S4: Multiply your revenue streams

English edition. S1–S6 are available in English; S7–S9 are published in Japanese.

From one operator in a field, to someone who moves the field.

On this page

• What you get from this skill
• Why it matters
• Copy-paste path
• The core: B2C, B2B, B2M
• Step 1: choose your center business and set revenue stages
• Step 2: map every role around it
• Step 3: expand the map into new business
• Stage 2: replicate the unit
• Cutting what you added
• The compounding cycle
• Where S4 ends
• Practice: design a high-margin new business

What you get from this skill

A center business at the core, with several revenue models around it — reaching not only end customers but other businesses and the industry as a whole. Your expertise starts working as a reference point for your field, and an ecosystem begins to run on its own.

Why it matters

From “one service” to “one market”

Most people think “my business = serving end customers.” But once you understand your market deeply (S2) and your voice reaches people (S3), something else becomes visible: the same expertise opens several doors to revenue. This isn’t about side hustles. It’s using one deep understanding of a market in different revenue shapes.

Your mental model of revenue sets your ceiling

Whether you can scale is less about ability or effort than about how you picture your own revenue.

If you think "I'm in the restaurant business":

Revenue = average ticket × seats × turns × days open

→ no amount of effort gets you outside this equation
→ your market is a few miles around the door
→ margins are capped by food cost

The equation is correct. The problem is believing it’s the only one.

You can change the business model and keep the core. What a restaurant owner really has isn’t the building — it’s deep, daily, first-hand knowledge of the market: sourcing, cost control, how staff grow, the early signs that regulars are drifting away. That knowledge lives in the restaurant but isn’t trapped there.

Model	Reach	Typical gross margin	What caps revenue
Product / venue (serve customers on site)	a few miles	30–60%	seats × turns
Service (help others in the industry)	national	50–80%	your hours plus a few contractors
License (standards, certification, data)	the whole industry	70%+	effectively nothing but adoption

Same core, different model — and reach and margin jump together.

[Inside the frame]
"We're small; there's no more room to grow."
"Teaching competitors would only hurt us."
"Setting industry rules is for big companies."

          ↓ step outside

[Outside the frame]
"The seats are a machine that generates knowledge, not the revenue itself."
"Teaching others turns my judgment into the standard."
"Rules should be written by whoever knows the field best."

One caution: stepping outside the frame doesn’t mean neglecting the business you have. Keep your center business whatever else you add. It’s your source of first-hand knowledge; abandon it and everything you say gets stale.

---

Copy-paste path: the shortest route through S4 (first pass)

Paste these into your AI agent one at a time, in order. Replace everything in 【brackets】 first. S4 alternates between widening and narrowing — score and judge every idea yourself before moving on. The reasoning is in the steps below.

① Pick one center business (Step 1)

My candidates for a center business are 【A】, 【B】, and 【C】.
Build a scoring table, 5 points each, 25 total, on: how often it repeats / how much first-hand data it produces / closeness to customers / profit potential / how well the knowledge transfers to other businesses.
Pull the evidence for each score from actionlog/ and ontology/. I make the final call.

② Design the revenue stages (Step 1)

My center business is 【business】; my revenue type is 【product / brokerage / service / license】.
For Stage 1 (up to ~$250k) and Stage 2 (~$250k–$2.5M), fill in these seven items:
who, what, at what price / the revenue equation (Stage 2 must include the number of parallel units) / gross margin and why / how my own time is split / the AI, contractor, and staff setup / the one bottleneck / the trigger to move to the next stage (quantified)
Always express money as multiplication — no vague "low six figures." Finally, judge where I am now.
Save to research/revenue-stages.md.

③ Map the roles, then let an outsider fill the gaps (Step 2)

My center business is 【business】, and every day I run 【what the work looks like】.
List every industry and role connected to it, and save it to research/role-map.md.

Then, as an outside analyst who knows nothing about my business, list 30 industries or roles that are NOT on that list.
- Exclude anyone I already do business with
- Prioritize those who would gain if my business grew, or who would want my first-hand data or experience
- One line each on why they're connected; order them by how unlikely I am to have noticed them

④ Find your objective value in the record (Step 2)

You're an outside analyst who knows nothing about my business.
Read actionlog/, the journal entries in inbox/ (last 3–6 months), and ontology/organizations/.
1. List recurring tasks by frequency
2. For each: who would value it, why, and roughly what they'd pay
3. What I'm most likely underrating
4. First-hand data only I have, as shown in the record
Ignore my self-description; judge only from what the records show.

