# Mi Analyze Swot

> Build a SWOT from fused evidence with quadrant discipline, ranked entries, and the S-O and W-T crossings that make it a decision. Use when the evidence exists and a position must be stated.

- Skill: `productside/mi-analyze-swot` (Agent Skill, multi-file: 4 files)
- Install (CLI): `npx skillmds@latest add productside/mi-analyze-swot`
- Raw SKILL.md: https://api.skillmd.com/api/skills/productside/mi-analyze-swot/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- License: CC-BY-NC-ND-4.0
- Author: Productside (https://skillmd.com/u/productside)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/productside/mi-analyze-swot

---


# SWOT From Evidence

## Purpose

Turn fused evidence into a position that survives a hostile read.

A SWOT is not four lists. It is a claim about where you stand, and the part that makes it a decision is the **crossings**: which strength lets you take which opportunity this quarter, and which weakness gets exposed by which threat. Without those, a SWOT is a workshop artifact that gets photographed and never referenced.

This skill consumes evidence; it does not manufacture it. Where a quadrant has no cited signal behind an entry, the honest entry is **"no evidence found"** — not a plausible sentence. A quadrant filled to look complete is the failure mode that makes SWOT the most-mocked framework in product management, and it is entirely avoidable.

## When to Use It

Use it when the evidence exists and a position must be stated — a board question, a planning cycle, a positioning reset. Use it on a competitor as readily as on yourself.

Do not use it when:

- **You do not have the evidence yet.** Run the sweeps and `mi-fuse-all-source` first. A SWOT built from opinion is a record of the room's mood.
- **The question is industry structure.** That is `mi-analyze-five-forces` — SWOT reads one player's position, forces reads whether the industry is worth being in.
- **The question is where to grow.** That is `mi-analyze-ansoff`.
- **You want a facilitation exercise.** This produces an evidence-bound artifact; it does not run a workshop.

## Input

Require:

- fused evidence, or sweep outputs, with sources and dates
- whose SWOT this is — us, or a named competitor
- the decision it supports

Anything supplied in the invocation, attachments, a fusion brief, or earlier conversation counts as context already given. If the evidence is thin in one quadrant, that is a finding to report, not a gap to fill.

**Example invocation:** `Use $mi-analyze-swot on us, using the fusion brief in this thread. Decision: what we defend and what we concede in next year's plan.`

## Key Concepts

**Quadrant discipline is the whole game** — Strengths and weaknesses are **internal and current**. Opportunities and threats are **external and not controllable**. Almost every bad SWOT fails here, and the two failures are specific and nameable: a competitor's product filed as a weakness (it is a threat — it is theirs, not yours), and an "opportunity" that requires you to act. Making the corrections visible teaches the rule better than stating it.

*Violation signal:* An entry in strengths or weaknesses describes something another company does.

**An opportunity that requires you to act is a strategy** — "Expand into healthcare" is not an opportunity; it is a plan. The opportunity is the external condition that makes it viable: a regulation creating demand, a competitor exiting, a buyer population growing. Filing strategies as opportunities is the most common way a SWOT quietly becomes a wish list, and it is invisible because wish lists read as ambition.

*Violation signal:* An opportunity entry starts with a verb you control.

**Five per quadrant, ranked on a stated basis** — Rank strengths on **defensibility**, weaknesses on **exploitability**, opportunities on **fit to your strengths**, and threats on **likelihood times damage**. Say which basis you used. An unranked quadrant treats a durable moat and a temporary advantage as equivalent, and the ranking is where the judgment lives.

*Violation signal:* Entries appear in the order they were thought of, with no basis named.

**Customer voice weighs heaviest in weaknesses** — What your own users say about you outranks internal opinion about yourself. Teams are systematically wrong about their own weaknesses in a predictable direction: they list what they know they have not built, and omit what customers actually struggle with. A weakness with a customer quote behind it is worth five without.

*Violation signal:* The weaknesses quadrant is a roadmap backlog, with no customer evidence in it.

**The crossings are the so-what** — **S-O:** which strength lets you take which opportunity, and what to do this quarter. **W-T:** which weakness gets exposed by which threat, and what protects it. Optionally S-T (defend) and W-O (invest to qualify). A SWOT without crossings is a list; with them it is a plan with evidence attached.

*Violation signal:* The document ends at the fourth quadrant.

**"No evidence found" is a valid entry** — A quadrant with two well-sourced entries beats one with five, three of which are plausible sentences. Empty space in a SWOT is informative: it says where you are not looking, or where nothing is currently true.

*Violation signal:* All four quadrants contain exactly five entries.

## Guided Context Capture

Follow [Guided Context Capture](../../reference/guided-context-capture.md).

