slug: so-what-test view: skill layer: slide agent: architect audience: llm companion: corpus/storymakers/frameworks/slide/so-what-test.md

So What? Test — operational reference

One sentence. Point at any artefact, ask "so what?", loop until the answer is an implication for this audience — that terminal sentence is the slide's title.

Use this when

  • A slide has a chart, number, list or finding.
  • A draft title is a description ("Revenue grew 8%") not a decision ("Reallocate spend toward premium").
  • The deck circulates without its author — implication must live on the page, not in the speaker's head.
  • The audience is being asked to do, decide or believe something.

Use something else when

If… Then…
Process documentation (SOP, runbook, manual). Description IS the artefact. Don't force a so-what.
Regulatory disclosure (10-K, MD&A, prospectus). Description IS the artefact. Editorial overreach risk.
Reference / appendix exhibits. Job is lookup, not argument.
Discovery deck, hypothesis tree, problem-board. Provoke so-whats; don't pre-empt them.
Audience already holds the implication. Trim the slide; the so-what is in their head already.

Decision tree

The recursion is the easy part. The terminating condition is the discipline:

state the artefact as a fact
└── loop:
    ├── ask "so what?"
    ├── write the answer
    └── inspect:
        ├── another fact?                  → loop
        ├── implication for the analyst    → loop (climb one rung)
        ├── implication for a previous
        │   audience                       → re-target; loop
        ├── philosophical regress
        │   ("…because capitalism")        → back up one rung; STOP
        └── implication for THIS audience
            (verb they can take, on their
            calendar, no further so-what
            required to act)               → STOP. that's the title.

Recipe (per slide, ~3 minutes)

  1. State the artefact as a fact. Read the chart / bullet / table aloud; write the sentence.
  2. Ask "so what?" Write the answer.
  3. Inspect. Implication this audience can act on, on their calendar? No → return to step 2. Yes → step 4.
  4. Promote the terminal sentence to the slide title; demote the chain into the body, closest-to-audience at the top.
  5. Audience sanity-check. Name the audience by role. Could they take the action tomorrow with what's on this slide?
  6. Headline pass. Complete sentence with a verb? No unfamiliar jargon?

Anti-patterns to refuse

  • Description-as-title"Revenue grew 8% YoY". Reject.
  • Topic-as-title"Pricing analysis". Label, not implication.
  • Single-pass — asked once, shipped. The test is recursive.
  • Philosophical regress"…because capitalism." Back up one rung.
  • Speaker-notes implication — fails the moment the deck is forwarded without its author.
  • Audience-drift implication — implication for the previous room. Re-run the test for the current audience.

Output checklist

A slide passes the so-what test when:

  • The title is a declarative sentence with a verb.
  • The title states an implication, not a description.
  • The implication is actionable by this audience on this calendar.
  • The chart / table / list on the slide supports the title (the title is not a non-sequitur dropped onto unrelated evidence).
  • The reader can read the title alone and end up holding the slide's contribution to the deck's argument.
  • No part of the so-what lives only in the speaker notes.

Canonical signals (for matchers)

  • Keywords: so what, implication, this means, therefore, we recommend, the takeaway, what this means for, bottom line, key insight.
  • Component kinds: title (the implication itself), callout (the so-what surfaced as a margin note), paragraph (chain of reasoning), list (descending so-whats as bullets).
  • Slide-level shape: an action-title that's a verb-led recommendation
    • a single supporting chart/table/list + a closing callout naming the decision.

References

  • Rasiel, The McKinsey Way, 1999 — canonical written source.
  • Ohno, Toyota Production System, 1988 — Five Whys, the inverted cousin.
  • Minto, The Pyramid Principle, 1973 (rev. 2009) — answer-first doctrine; the so-what is the apex.
  • Pink, To Sell Is Human, 2012 — "information becomes insight only when someone has to do something with it."