⑤ Turn every new contact into ideas (Step 2)

Add this to the rules file (AGENTS.md):
- When a new person file is added to ontology/people/, read research/role-map.md and research/revenue-stages.md,
  and propose 3 business ideas we could build together (same industry: synergy; different industry: where my strengths meet theirs)
- Append them to research/ideas-by-role.md. I decide whether to act.

⑥ Expand, then filter (Step 3)

Read research/role-map.md, ontology/people/, and research/revenue-stages.md.
1. One new business possibility per role (first pass)
2. Estimate, as an equation, what share of the target stage's revenue each could carry
3. Sort into lead candidates (10%+), supporting, and parked
4. For the top three leads, answer six questions:
   the role's recurring problem / data only we have / what they have that we don't / what we borrow and what we give / the outcome and who pays / the minimum proof
   Write "blank" where you can't answer. Don't fill in by imagination.

⑦ Design a high-margin new business (Practice, steps 1–4)

Read research/revenue-stages.md, research/role-map.md, and the list of first-hand data from ④.
Propose 3 new businesses from "first-hand data only I have × someone on the role map."
Conditions: no new inventory, equipment, or location / my hours don't grow with volume / reuse records, experience, and results I already have / the payer can explain the value in money
For each: who / what / price / costs / my hours per job / gross profit (count my hours as a cost)
Then count how many of the five margin factors lean to the high-margin side.

⑧ Write the 30-day verdict criteria first (Practice, step 7)

For 【business】, write the verdict criteria before I start.
[Funnel verdict] Within 30 days: at least 【n】 inquiries / at least 【n】 paid sign-ups. If not, how to tell whether the cause is demand, the funnel, or the offer.
[Outcome verdict] How many months until the outcome is clear, and what number I'll check then.
Include the 30-day date (I'll put it on my calendar).

Publish the landing page using S3’s copy-paste path ⑦⑧, after passing the pre-launch gate below.

---

The core: B2C, B2B, B2M

Split by who you deliver to, and there are only three revenue models. Treat them as parallel options, not a ladder.

B2C = consumers. B2B = businesses. B2M = the market itself — not one company but the whole industry: standards, certifications, licenses, marketplaces, data. Something built once that everyone in the field uses.

Model	To whom	What	Price	Reach	Strength	Weakness
B2C	end customers	direct service	low	narrow	richest first-hand data; fast feedback	needs volume; your hours cap it
B2B	peers and businesses	support, setup, systems	high	national	big tickets; predictable	needs standardization; long sales cycles
B2M	the whole market	standards, certification, licenses, marketplaces, data	usage or recurring	industry-wide	no ceiling; high margin	needs credibility and legitimacy; slow

There’s more than one starting point

Start	Path	Typical
A: B2C first	B2C (shop, classroom, field) → peers ask for help → B2B → your judgment becomes the standard → B2M	restaurants, tutoring, real estate, salons — people who already run a front line. The front line keeps producing knowledge, so what you say never runs dry. This guide mainly assumes this path
B: B2B first	B2B (systems or supply for businesses) → once supply is stable → B2C or B2M	IT, manufacturing, wholesale. Launch to consumers too early and demand swings break your supply. Harden the system on business orders, then go to consumers
C: B2M first	B2M (media, community, standards, associations) → B2B to the businesses that gather → then B2C	research, associations, communities. Build standing in the market before having a first customer — it happens

Path C works when research, a public role, or lived experience is itself a source of legitimacy. But a platform without substance hollows out. If you have none of those, run at least one small real business first — S3’s “prove you’re real” applies.

Combine, don’t graduate

Model	Why you keep it
B2C	the source of first-hand knowledge. Drop it and your B2B and B2M content goes stale
B2B	steady cash flow, and the test of whether your standard works elsewhere
B2M	the ceiling remover. Without it, you stay capped by hours and headcount

The move is simple: from the one that’s already running, choose the next one to add.

Decide what to add by conditions, not by vibes

“Peers keep asking me for advice — time for B2B.” Not enough. Being asked and being paid are different things. Judge in two layers: the necessary conditions (which no path can skip), and the evidence that shows you meet them (which can come from many routes).

Model	Four necessary conditions
B2B	① a defined deliverable (scope, period, responsibility) ② explained in the buyer’s economics (their revenue, cost, or risk, in their numbers) ③ reproducibility — evidence beyond your personal instinct (it works with other people, elsewhere; how to build that is in S6) ④ you know who has budget and authority
B2M	① a problem shared by many parties ② legitimacy to set the standard (“why you?”) ③ a decided way to charge (standard, certification, license, marketplace, data) ④ neutrality built into the structure (conflicts of interest separated up front)
B2C	① supply survives demand swings ② unit cost is known ③ one clear path from inquiry to delivery to follow-up ④ a way to capture first-hand information — the real value of B2C

B2B’s ③ matters most. What only you can do isn’t a product — it’s personal instinct. That’s why most B2B offers don’t sell.

