---
name: "convention-comms-frameworks"
type: convention
description: 'The format-agnostic catalogue of communication frameworks — what each one is for, where it came from, its steps in plain English, what it suits, where it falls short and how it overlaps with its neighbours. Holds AAA (Altitude, Audience, Action), the pyramid principle, SCQA, SCR, BLUF, SBAR, What / So What / Now What and Options → Trade-offs → Recommendation, plus three components used inside any of them: the decision line, the most obvious objection and the rule of three. Carries the purpose-to-framework selection table. Read by skill-message-outline before any outline is written, and by any reviewer checking a finished draft against an approved message brief, so the shape of an argument is chosen from a named definition rather than improvised per piece. Names no format — the slide, email and document wrappers reach it only through the narrative skill. Surfaced by APP-2331; grown as frameworks are reused.'
license: CC BY-NC-SA 4.0
---

# Communication frameworks

The definitions `skill-message-outline` reads. Each entry describes one way of ordering an argument in enough detail to **build** against and to **check** against — the two jobs the skill does with it. Nothing here mentions slides, emails or documents, and nothing here should: a framework is a shape for an argument, and the format is a later, thinner decision.

The problem this exists to fix is cheap to state and expensive to live with. Written output arrives without a spine, the Owner sends it back, and the next draft argues a different case because nobody wrote the case down first. The fix is to settle the argument before the prose, and to keep the argument's shape independent of the thing it is eventually poured into (APP-2331).

The split follows `core-operating-model`'s classify-by-nature test. A framework 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 one, applies it and checks a draft against it sits at Layer 1 as `skill-message-outline`. This is the same relationship as `convention-user-modelling` → `skill-user-modelling` and `convention-deck-narrative` → `skill-deck-narrative`, and it earns the same benefit: a framework can be added, sharpened or retired without touching the behaviour, and the behaviour stays one loop rather than growing a branch per framework.

## How to read an entry

Every framework carries the same six fields, in this order:

* **Use when** — the purpose it serves. The selection table below is the index of these.
* **Origin** — where it comes from, or an explicit statement that no origin was established. An entry never attributes a framework it cannot place.
* **The steps** — the order, in plain English, with what each step is *for*.
* **What it suits** — the readers and occasions it is strongest with.
* **Where it falls short** — how it actually breaks, and what it has no rule for. Every entry carries this field, because a catalogue without failure modes is only a list, and a list tempts a writer to pick the first plausible row.
* **Overlap** — what it shares with its neighbours here, and the one thing that distinguishes it. Most of these frameworks agree on answering first and supporting with two or three points; the differences are narrower than the names suggest, and knowing where they are is what makes a selection defensible.

## Altitude, and why it is not a field

Three frameworks below are explicitly about the *level* a piece is pitched at, and every other one assumes it silently. Altitude is the height at which the reader can act: a board reads the consequence, a delivery lead reads the sequence, an engineer reads the mechanism. The usual failure is to write at the altitude the work was done at rather than the one the reader decides at, which produces a piece that is accurate, complete and unusable.

Altitude is handled inside the AAA entry rather than being hoisted into a field of its own, because for every other framework it is a property of the **brief** — the reader's section of the message brief sets it before a framework is chosen. Where no brief exists, altitude is the first thing to settle and the most commonly skipped.

## Selection table — purpose to framework

The canonical index. `skill-message-outline` Pass 1 reads this table; it is held here rather than in the skill so that adding a framework is a one-file change, and so the selection rationale sits beside the definition it selects.

| Purpose | Framework |
| -- | -- |
| Any piece at all, as the default spine | **AAA** |
| Strategy, analysis, a recommendation that must be reasoned to | **Pyramid principle**, introduced by **SCQA** |
| The same, compressed to a page or less | **SCR** |
| A busy reader who may stop after the first paragraph | **BLUF** |
| An urgent escalation, a handover, a thing going wrong now | **SBAR** |
| Findings, a retro, an audit read-out, a post-incident note | **What / So What / Now What** |
| A choice that someone else has to make | **Options → Trade-offs → Recommendation** |
| Anything asking for a decision, inside whichever structure | **The decision line** |
| Anything a reader will argue with | **The most obvious objection** |
| Anything that must be remembered rather than filed | **The rule of three** |

Where a piece genuinely serves two purposes, it is two pieces, or one piece with a stated primary purpose and the secondary demoted to an annexe. A framework applied to a purpose it does not serve is the mismatch Pass 3 of `skill-message-outline` exists to flag.

