's takeaway.

**Opening instruction:** state the overall question the whole deck answers before chapter 1.

**Closing instruction:** the roll-up takeaway is a synthesis, not a list of the chapter takeaways restated.

**Proof placement:** within each chapter, adjacent to that chapter's takeaway.

**Detection signal:** repeating question-data-takeaway units with no through-argument between them.

**Evaluation signals:** does each chapter end with a stated takeaway, or trail into the next chapter's data; does the roll-up actually synthesise, or just recap.

**Common failure:** chapters have data but no stated takeaway — the deck reads as a data dump the audience has to interpret for itself, which is the work they came to have done.

---

## What this convention does not hold

**Voice.** A structure says what comes in what order; it says nothing about register. Voice comes from the domain's own convention — `convention-mrp-voice` for Mr P, the venture's `skill-content-tov` or brand voice elsewhere — and the two compose without touching: pick the structure here, set the voice there. The structure names above were cross-checked against the voice conventions and none collide (APP-563).

**Visual weight.** Which archetype an outline line becomes, and how the deck's rhythm runs, is `convention-deck`. A structure's section split and the deck's weight rhythm are different axes: SCQA says the Answer is 6–8 slides, `convention-deck` says those slides cannot all be the same tier.

**Non-deck long-form.** These structures are format-agnostic in principle, and a later extension to `skill-content-studio`'s long-form choices is plausible — but it is not in scope here, and the entries are written against deck sections and slide counts. If the need is real, capture it to the Pulse queue rather than stretching this convention to cover a format it was not evidenced against (drain decision, APP-562). The capture arrived and the prose-document case now has its own convention: `convention-document-opening` holds how a client-facing document opens, and the boundary between the two is the format each was evidenced against (APP-1939).

## Quick checklist before an outline is approved

- [ ] Purpose named, and a structure selected from the table for *that* purpose?
- [ ] Section order and job match the entry — nothing dropped, nothing reordered?
- [ ] Section ratios hold at this deck's length (baseline 12, scaled proportionally)?
- [ ] Opening instruction followed — especially the "no agenda slide first" ones?
- [ ] Closing instruction followed — the last slide is what the entry says it is?
- [ ] Proof placed where the entry says, not batched at the end?
- [ ] The structure's common failure checked for by name?
- [ ] Voice sourced from the domain's own convention, not from here?