Evidence can come from many routes: your own operating track record (three months running it, with improved numbers), research, having built it elsewhere, a public role, lived experience, or simply being earliest and most consistent. Any route works, as long as someone else can verify it. No route at all means no evidence — be honest about that.

Add a model without its conditions and it eats the credibility of the ones you have: B2B that doesn’t reproduce becomes “all talk”; B2M without legitimacy becomes a self-declared badge; B2C that runs out of stock spreads fastest of all.

B2M means deciding how you charge

The most common B2M failure: more activity, no revenue model. Talks, associations, and collaborations show influence; none of them pays by itself. Choose one to start:

Model	Revenue	Example	Fits when
Standard	usage or compliance fees	a shared data format or process standard	data or processes are comparable across companies
Certification	exam, renewal, partner fees	certified instructors or locations	inconsistent quality is an industry problem
License	fees for materials, brand, or tech	curriculum licensing, franchising	a reproducible package exists
Marketplace	transaction or listing fees	matching platforms	you have access to both buyers and sellers
Data	subscriptions, analysis	industry databases and dashboards	first-hand data keeps flowing in

Governance before launch: third-party review and oversight; renewal and revocation criteria; an appeals channel; a published change history; data-handling rules; and conflict-of-interest separation between your own sales and the certifying or reviewing role. A system where the certifier is also the biggest beneficiary looks like favoritism from outside. Only systems that separated these first last.

The three models are about who you deliver to. There’s also how you deliver — designing for each customer individually from their data. That’s covered in S6.

---

The method: build revenue doors around your center business

Three steps that alternate widening and narrowing:

Step 1  Choose your center business and set revenue stages
        └ fix one core + put Stage 1/2 in equations (= the tool for narrowing)
              ↓
Step 2  Map every role around the center business
        └ list them yourself → let an outsider fill the blind spots (= widening)
              ↓
Step 3  Expand the map into new business
        └ first pass for volume → second pass for depth → filter by stage

You run these three steps twice. First pass (Stage 1): find the unit that sells. Second pass (Stage 2): the center business becomes that unit, the role map becomes the people who could replicate it, and the expansion becomes the replication system. If you’re in Stage 1 now, read only the first pass.

Step 1: choose your center business and set revenue stages

Where the core is (1-A) and how much that core should earn (1-B). A number without a core is wishful; a core without a number gives you no way to throw ideas away later.

1-A: choose the center business

One condition: the point where you’re most deeply committed to the market — where expertise and experience pile up fastest, not necessarily where the money is. You don’t start it from scratch: if there’s a place where field knowledge is generated every day, that’s it.

If several candidates come to mind, score them and pick one (prompt ①). Keeping two “for now” doubles your role map and leaves you half-in two markets.

Criterion	What to look at
Repetition	how often that front line runs — daily, or a few times a year
First-hand data	how much and how detailed
Closeness to customers	how raw and fast the feedback is
Profit potential	profitable now, or soon
Transferability	can the knowledge be reused elsewhere

Scoring example (a food business):

Candidate            Rep  Data  Close  Profit  Transfer  Total
In-store baking       5    5     5      3        5        23  ← chosen
B2B consulting        2    3     4      4        3        16
Podcast               4    2     2      1        3        12

Profit doesn’t need the top score. Consulting pays more per job, but baking wins on repetition and data. The center business is where knowledge accumulates most richly, not where it pays most.

“Where knowledge comes from” and “what makes money” can differ.

	Knowledge source (= center business)	Profit product
Role	keeps generating first-hand knowledge	generates cash
Chosen by	repetition, data, closeness	price, margin, close rate
Across stages	kept at every stage	replaced as stages change
Example	running the shop / teaching students / on-site work	premium course / setup service / license

Usually they differ. The danger is not noticing, mistaking the profit product for the center — and shrinking the knowledge source because “it doesn’t pay.”

1-B: design the revenue stages

Set the target before Step 2. The role map is a widening tool; AI will generate endless ideas. Without a narrowing tool you end up with a long list and nothing chosen. A target number gives you grounds to discard ideas that can’t reach it.

First, define “revenue” for your business type.