Open with: "This is a SWOT built from evidence you already hold. Four quadrants, max five entries each, every entry sourced and ranked, ending with the S-O and W-T crossings. Forty-five to ninety minutes. Where evidence is missing I will write 'no evidence found' rather than fill the box. Up to three setup questions. Choose **1. Guided**, **2. Context dump**, or **3. Best guess**."

In Guided mode, ask one at a time, hard cap of three:

1. `SWOT Setup Q1/3` — Whose SWOT is this: ours, or a named competitor's?
2. `SWOT Setup Q2/3` — What decision does it support?
3. `SWOT Setup Q3/3` — Is there customer evidence about weaknesses — reviews, win/loss, support themes — that I should weight above internal opinion?

**Context dump is the expected mode**, since this skill normally begins with a fusion brief or sweep outputs. Extract entries, sort them into quadrants, show the corrections you made, and ask only about gaps. In **Best guess** mode, build from available evidence, mark unsupported entries as "no evidence found," and name each assumption. On silence: our own SWOT, planning-cycle framing, customer evidence weighted first. Proceed.

## What It Produces

Complete the [SWOT](template.md):

- four quadrants, max five entries each, every entry with a source, date, and F/I/A label
- ranking within each quadrant on the stated basis
- **quadrant corrections made visible** — what was moved, and why
- S-O and W-T crossings, each with a named move and an owner
- entries marked "no evidence found" where they belong
- assumptions to validate, Final Step block

## Workflow

1. **State web access in one line,** and state that this run consumes evidence rather than collecting it.
2. **Establish whose SWOT this is.** The same evidence produces different quadrants for us and for a competitor.
3. **Sort candidate entries into quadrants, applying discipline as you go.** Internal and current on the left; external and uncontrollable on the right.
4. **Show your corrections.** When you move a competitor's product from weaknesses to threats, or a strategy from opportunities out entirely, say so. The corrections teach the rule.
5. **Attach a source, date, and label to every entry.** Anything unsupported becomes "no evidence found," and the quadrant stays short.
6. **Weight customer voice heaviest in weaknesses.** Reviews, win/loss, support themes outrank internal opinion.
7. **Cap at five per quadrant and rank** on the stated basis: defensibility, exploitability, fit to strengths, likelihood times damage.
8. **Write the S-O crossings:** which strength takes which opportunity, and the move this quarter.
9. **Write the W-T crossings:** which weakness meets which threat, and what protects it.
10. **Name the assumptions** that would most change the picture, and stop.

## Human Decision Gate

Present the crossings first, then the quadrants. Highlight:

- which entries were moved between quadrants, and why
- which quadrants are short because evidence was missing
- the W-T crossing with the highest likelihood-times-damage
- whether the weaknesses quadrant reflects customers or internal opinion

Use an Adaptive Decision Ladder: `Act on the top S-O and W-T crossings`, `Collect evidence for the thin quadrant first`, `Run the same SWOT on the competitor for contrast`, or `Other (specify)`.

## Evidence and Attribution Rules

- Label every entry **Fact**, **Inference** (chain shown), or **Assumption** (basis stated).
- **Do not invent:** competitor capabilities, market conditions, customer sentiment, regulatory changes, or internal metrics. A SWOT is presented to leadership as a considered position, which makes an invented entry uniquely durable — it will be repeated for a year.
- Every entry carries a source and a date, or reads "no evidence found."
- An internal opinion about your own weakness is an Assumption unless customer evidence supports it.
- Threats must be external and outside your control; if you can decide it, it is not a threat.
- Do not average conflicting evidence into a moderate entry.

## Common Failure Modes

- Filing a competitor's product as a weakness rather than a threat
- Filing a strategy as an opportunity
- Filling all four quadrants to five because the template has five rows
- Listing the roadmap backlog as weaknesses, with no customer evidence
- Ranking nothing, so a moat and a temporary edge look equivalent
- Ending at the fourth quadrant with no crossings
- Building the SWOT from the room's opinion rather than from fused evidence
- Writing plausible entries where evidence is absent

## Assets and Examples

- [SWOT template](template.md)
- [Synthetic worked example](examples/worked-example.md)
- [Weak example](examples/weak-example.md)

## Final Step

```text
## Final Step

1. Act on the top S-O and W-T crossings (Recommended)
2. Collect evidence for the thin quadrant before presenting this
3. Schedule a semi-annual refresh so the position is tracked
4. Run the same SWOT on the competitor for contrast

Reply with `1`, `2`, `3`, `4`, `1 and 3`, `Verbose Mode`, or a custom path.
```

## Sources

- [The act layer: SWOT discipline rules](../../reference/frameworks.md)
- [Fusion: what feeds this framework](../../reference/fusion.md)
- [The disciplines](../../reference/disciplines.md)
- [Competitive research compendium and runnable prompts](https://github.com/Productside/Productside-Market-Intelligence-Skills)
- [SCIP Code of Ethics](https://www.scip.org/page/CodeofEthics)