---

## 1. AAA — Altitude, Audience, Action

**Use when:** the default spine for anything written to a person who has to do something with it. The other frameworks answer *how to arrange the argument*; this one answers *whether the piece is aimed correctly at all*, so it is applied first and composes with whichever structure is chosen beneath it.

**Origin:** **not established.** No published source for AAA as Altitude, Audience, Action was found in the 2026-10-13 research pass, and a different AAA — Awareness, Advice, Action — exists in adjacent advice and marketing writing. The definition below is therefore **ours, stated here**, and should be cited to this convention rather than attributed to anyone. If a prior source is later established, record it here and keep the definition stable; a found origin changes the citation, not the practice.

**The steps:**

1. **Altitude** — decide the level the reader can act at, and write everything at it. Consequence for a decider, sequence for a delivery owner, mechanism for a practitioner. Detail below the reader's altitude goes into an annexe or goes away; detail above it leaves the reader unable to act.
2. **Audience** — state what this particular reader is measured on, what they already know (so it is not re-explained), how they prefer to receive information, and what they usually push back on. One named reader, not a category.
3. **Action** — state what they must do, in a form someone could act on without a follow-up. "Thoughts?" is not an action; "approve the 30-day window by Friday, or it ships at 21" is.

**What it suits:** everything, as a pre-flight. It is particularly load-bearing for a reader whose altitude differs sharply from the writer's — an Owner reading a practitioner's work, a client reading a delivery team's.

**Where it falls short:** three known gaps, all of which are why the components below exist.

* **No rule for bad news.** AAA will happily carry a cheerful piece that never states the problem. Bad news needs stating plainly with an owner and a plan, and nothing in AAA forces it.
* **No defined decision line.** Its Action step names what to do but not *who decides*, *by when*, or *what happens if nobody answers* — which is the difference between an action a reader recognises and one they execute. Use the decision line entry.
* **Weak asynchronously.** It assumes a reader who can ask a follow-up question. A piece read once, out of hours, with no reply path, needs to be self-sufficient, and AAA has no test for that. The test to add: could the reader act correctly having read it once and asked nothing.

**Overlap:** its answer-first instinct is BLUF's; its Action step is the decision line with less in it; its Audience step is the message brief's reader section, compressed. **Altitude is the one field none of the other frameworks carry**, and it is the reason AAA sits above the others rather than beside them.

## 2. The pyramid principle

**Use when:** strategy, analysis, or any recommendation the reader will want to see reasoned rather than asserted.

**Origin:** Barbara Minto, developed at McKinsey and set out in *The Minto Pyramid Principle*. It is the oldest and most widely taught of the frameworks here, and several of the others are compressions of it.

**The steps:**

1. **State the answer first** — the single governing thought, as one sentence that a reasonable person could disagree with.
2. **Group the supporting arguments beneath it**, two to four of them, each one a claim in its own right rather than a topic.
3. **Make the group mutually exclusive and collectively exhaustive** — no two supports saying the same thing, nothing material left outside them.
4. **Put the evidence under each support**, not batched at the end.
5. **Introduce the whole with SCQA** (entry 3), so the reader arrives at the answer already holding the question it answers.

**What it suits:** a reader who has to be able to interrogate the argument — a board paper, a strategy note, a recommendation that will be tested. It is the strongest framework here for pieces that will be read more than once.

**Where it falls short:** it is heavy. Building a genuinely MECE group is slow, and under time pressure the grouping degrades into a list of whatever was to hand, which looks like a pyramid and reasons like a pile. It also assumes a reader willing to follow a structure downward; a reader who stops after two sentences gets the answer but none of the support, which is fine, and a reader who skims the middle gets supports without evidence, which is worse. For short pieces, use SCR.

**Overlap:** SCQA is its introduction, not an alternative to it. SCR is it, compressed. BLUF is its first step alone. The rule of three is the usual shape of its supporting group.

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

**Use when:** introducing a piece whose reader needs to arrive at the recommendation already holding the problem. The standard opening for strategy and analysis.

**Origin:** Barbara Minto, as the introduction device of the pyramid principle.

**The steps:**

1. **Situation** — shared, uncontroversial context. If the reader would push back on it, it is already Complication.
2. **Complication** — what changed, or the tension that makes the situation no longer workable.
3. **Question** — the one question the piece answers. Often implied rather than written, but it must exist and must be singular.
4. **Answer** — the recommendation, which then opens out into the supporting group.

**What it suits:** openings. It is a few sentences to a few paragraphs, and it does the work of making a recommendation land as analysis rather than assertion.