Type	Examples	Count as revenue	Typical gross margin	Watch out
Product	restaurants, retail, making things	sales	30–60%	the gross profit after materials is the real business
Brokerage	real-estate agency, recruiting, agencies	commissions, not transaction value	70–90%	counting transaction value inflates revenue 10–30×
Service	consulting, education, trades, professional services	fees billed	50–80%	your time is the cost — measure profit per hour
Asset	owning and renting property	rent	—	separate sale proceeds from cost
License	courses, standards, certification, software	usage and license fees	70%+	near-zero marginal cost; the star of Stage 2

Write your type in one line. Across several types, business revenue can be added up, but never average the margins — a blended 63% from 50% and 80% tells you nothing about which to grow. Don’t add asset income at all: it’s capital income, not business income. Mix it in and you can “hit the target” without the business growing.

Three stages. Growth isn’t a smooth line; the structure changes, and so does the verb.

Stage	Annual revenue (rough guide)	Essence	Verb
Stage 1	up to ~$250k	no sellable unit yet; your hours cap revenue	find, redefine
Stage 2	~$250k–$2.5M	the unit exists; how many you can run in parallel sets revenue	replicate
Stage 3	$2.5M+	the system exists; organization, capital, quality control	industrialize

The dollar figures are a rough guide; set thresholds that make sense for your type and region. What matters is the shape: the first step is a 10× jump, and you can’t cross 10× by doing more of the same.

S4 covers Stages 1 and 2. Stage 3 belongs to S5 (money data), S6 (team), S7 (AI-centered team, Japanese), and S8 (capital planning, Japanese). Stage and B2C/B2B/B2M are independent — combine them.

Write every stage as an equation, because “low six figures” can’t be checked. And the shape differs between stages:

Stage 1:  price × customers × frequency
          e.g. $40,000 × 3 clients + $1,000 × 6 clients × 12 months = $192,000
               ↑ your own hours hit the wall around here

Stage 2:  price × customers × frequency × [parallel units] − management loss
                                            ↑
                              stores, licensees, instructors, SKUs, units, crews
                              = units that run without you
          e.g. $400,000 per store × 6 stores = $2,400,000
               $100,000 per licensee × 10% royalty × 240 locations = $2,400,000
               $400 average order × 6 per year × 1,000 business customers = $2,400,000

Stage 2 adds one variable: parallel units. In Stage 1 you grow by raising price, not volume (volume at the same price breaks your body first). In Stage 2 you change neither price nor volume — only the number of units.

Fill in seven items for Stages 1 and 2 (prompt ②): who/what/price; the equation; gross margin and why; your time split; the AI, contractor, and staff setup; one bottleneck (cash, people, product, credibility, or licensing — list several and you fix none); and a quantified trigger for the next stage. Then mark where you are honestly, or you’ll just admire a Stage 2 picture.

Step 2: map every role around it

List every person and business your center business touches. For a restaurant: drinks distributors, food suppliers and markets, the landlord, equipment makers and repair techs, staff, nearby restaurant owners, review sites and social media, customers, the health department, the accountant, delivery services. Save it as research/role-map.md. Every role is a potential business partner.

The blind spot: you only list who you already pay or who pays you. But there are usually as many people who would gain from your business without any deal yet — and you can’t think of them, which is exactly why you’ve never met them.

Type	What it means	Restaurant example
Indirect beneficiaries	gain when you grow, no deal yet	nearby property owners (a busy restaurant lifts the street)
People who want you as material	your front line itself has value	trade media, researchers, culinary schools, city case-study programs
Same problem, different solution	look like competitors, complement you	prepared-food makers, caterers, cooking schools
Upstream of upstream	your suppliers’ suppliers	farmers, fishers, co-ops
Downstream of downstream	your customers’ world	the companies your regulars work for (corporate dining)
Public bodies	seen only as regulators	economic development offices, chambers of commerce, small-business centers
Adjacent businesses	same customer, different moment	salons, gyms, photographers, event venues
Future people	not customers or suppliers yet	students, career changers, successors, newcomers to the area

Most people are missing five or more. You won’t find them by asking yourself — you can’t search your memory for what’s outside your awareness. Use AI as an outsider who doesn’t know you (prompt ③), and force a number: ask for 30, or it stops at the safe ten. The surprising ones come late in the list.

Your own value is the other half of the blind spot. What you do every day stops looking valuable to you — “I cook for a hundred people a day while managing food cost,” “I track twenty students’ understanding individually every week,” “I walk through thirty vacant houses a year.” Rare from outside; ordinary to you. Your actionlog/ holds it without your self-assessment bias (prompt ④). Sort by frequency: what you repeat most is what you take for granted and what you’ve accumulated most of. Mark the items that make you think “that’s valuable?” — that discomfort is exactly where the gap is. Three months of log makes this much sharper; a month or two still works.

