hou · the method
A niche is an intersection, and an intersection can be read.
Nichigai holds one strategic model: the market you serve, the problem you own, the skills you bring, and who you are. Everything the workspace shows: the score, the economics, the adjacent moves, the content pillars: is derived from that one object. Nothing keeps its own copy, so nothing can quietly disagree with anything else.
shiki · the equation
The four questions.
Asked in this order because each answer makes the next one answerable. Self first: it is the only one nobody else could answer for you.
Who should I serve?
Can you actually reach them, and can they actually pay? Name them the way they would name themselves.
What expensive problem should I own?
Is it painful, costly and frequent enough that they already want it fixed, and would anyone be able to tell once it was?
What can I bring to solving it?
Any single skill is replaceable. What matters is whether the combination covers the whole outcome, and how rare that combination is.
Where does who I am create an advantage?
Not what you can do, since anyone can learn that. What you have lived, what you came through, and what you cannot stop caring about.
han · judgement
What makes the score move.
Four rules stop the model flattening out. They are the reason a sharpened position moves the number instead of nudging it.
Contrast, not averages
Each dimension score passes through a logistic curve. An average of averages always lands near fifty and tells you nothing; a curve makes the middle of the range mean something.
Weak-link penalties
A dimension is capped by its worst meaningful factor, and opportunity is dragged down by the weakest of market, problem and skill. A position is only as sellable as the arm carrying the least weight.
A geometric blend
Niche ≈ opportunity^0.55 × advantage^0.45, with a further penalty when either side falls below 28. AI cybersecurity for global banks: enormous opportunity, no personal advantage: stays in the low twenties, where it belongs.
Coherence costs
Serving three unrelated markets is harder to sell than serving one, so the score pays for the breadth it bought. Everyone scores zero coherence, because a position that excludes nobody is not a position.
Online presence for everyone.
- Market
- 22
- Problem
- 30
- Skills
- 44
- Self
- 28
A position that excludes nobody is not a position. Nothing here tells a buyer why you, and nothing tells you what to say next.
Recall systems for private dental practices losing $40k a year to no-shows.
- Market
- 88
- Problem
- 94
- Skills
- 90
- Self
- 86
A named buyer, a costed problem, and a stack that covers the whole outcome. The first sale is a conversation, not a campaign.
atai · what it is worth
What you get back.
A score on its own is a grade. What you can act on is the economics underneath it, and every one of these is a range with a confidence attached, never a decimal pretending to be a fact.
Engagement value
What one piece of this work is plausibly worth, as a range.
Annual problem cost
What the problem costs the buyer over a year: the number your price sits against.
Time to first sale
How long the market's sales cycle and accessibility suggest it takes to land one.
Confidence
How much of the above the model is willing to stand behind, stated plainly.
michi · the way you are going
You are read twice.
When an opening is matched against you, your record is only half of it. The other half is direction: the problems you keep choosing, what you are actively learning, what you have made on your own time. It is read separately rather than averaged into a single number.
Capability
What you can do today: coverage of what the problem actually needs, direct experience, and whether you hold the things that genuinely cannot be taught on the job.
Trajectory
Whether this is where you are already heading. Nothing is inferred about you: every signal is something you put in your own profile, and you decide what is visible.
This is why a position is worth writing down carefully. It is what lets somebody find you for work you have not formally done yet, which is the only way the people whose interests run ahead of their credentials ever get found.
gen · the limit
What this is not.
The structural numbers come from a seeded dataset of markets, problems and skills: saturation, pricing power, sales cycle, capability coverage. Your four answers are read against it. That reading is a judgement, and a judgement can be wrong. The interface never pretends otherwise: everything is a range, everything carries a confidence, and the weakest arm is always named so you know where a second pass is worth most.