**Where it falls short:** the Question is the step that disappears. Without it the piece reads Situation → Answer, which is assertion with a preamble. The Situation is the step that gets abused: a writer who smuggles the argument into the context has removed the reader's chance to agree with the setup before being asked to agree with the conclusion. And SCQA is an **introduction** — treating it as the whole structure leaves a piece with a strong opening and an unsorted body.

**Overlap:** it is the front of the pyramid principle. SCR is SCQA with the Question left implicit and the Answer renamed.

## 4. SCR — Situation, Complication, Resolution

**Use when:** the same ground as SCQA and the pyramid principle, at a length where a separate question step costs more than it earns. A page or less.

**Origin:** the compressed consulting form of Minto's shape. No separate published origin was established; it is best treated as a working variant of the pyramid introduction rather than a rival framework.

**The steps:**

1. **Situation** — where things stand, briefly and uncontroversially.
2. **Complication** — what changed and why it cannot stand.
3. **Resolution** — what is being done or recommended, with the action inside it.

**What it suits:** short updates, a note that must fit in a message, the body of an escalation. It is the most portable of the structures here and the easiest to hold in the head while writing.

**Where it falls short:** it has no room for alternatives, so it reads as a decision already taken. Where the reader expects to weigh options, SCR will feel like being told. And because the question is implicit, two readers can finish the piece believing it answered different questions — which is exactly what the Q in SCQA prevents, and why SCR gets riskier as the piece gets longer.

**Overlap:** SCQA with one step folded away. Where a reader might disagree on which question is being answered, use SCQA instead.

## 5. BLUF — bottom line up front

**Use when:** the reader is time-pressured and may stop reading at any point, and the cost of them stopping early with the wrong impression is high.

**Origin:** US military staff writing, where leading with the bottom line is the standard form for a memorandum. It is a writing convention rather than a reasoning method, which is the thing most often misunderstood about it.

**The steps:**

1. **First sentence is the conclusion, or the decision needed.** Not the topic, not the context, not what the piece is about.
2. **Then the supporting points, most important first**, each one able to be the last thing read.
3. **Then detail**, on the understanding it may not be reached.

**What it suits:** updates, escalations, anything going to a reader who triages their reading. It is the cheapest framework here to apply and the one with the best ratio of effort to effect.

**Where it falls short:** it is an ordering rule, not an argument. BLUF will faithfully front-load a conclusion that was never reasoned to, and it gives no help at all in working out what the bottom line *is*. Its characteristic failure is a first line that is a topic ("Q3 update") rather than a conclusion ("Q3 revenue is on track; hiring is the risk to Q4") — the same failure as a label-shaped heading, arriving from the ordering side. It also suits bad news badly on its own: front-loading a problem with no owner and no plan beside it reads as an alarm rather than a report.

**Overlap:** it is step 1 of the pyramid principle used alone. Where the reader will interrogate the argument, BLUF is not enough and the pyramid is the framework.

## 6. SBAR — Situation, Background, Assessment, Recommendation

**Use when:** something is going wrong now, or responsibility is changing hands. Escalations, handovers, incident notes, anything where the reader must be brought up to speed and asked for a call in the same breath.

**Origin:** the US Navy, carried into clinical handover in the early 2000s — notably by Kaiser Permanente — and now standard in healthcare communication. Its design constraint is worth keeping in mind when choosing it: it was built for a verbal handover between people who could not assume shared context and had no time to establish it.

**The steps:**

1. **Situation** — what is happening, right now, in one or two sentences.
2. **Background** — only the context needed to interpret it. Not a history.
3. **Assessment** — what you think it means. This is the step that carries the writer's judgement, and the one that distinguishes SBAR from a status report.
4. **Recommendation** — what you are asking for, specifically, including what you will do if you hear nothing.

**What it suits:** urgency with a named recipient. It is also unusually good at bad news, because Assessment forces an interpretation and Recommendation forces an ask, so a problem cannot be reported without a view and a next step.

**Where it falls short:** it is shaped for a single live incident and gets baggy when stretched over a standing report — Background grows, Assessment hedges, and the piece turns into a chronology. It also presumes the writer is competent to assess; where the writer genuinely does not know what the situation means, an honest SBAR says so in the Assessment rather than manufacturing one, and that is the step most often fudged.

**Overlap:** Situation and Background together do SCQA's Situation and Complication. Assessment is the step none of the other structures name, and is the reason to choose SBAR over SCR for anything urgent.

## 7. What / So What / Now What

**Use when:** findings, a retro, an audit read-out, a post-incident note — anywhere observation has to become action without the reader being asked to do the interpreting.

**Origin:** Terry Borton's 1970 reflective cycle, later formalised for professional practice by John Driscoll. It arrives from reflective learning rather than from business writing, which is why it is better at honesty than at persuasion.