Make it real. The AI lists types, not people. Pick about five, and find at least one real business or organization of that type in your area, check whether anyone in people/ or organizations/ knows them, and if not, who could introduce you. Add each to ontology/organizations/. Redo the map every quarter.

Tip: every new person is a new set of ideas. With the rule from prompt ⑤, each time you add someone to people/, the agent proposes what you could build together. Keep adding the people you meet and ideas pile up as fast as your network grows.

Mini-exercise: three real people, 15 minutes. Pick three people from a different industry (in a class, the people in the room). Don’t guess — ask them, openly: “What takes most of your time these days?” Listen for “every month” or “every week” — one-off problems don’t make businesses. Write it in their own words; tidied summaries collapse into generic problems.

ontology/people/【name】.md
- Industry and role / what they deal with
- What they're struggling with (their own words)
- What they have that's real (location, license, customers, equipment, reputation, venue)
- Where they connect with me

Run the rule, add ideas to research/ideas-by-role.md, don’t act today, and mark the ones that made you think “I’d never have thought of that.”

Step 3: expand the map into new business

Stop seeing roles only as customers or suppliers. Expand all of them quickly for volume (first pass), dig into the promising ones (second pass), then filter by your revenue stages.

Role	Possible business
Drinks distributor	Feature their products in your content → paid partnership
Landlord	Help fill empty units → produce the concept of the new tenant
Staff	Turn their growth into content → a training program for the industry
Nearby restaurant owners	Joint events → a local restaurant community → B2M
Health department	Food-safety knowledge → licensing help for new owners
Accountant	How restaurants should read their numbers → owner seminars

The more you know about the real person in each role, the better the ideas:

Without background:  "Ideas for working with a wine shop" → generic
With background:     read people/wine-shop-owner.md (focused on natural wine, open to small collaborations)
                     → "A limited pairing menu with their wines, promoted on both sides —
                        new customers for them, a reason to visit for us"

Second pass: six questions. Ideas like “turn X into content and sell consulting” are endless and alike. For three to five promising ones, answer:

Question	Fill in
① The role’s recurring problem	what they struggle with every month or week
② Data only you have	records, information, experience nobody else has
③ What they have that’s real	location, license, members, reputation, equipment
④ Borrow / give	what you borrow from ③, and what you give back
⑤ Outcome and payer	whose result improves, and who pays (not always the same person)
⑥ Minimum proof	the least evidence needed to show it works

A blank ② means anyone could do it. A blank payer in ⑤ means it won’t earn, however good. ③④ work fastest: asymmetry is a resource. Borrow their track record and venue as your offer’s “proof of being real”; in return, lift their customer flow and efficiency with your system. You don’t have to build what you lack — someone on the map already has it.

Filter by stage (prompt ⑥):

Verdict	Criterion	Do
Lead candidate	can carry 10%+ of the target stage	consider the order to start
Supporting	small, but feeds a lead or builds trust	run alongside a lead
Parked	small and unconnected	keep in research/; don’t start

Don’t delete parked ideas — some become leads when your stage changes.

---

Stage 2: replicate the unit

This is the second pass: take the unit that sells and make it run without you, multiplied. No new ideas — same unit, new arrangement. What you sold to 12 clients becomes 240 locations, 6 stores, 1,000 business customers.

Why the Ontology moves Stage 2 within reach of one person

Traditionally, Stage 2 meant hiring. The real wall was human: a founder can directly manage roughly six people. Add parallel units and management cost grows with them — store managers, then managers of managers. That’s the ”− management loss” in the equation: revenue rises and profit stalls.

Management cost breaks into four things:

Management cost	Traditional fix	With the Ontology
Knowing the state (who, what, how far)	meetings, reports	state is structured in people/ and organizations/; AI reads across it
Passing on judgment (what to do when…)	calls, manuals	decision criteria live in concepts/; every unit judges by the same rules
Checking quality	site visits, audits	AI shows the gap between records and the standard (S6)
Records and handover	handover docs, minutes	the action log fills itself (S3)

All four used to mean “people watching people,” so cost grew with headcount. Build the structure once, and the AI’s cross-reading costs almost nothing extra: watching 60 units costs about the same as watching 6. That’s the management-cost revolution — a founder plus AI plus a few contractors can run what used to take a team of fifteen.

