| name | so-what-generator |
| description | Turns an observation, data point, chart, or finding into a real insight — the implication and the action — instead of just restating what happened. Use this skill whenever someone shares a fact, metric, chart, or table and needs to know what it means, asks "so what", "what's the takeaway", "why does this matter", "what should we do about this", or is presenting findings that read as observations rather than conclusions. Trigger it whenever a user has data but hasn't extracted the decision from it, even if they don't use the phrase "so what". This is the instinct that turns a background slide into an actual recommendation.
|
So-What Generator
A senior decision-maker does not want to know that sales fell 12% in Q3. They want to know what that
means and what to do about it. The gap between those two is the difference between an observation and
an insight, and closing it is the single most valuable habit in strategy work.
An observation restates the data: "churn rose to 8% last quarter." An insight states the
implication and the action: "churn rose to 8% because the mid-tier cohort is leaving for a cheaper
rival, so we should defend that segment on value before anyone reaches for a price cut." Your job is
to help the user cross that gap, honestly.
The method
Take the finding and ask "so what?" repeatedly, each answer becoming the thing you interrogate next,
until you land on an action or a decision. Once is rarely enough; the first "so what" usually
produces another observation. Chain it at least twice.
Worked example:
- Finding: "Our biggest customer segment has the lowest margin."
- So what? "We're growing fastest in our least profitable segment."
- So what? "Our growth is quietly diluting overall margin even as revenue rises."
- So what (the action): "We should either reprice that segment or shift acquisition spend toward the
higher-margin one, before the board mistakes revenue growth for health."
That last line is the insight. It names an implication and points to a decision. Stop when you reach
something a leader could act on.
Two disciplines keep this honest:
- Separate observation from insight explicitly, so the user sees the move you made and can
learn it. Show the observation, then the chain, then the insight.
- Be honest when the data can't bear a strong conclusion. If a finding is suggestive but thin,
say what it hints at, then name what you'd need to confirm it. A manufactured dramatic insight
the data can't support is worse than an honest "here's the hypothesis, here's what would confirm
it."
Output format
- Observation: what the data literally says.
- So-what chain: two or three steps, each pushing toward an implication.
- Insight: one sentence a senior could repeat in a meeting, naming the implication and the
recommended action.
- Confidence note (when relevant): if the data is thin, what would strengthen the conclusion.
If the insight is big enough to warrant a decision document, hand off to
one-page-recommendation-writer to turn it into the answer-first page.
How to run it
Default to delivering the insight, but always show the chain so the reasoning is transparent and
challengeable. Switch to coaching when the user signals they want the practice: ask them for their
own "so what" first, then push it further when they stop at the first restatement — that
stopping-too-early is the exact habit to break.
Where this breaks
The "so what" chain can run too far, manufacturing grand strategic conclusions from a single data
point. Every step should stay tethered to what the data actually supports; when you feel yourself
reaching, stop and flag it. Also, some findings genuinely are just context, not insight — not every
number has a decision hiding inside it, and pretending one does wastes the reader's trust.
Style
Land the insight in one crisp sentence. Don't bury it under caveats — state it, then caveat once if
needed. Define terms plainly, no em dashes, short paragraphs.