**The steps:**

1. **What** — the observation and the data. What was seen, stated without interpretation.
2. **So what** — what it means. The implication, the root cause, the consequence if nothing changes.
3. **Now what** — the actions, each with an owner and a date.

**What it suits:** anything where the credibility of the recommendation rests on the reader believing the observation. Keeping observation and interpretation visibly separate is what earns that.

**Where it falls short:** the middle step is the one that vanishes. What jumps to Now What and the recommendations arrive unearned, because the reader was never shown why the observation matters. The second failure is a So What that restates the What in different words rather than interpreting it, which is harder to spot and does the same damage. And Now What degrades easily into "areas for further exploration", which is a finding with the action removed.

**Overlap:** the three steps map onto SBAR's Situation, Assessment and Recommendation, minus Background. Choose this for findings read at leisure, SBAR for a situation someone must act on now.

## 8. Options → Trade-offs → Recommendation

**Use when:** a decision belongs to someone else and they need to be able to take it differently from the way you would.

**Origin:** **no single published origin established.** It is standard decision-brief practice rather than a named framework, and the entry below is the form we use. The decision-rights vocabularies — RAPID and DACI — are a different thing: they name *who holds which part of a decision*, and they feed the decision line below rather than supplying a structure.

**The steps:**

1. **Frame the decision** — what is being decided, and by when, before any option appears.
2. **Present the options evenhandedly**, two to four, each stated as its own best case.
3. **Compare the trade-offs directly**, against named criteria that are the same for every option.
4. **Give the recommendation with its reasoning**, including what would change it.
5. **State what is needed from the reader** — the decision line.

**What it suits:** a reader with the authority and the appetite to decide, especially one who will be held to the decision later and needs the alternatives on record.

**Where it falls short:** it is slow, and it is easy to fake. The characteristic failure is a stacked deck — options presented at nominally equal weight when one is obviously straw — and readers notice, at a cost in trust higher than simply recommending would have been. The second failure is revealing the recommendation before the trade-offs, which turns the options section into theatre. Where the decision is genuinely yours to take, do not use this framework; state the decision and the reasoning.

**Overlap:** it is the pyramid principle with the supporting group replaced by alternatives. It always ends in the decision line.

---

## Components

Three things that are not structures. Each one is used *inside* whichever structure was selected, and a piece can carry all three.

## 9. The decision line

**Use when:** anything that asks for a decision. This is not optional and it is not a style choice — a piece that asks for a decision without one has asked for a conversation.

**Origin:** composed here. The decider half comes from decision-rights practice — RAPID (Bain) and DACI both name who holds the decision — and the deadline and the default are the additions that make an ask work when the reader is asynchronous and may never reply.

**The steps:** one line, carrying three things.

1. **Decider** — the named person whose call this is. A team is not a decider.
2. **Deadline** — the date by which the call is needed, and what it is driven by.
3. **Default if no answer** — what happens if nobody responds. This is the part that is almost always missing, and the part that does the most work: it converts silence from a blocker into an outcome, and it tells the decider the cost of not reading.

**What it suits:** every ask, in every framework. It is the single highest-yield line in this convention.

**Where it falls short:** a default has to be one the writer can actually execute and is willing to live with. A bluffed default — "otherwise we proceed as planned" when nothing of the sort will happen — is worse than none, because it is discovered exactly once. Where no honest default exists, say that the work stops and name what stops.

**Overlap:** it is AAA's Action step, completed. It is the closing step of Options → Trade-offs → Recommendation and of SBAR.

## 10. The most obvious objection

**Use when:** anything the reader will argue with — which is most things worth writing.

**Origin:** named and taught by Wes Kao in her writing on executive communication.

**The steps:**

1. **Name the single most obvious objection** the reader will raise. Not the easiest one to answer, the first one they will actually think of.
2. **Raise it yourself, in the piece**, in the reader's own terms rather than a softened version.
3. **Answer it**, or concede it and say what that costs.
4. **Place it immediately after the claim it threatens**, not in an annexe. An objection answered late has already been raised in the reader's head and gone unanswered for several paragraphs.

**What it suits:** recommendations, asks, anything going to a reader whose pushbacks are known — which the message brief should already have named.

**Where it falls short:** answering a weaker objection than the real one is worse than answering none, because it signals either that the real one was not seen or that it was avoided. And raising too many objections turns a recommendation into a defence; one is the discipline, and a piece needing three is usually a piece whose claim is too wide.