The honest limit: the Ontology cuts the cost of knowing, deciding, passing on, and recording — not the hands-on work. People who serve, build, or treat customers are still needed, quality still depends on them, and trust with partners is still maintained by humans. The benefit is large for courses, memberships, matching, and software; medium for standardized products and units (logistics remain); limited for stores and standardized services (only headquarters gets lighter). Know which you are.

Is it ready to replicate?

• The unit fits in one sentence — what, done how, delivered in what form (“Big custard pudding, cut to order, from a market stall.” “Personal statements, finished through dialogue with AI, in a four-month course.” “Everything about an inherited house, documented in one binder.”) Test: could a stranger build the same thing from that sentence? “AI consulting” fails. If you can’t say it, you’re still in Stage 1.
• The procedure is written down — no decisions left only in your head
• Someone other than you has delivered it successfully (S6)
• Unit cost and margin are known — skip this and revenue rises while profit falls

Eight replication patterns

Pattern	Unit	Example equation (~$2.4M)	Fits / hard part
P1. Replicate stores	store	$400,000 × 6 stores	food, beauty, retail / location mistakes, capital
P2. Standardize a service	standardized job	$20 × 100/day × 300 days × 4 locations	haircuts, repair, cleaning / staffing, price wars
P3. Replicate a curriculum or membership	instructor, class, member	$100/mo × 2,000 members × 12	schools, studios / instructor quality, churn
P4. Matching or software	account	$250/mo × 800 accounts × 12	vertical software, staffing / acquisition, churn
P5. Standard products to many customers	product, channel	$400 order × 6/yr × 1,000 customers	supplies, private label / inventory, logistics
P6. Standard units	home, room, building	$240,000 × 10 per year	homebuilding, lodging / land, rates, inventory
P7. Roll-up	acquired company	three $800,000 businesses	fragmented local industries / price, integration
P8. Organize high-ticket projects	project, team	$60,000 × 40 per year	B2B projects, construction / hiring, consistency

P8 has a ceiling around this size: raise prices, productize, or connect to another pattern. “More people, more projects” doesn’t break through.

The pattern follows the unit: a place (people come to you) → P1/P2; a person (teaching, treating) → P3; information or connection → P4; a product → P5; an asset → P6; a project (different each time) → P8.

Divide the target by the unit and it becomes countable: $2.4M ÷ $400k = 6 stores. ÷ $10k royalty = 240 locations. ÷ $240k = 10 units a year. The same counting as Stage 1, one digit up. Stage 3’s subjects — organization, quality, capital — are handed to S5–S8.

---

Cutting what you added

Depending on a single revenue stream is a risk — a pandemic, a disaster, a platform algorithm change can zero it overnight. And now that management cost no longer scales with headcount, you can run more businesses. But there’s a limit, and adding and cutting only work as a pair.

Why you must cut: a business uses your time even with zero sales — replies, relationships, renewals, inventory, updated listings. AI and contractors take the execution; judgment and relationships only come from you. Most people stuck in Stage 1 aren’t failing to create products — they’re carrying every Stage 1 business while adding the next.

The time portfolio. List every running business:

Business	My hours/month	Gross profit/month	First-hand knowledge	Can AI or contractors run it?
(A)			high / mid / low	yes / partly / no

Situation	Verdict
high profit, delegable	delegate (fastest win)
high profit, only you can do it	standardize so it can be delegated (top Stage 2 priority)
low profit, rich first-hand knowledge	keep (a center-business candidate; don’t judge by money)
low profit, thin knowledge, not delegable	stop

Never cut a knowledge-rich business for its numbers — that’s where your expertise comes from. Then add up the hours column: if it exceeds the time you have, your stage won’t rise. Lower the total before adding anything.

Stage	Main revenue streams	Why
Stage 1	1–2	you’re on the front line; spreading thins your knowledge
Stage 2	up to 3	you own unit design and partner relationships; beyond 3, everything goes half-done
Stage 3	3+ possible	only with a responsible lead per business (S7)

Three ways to cut: delegate (keep delivering, remove your judgment), freeze (stop new sign-ups, serve existing customers), close — always with an explanation and a handover for existing customers. Don’t delete closed businesses from the Ontology: record what you tried, how far it got, and why it stopped. “No demand” doesn’t come back; “wrong stage” is a restart candidate when your stage rises.

Judge every new thing at 30 days — on numbers that can move in 30 days, that you can count yourself, and that lead to revenue. Views and time-on-page move, but if they don’t lead to revenue, they aren’t KPIs.

