Everyone sees the same category.
Our job is to think what no one has, and hand back a sharper question to build on.
One worry in. One question worth building on, out.
A client worry — "we need to grow," "we're losing our audience" — urgent and unstructured.
A reframed core question, a strategic direction, and the next jobs — every claim sourced.
This is the front end of strategy: problem definition and strategic direction, before anyone writes a positioning or a brief.
Every box runs on one engine — the reason the output is a reframe, not a report.
Six working habits fire inside each step. They're what keep the tool from handing back the safe, average reading a search would give you.
Six boxes carry the worry to the question.
The next slides open each box: what it does, what goes in and what you do, and how AI sharpens it.
Frames where the brand is and where it needs to go, and captures the problem in the client's own words — the oldest instrument in planning, Stephen King's "where are we, where could we be."
- Read the brief, the business ambition, current performance
- Write the problem exactly as the client stated it, before touching it
- Desk research on the brand's own numbers and goal
- Pulls category size, share and decline to ground "where you are" in numbers, not the client's assertion
- Cross-references the brief's claim against market data — is this a brand problem or a category one?
Engine move · Set aside the obvious — the conditioned read gets named here, not accepted.
Reads every player twice — what they do, and the story they tell — then types every white space: genuinely open, a trap, or a story to retell.
- The competitive set: campaigns, product UI, packaging, positioning
- Audit the codes and conventions the category shares
- Name the story each brand tells — rom-com, prophecy, court
- Reads live brand material at scale and cross-references public filings and user numbers to weight each read
- Maps each brand onto the handful of story shapes every plot reduces to — the quest, the rom-com, the rebirth — to name the myth it tells without meaning to
- Flags a white space as a trap when its story already belongs to a rival
Engine move · Go to the edges — the fresh reading comes from outside the category's own conventions.
Finds the human tension the brand can lean into — both sides defensible — and roots the live signal to the deeper pattern it repeats.
- Social listening, cultural signals, audience research
- Spot the tension both sides of an audience would sign
- Reject one-sided tensions — those are judgments in disguise
- Mines Reddit, reviews, forums and trend knowledge graphs for what people say right now
- Roots each signal to a fundamental — sociology, anthropology, philosophy — so a trend never floats free
- Runs the both-sides gate and kills accusations dressed as tensions
Engine move · Steal across domains — a 2025 signal explained by a 1925 idea.
Excavates what the brand already owns against this challenge — the name, the heritage, the assets it's ignoring. The permission is usually already there.
- Brand assets: name, founding story, packaging, prior work
- Look for what the brand encodes but its marketing ignores
- Check whether a competitor has already started the move
- Cross-references name etymology, heritage and packaging against the opportunity
- Surfaces latent meaning the current marketing walks past — excavate, never import
Engine move · Excavate, don't import — the strategy is dug out of the brand, not pasted on.
Slams every candidate reading against real data — market, competitor, culture — and kills the ones that don't survive, with reasons on the record.
- The candidate readings from the four boxes above
- Stress-test each against the evidence, not against taste
- Keep the survivors; record why the rest died
- Tests each reading against the full evidence stack at once
- Tags every claim sourced, inferred or hypothesized
- Rejects readings the data contradicts and readings the audience would never sign
Engine move · Back every claim — nothing reaches the center untested.
Restates the problem as a question worth building on — the reframe — and keeps a rival question live rather than collapsing to one answer.
- Everything the five boxes shipped
- Converge the four lenses through the pressure ring
- Land a question a creative team can run with
- Synthesizes the lenses into one central question
- Holds a rival reading on the table — the recommendation stays the client's to make
Engine move · Invert it, and be interesting — the reframe lands in two seconds or it isn't one.
A worked case: Kindling
A fictional challenger dating app, run through every box on real category data — from the worry it walked in with to the question it leaves on.
Run the Kindling case →