**Overlap:** it is the brief's objection field, carried into the piece. Inside the pyramid principle it usually attaches to one support rather than to the governing thought.

## 11. The rule of three

**Use when:** the reader has to remember the piece rather than file it, and when a supporting group needs a size.

**Origin:** classical rhetoric; no single origin, and none should be claimed. The grouping habit is ancient and the supporting cognitive claims are vaguer than they are usually presented, so use it because it reads well and holds, not because of a cited study.

**The steps:**

1. **Reduce the supporting group to three**, by merging or cutting, never by padding.
2. **Make the three parallel** in grammar and in kind, so they read as one set.
3. **Return to them at the close**, in the same order and the same words.

**What it suits:** a governing thought with supports, a close that must be repeatable, a verbal piece.

**Where it falls short:** three is a shape, not a truth. Forcing a fourth material point out to make the shape is a structural lie, and inventing a third to fill it is padding the reader can feel. Where the honest number is two or four, use it and lose the rhythm.

**Overlap:** it is the usual size of the pyramid principle's supporting group, and the usual shape of a BLUF piece's middle.

---

## What this convention does not hold

**The brief.** What the piece is *about* — the point, the reader, the ask, the evidence, the bad news — is the message brief, which is Atlas's (APP-2332). This convention holds shapes; the brief holds the content a shape is applied to. Selecting a framework before the brief exists is choosing a container for contents nobody has named.

**Voice.** A framework says what comes in what order and nothing about register. Voice comes from the domain's own convention — `convention-mrp-voice` for Mr P, a venture's brand voice or `skill-content-voice` elsewhere, `convention-comms-owner` for anything the Owner reads. Pick the structure here, set the voice there; they compose without touching.

**Format.** Nothing here names slides, emails or documents, and nothing here should be amended to. Format lives in the wrappers: `convention-deck` and `skill-deck-narrative` for a deck, `convention-document-opening` for how a client-facing document opens. A framework that acquires a slide count has stopped being format-agnostic and belongs in the deck convention instead.

**Owner-facing shape.** The decision-and-action header, the one-item-at-a-time live review and the sign-off brief are `convention-comms-owner`. Where the reader is the Owner, that convention governs the surface and this one governs the argument inside it.

## The deck structures, and the overlap window

`convention-deck-narrative` holds eleven narrative structures — PAS, SCQA, Before/After/Bridge, Hero's Journey, ABT, Three-Act, Big Idea + Rule of Three, What / So What / Now What, Inverted Pyramid, Options → Trade-offs → Recommendation, and Chapter structure — written against deck sections and slide counts. They are format-agnostic in principle and **this convention is their destination** (APP-2331).

They have not moved, and this pass did not move them. The conversion happens when the deck artefacts are next touched, under APP-2333, which is held precisely because its structures had nowhere to go until this convention existed. Until then:

* **Four entries exist on both sides** — SCQA, What / So What / Now What, Options → Trade-offs → Recommendation, and the rule of three. The deck entries carry slide splits, detection signals and evaluation signals, and remain the definition for deck work; the entries here are the format-agnostic definitions and remain the definition for everything else. Where the two disagree on substance rather than on format, **this convention wins**, and the disagreement is a defect to fix in the conversion rather than a choice to make per run.
* **Seven entries exist only there** — PAS, Before/After/Bridge, Hero's Journey, ABT, Three-Act, Big Idea + Rule of Three, and Chapter structure. A piece that needs one of those reads `convention-deck-narrative` directly in the meantime.
* **The conversion rewrites, it does not copy.** Each arriving structure is re-expressed in the six-field entry shape above, loses its slide splits to `convention-deck`, and keeps its detection and evaluation signals in whichever artefact does the checking.

## Quick checklist before an outline is approved

- [ ] Is there an approved message brief? (No brief, no framework selection — the brief sets the altitude and the reader.)
- [ ] Purpose named, and a framework selected from the table for *that* purpose?
- [ ] AAA applied first — altitude set to the reader's decision level, one named reader, an action someone could execute?
- [ ] The selected framework's steps all present, in order, nothing dropped?
- [ ] The framework's named failure checked for by name?
- [ ] Decision line present wherever a decision is asked for — decider, deadline, default if no answer?
- [ ] Most obvious objection raised in the reader's own terms, and placed beside the claim it threatens?
- [ ] Any bad news stated plainly, with an owner and a plan?
- [ ] Supporting group sized honestly, not padded to three?
- [ ] Framework and reason stated in one line at the top of the outline, so a reviewer can check it rather than guess it?
- [ ] Voice sourced from the domain's own convention, not from here?
- [ ] Nothing in the outline naming a format?