Business	Visible in 30 days	Not visible in 30 days
Tutoring, education	inquiries, trial sign-ups, price acceptance	grades, admissions
Food, retail	orders, repeat rate, average ticket	becoming a staple, brand recognition
Real estate	consultations, viewings, quote-to-contract rate	closed sales, rental performance
Consulting, B2B	meetings booked, proposal acceptance, price acceptance	client KPIs

At 30 days you judge the funnel, not the outcome. Write down when the outcome becomes clear (for tutoring, after exams; for farming, after harvest) before you start. Past 30 days: above target, continue; below, find whether demand, funnel, or offer failed and rebuild; no movement at all, stop. Postponed businesses quietly eat your hours.

---

The compounding cycle

The power of S4 is that the cycle keeps running. The richer your Ontology, the better the AI’s ideas; every test updates the Ontology; the next ideas get better still. That’s compounding.

Monthly, you test revenue hypotheses — not new businesses. Launching a full business every month multiplies purchasing, inventory, licensing, training, and support, and destabilizes your center business.

Cycle	What runs	Examples
Monthly	revenue hypothesis tests	change the price, the format, the audience, the funnel; one new landing page
Quarterly	decide what becomes a business	turn responsive tests into ongoing offers
Yearly	revise the standard	product line, procedures, pricing

One or two real new businesses a year is healthy — the result of twelve monthly tests. Know your minimum cycle: services, information products, and referrals can go monthly; physical products quarterly; licensed businesses only after the license; capital-heavy moves quarterly to yearly.

① Have the AI read the Ontology (people/, organizations/, concepts/)
        ↓
② Choose one hypothesis to test with the AI (30–60 min)
        ↓
③ Pass the pre-launch gate; build and publish a landing page (1–3 hours)
        ↓
④ Look for first revenue
        ↓
⑤ Judge at 30 days: continue / rebuild / stop
        ↓
⑥ Write the result back into the Ontology → ①  (record the reason even when you stop)

Read ontology/people/, organizations/, concepts/, and research/revenue-stages.md.
Given my resources, network, and market understanding, propose 3 revenue hypotheses I can test within 30 days.
Not whole new businesses — change one of: price, format, audience, or funnel.
For each: who, what, price, and what result would mean "the hypothesis was right."

The pre-launch gate. A landing page takes orders the moment it’s live. An order you can’t fulfil is an incident.

Check	What
Start date	when you can actually deliver. Never imply you can sell before you hold a required license
Capacity	how many per month, and what happens beyond that
Required disclosures	ingredients and allergens for food, license numbers for real estate, business identity, refund and consumer-rights terms required where you sell
Storage and shipping	refrigeration, breakage, delays
Refunds and cancellations	unwritten terms mean every dispute lands on you

Before a license, you can still run demand surveys, tasting or trial sign-ups, waitlists, and share your plans. You can’t take priced orders, bookings, or pre-sales. A waitlist still measures demand.

Hosting: a private GitHub repository plus Cloudflare Pages, one repository per landing page so private files never leak in. First revenue: before paying for ads, pitch directly to people already in your Ontology — it’s faster. Log what landed and what didn’t, and update the related people/ and organizations/ files. That’s the feedback that makes the next cycle smarter.

Twelve tests in a year should leave one or two standing — the rest delegated, frozen, or closed. Twelve still standing isn’t success; it means nothing was judged.

---

Where S4 ends

S4’s job ends at finding the unit that sells (Stage 1) and designing how it replicates (Stage 2). Turning it into an execution plan is not S4’s job.

Over-planning drags you back inside the frame. An idea from outside your frame always lacks information at first — that’s why you never thought of it. Demand “proven feasible before starting,” and only ideas from what you already know survive: the frame again. The equations and margin math in S4 are for judging whether an idea could be a lead — getting the order of magnitude right is enough. The rest fills in after you start; the 30-day criteria are what let you stop.

“I’ll gather more information first” is the signal you’re slipping back. Ask: can this information be gathered at a desk, or only by starting? Waiting for the second kind isn’t research. It’s postponing.

Once there’s a brief, a page, and 30-day criteria, move the work into projects/<name>/ and route issues by type:

Issue	Go back to
Rework or sharpen the idea	S1 — rewrite the brief
Not enough on the counterpart, market, or competitors	S2
How to sell and promote it	S3
Proposals don’t close	the S1 ⇄ S3 loop (below)
You can’t get out of “only I can do it”	S6
Cash and upfront investment are unclear	S5

Deals close as a result of a cycle. As prices rise, buyers want to see “how it actually goes,” not a better pitch. So add a cycle, not sales technique:

① Let them experience a small piece (trial, pilot, partial delivery)
② Get their feedback
③ Rewrite the brief (S1)
④ Publish the feedback (S3)
⑤ The next person sees it and asks for a trial → ①

The number of cycles decides the number of deals. You can run it with zero customers and nothing to sell yet. And each pass fills in B2B’s conditions: the deliverable gets defined (③), value gets explained in their words (②), repeating ① elsewhere becomes reproducibility, and ④ builds your track record.

A free trial isn’t a discount. Discounts shrink value; a trial proves it. Give away the experience, never the finished deliverable — or you can’t charge for it afterward.

Beyond S4: real influence

What this framework finally gives you is influence — not follower counts, but standing: your standards and certifications in use, projects moved forward with others in your region. Influence doesn’t replace a revenue model; it’s what’s left after the revenue model works. Measure it by things others can verify: how many businesses use your standard, how many people you’ve certified and how many stay, the numbers your standard improved — and above all, how often deals happen inside your ecosystem without you. When it runs without you at the center, it has become part of the market.

---

Practice: design a high-margin new business

Join the tools into one line and build one new business in about two hours. Output: one brief, one landing page, 30-day criteria.

Why design from margin. Gross margin isn’t improved later; it’s mostly decided at the idea stage. Changing a product business into a service business midway isn’t an improvement — it’s building a different business. You get to choose the model now.

Factor	High-margin side	Low-margin side
Cost	reuses what you already have (near-zero marginal cost)	new materials every time
Your hours	don’t grow with volume	you’re tied to every job
Inventory	none	buy, store, waste
Real assets	borrowed from a partner (venue, license, members, reputation)	built yourself
Payer	can explain the value in money (businesses)	price-sensitive, hard to raise prices

Only ideas with three or more factors on the high-margin side go forward.

1. Materials (10 min) — revenue-stages.md, the role map with the outsider’s additions, and the list of first-hand data only you have. Without the third, go back to Step 2.
2. Three combinations (20 min) — “first-hand data × someone on the map” (prompt ⑦). Never from a blank page.
3. Filter with the five factors (10 min) — pick one with 3+ on the left; on a tie, the one that uses your data most.
4. Answer the six questions (30 min) — any blank, move to the runner-up. Imagined answers are where it breaks later.
5. Equation and margin (20 min) — gross profit = revenue − (variable costs + your hours × your hourly rate), with the rate taken from your center business’s gross profit per hour. Always count your hours. Can it carry 10% of the target stage? If not, it’s “supporting.”
6. Gate and landing page (40 min) — pass the pre-launch gate (switch to a waitlist if a license is needed), then the S3 eight-element brief, and publish with real proof above the fold.
7. 30-day criteria first (10 min) — prompt ⑧, and put the date on your calendar.

Done when nothing was filled in by imagination.

Symptom	Cause and fix
No first-hand data appears	the action log is too thin — firm up S3 Step 1
No clear payer	split beneficiary and payer; can a business pay instead?
Margin won’t rise	you’re still in your usual model — try service or license
All filled in, but dull	it doesn’t use your data; anyone could do it
All three ideas look alike	the role map is only existing partners — redo the outsider step

---

Teaching S4: three 2-hour sessions

Prerequisite: S1–S3. An action log of three months or more is ideal; one to two still works. Cover the first pass (Stage 1) only. Self-learners: leave a week between sessions — the homework is the real work.

Session	Theme	By the end
1	Center business, stages, role map	Score candidates and pick one; write revenue-stages.md with equations and judge where you are; build the role map and have the AI add 30 outsiders; interview three people and add them to people/
2	Expand, filter, and know what to cut	First pass, six questions on 3–5 ideas, stage filter; time portfolio; the monthly cycle and its limits for your type
3	Practice: high-margin new business	Three combinations → one by the five factors → six questions, equation, margin, gate, 30-day criteria; swap briefs in pairs for review; publish the page within 48 hours

Three facilitation rules: don’t let anyone keep two center businesses (the score decides); listen before showing AI ideas in interviews; say up front that switching to the runner-up is fine when the margin math sinks an idea — or people fudge the numbers.

---

Related skills

• Before this → S3: Take control of your voice
• Next → S5: Build your money data
• Out of “only I can do it” → S6: Manage your team with AI

---

SOVREN Framework is open-source. The English edition covers S1–S6; S7–S9 and the milestones are published in Japanese. The method is practiced in weekly live classes in Yokohama, Japan, with a report after every session.