FJSCALING · OPERATING AUTHORITY · COMPILED 11 JUL 2026 · 3-ROUND INTERVIEW + SYNTHESIS

The document that answers "what should I do."

Every decision in here was made at planning time, from your own answers. Nothing is decided live. When the wall hits, this document is item zero: open it, read the relevant tab, execute. Claims are tagged FACT / INFERENCE / UNKNOWN throughout — untagged sentences are structure, not claims.

01 · NORTH STAR

The definition

One paragraph. If a sentence in your head contradicts this paragraph, the paragraph wins until a checkpoint changes it.

HVAC companies in one test-cell metro pay FJScaling to fix their Google Business Profile until it ranks in the local Top 3. Entry offer €500 (audit + implementation). Core service €1,500/month, 3-month minimum term, no cancellation after month one, carrying the risk-reversal "Top 3 in 90 days or you don't pay." Upsell package €4,000. Clients are found through personalized email/DM first-touches → call → audit with offer → close call. Delivered solo, from a laptop, from anywhere. FACT — YOUR STATED PRICES & CHANNEL

The test cell is HVAC · Tampa, Florida — year-round AC demand, dense small-business pool, English-speaking. INFERENCE You said "I don't know if HVAC is the right place, but for now sure." Correct posture. The cell is a coordinate, not an identity. The kill rule in Tab 06 fires it automatically; the queue in Tab 06 replaces it automatically. You never have to defend this choice — you only have to send into it.

€6,000per month, after tax — phase-one done
12 APR 2027deadline — month 9, computed from your pipeline numbers, not picked
WHAT THE NUMBER BUYS — YOUR WORDS, KEPT CLOSE
An apartment significantly bigger than what I can afford now. Money to spend on dumbshit — clothes, anything I set my mind on. Taking my girl out without feeling stressed. Going to the gym without money stress.

Optimization order, your ranking, strict: FREEDOM → CRAFT → PEOPLE → STATUS. The rooms you named (circles of people not limited by money) sit behind the number. The number is the entry fee. FJScaling is the vehicle, not the destination — that framing is already documented and it holds.

The four-input status board

The nudge fires only when all four read LIVE. Two are live at compile time. Each DEAD input has exactly one flip action, already scheduled. Tap a card for what makes it live.

Defined RepLIVE
Live because: the rep exists in one sentence (Tab 03), completable in one sitting, controlled by you alone. Stage 1 never has to run unassisted again — the list runs it.
tap to collapse
Fast FeedbackDEAD
Dead because: no scoreboard exists — your words: "it's all in my head." Flips LIVE: Day One, Step 0 — one paper index card, two minutes. The mark is the feedback, delivered inside the credit window.
tap for flip action
Trusted SourceDEAD
Dead because: no seeded root of trust; the veto correctly rejects an adversarial information space. Flips LIVE: diligence window Mon 13 Jul → Sun 19 Jul. Decision at the 19 Jul, 20:00 review. Fallback pre-decided: least-bad of the top 3 encountered. No extension. (Your dates said Mon 13–Sun 20; the Sunday is the 19th — window corrected.)
tap for flip action
Safe FrameLIVE
Live because: the daily game is XP-denominated skill acquisition (Tab 03), ranked against no one. Money is banned from the daily view — it appears only at checkpoints, where the auditor is scheduled instead of ambient.
tap to collapse
EXECUTABLE Both DEAD inputs flip through actions already on the Day One list (Tab 08). Nothing on this tab requires thought. Go to Tab 08.
02 · THE MODEL

Who pays, for what, and the math between you and €6k

Real numbers where you gave them. UNKNOWN markers where you didn't. Assumed rates are tagged INFERENCE and get replaced by measured rates at checkpoints — never by feelings.

WHO PAYS

Owner-operators of HVAC companies not currently in the local Top 3 of Google's map pack, in the Tampa test cell. They already spend on getting found (ads, lead-gen services) — whether they'll pay you for this is untested. UNKNOWN — ZERO WITNESSED DEMAND, YOUR STATEMENT

THE OFFER LADDER

€500 entry — GBP audit + fixes.
€1,500/mo core — full optimization, 3-mo minimum, no cancel after month 1, Top-3-in-90-days-or-free.
€4,000 upsell — total package. FACT

CHANNEL

Personalized email or DM → book call → deliver audit with offer → second call to close. FACT Cold calls excluded below a €10k/mo baseline — encoded as constraint R5, Tab 06. FACT — YOUR REFUSAL

DELIVERY

Solo, laptop, location-free. Capacity ceiling: 4 clients at the current hour floor; at 4, the hour structure gets renegotiated with revenue as evidence. INFERENCE — YOUR ESTIMATE

Unit economics

LineValueStatus
Minimum contract value (entry + 3 × core)€5,000FACT — ARITHMETIC ON YOUR PRICES
4 clients on core retainer€6,000/mo grossFACT
€6,000 gross vs €6,000 after-tax targetGap exists. Closed by client #5 or one €4k upsell.UNKNOWN — ONE OSVČ CALC, DUE 9 AUG
Guarantee exposureIf Top 3 missed at day 90, client pays €0. Cost = your delivery hours. Contained by the 3-month-term condition.FACT — YOUR TERMS
Delivery ability ("3 focused months of SEO gets results")Your stated confidence. Zero delivered SEO projects behind it. Client #1 is knowingly the test case.INFERENCE — VERIFIED ONLY BY CLIENT #1'S RESULT

Pipeline math

Reps → conversations → clients → revenue. Two scenarios, both yours. The truth arrives by day 30 as a measured number.

ScenarioReplyCloseSends per clientFirst client at 55 sends/wk
Your pessimistic INF1%10%~1,000~18 weeks — unworkable; this is why rates get measured, not trusted
Expected case INF3%20%~167day 45–60
VOLUME MODEL 5 personalized touches per weekday + 15 per weekend day ≈ 55/week. Your "20–30/day" claim was cut: at a 1-hour weekday floor that is 2–3 minutes per touch, which is not personalized — it's spray. Week 1 measures your true throughput; the quota adjusts to the measurement at the 19 Jul review. INFERENCE UNTIL DAY 7
GATES DERIVED FROM YOUR NUMBERS First paying client by day 90 (11 Oct 2026). €6,000/month by month 9 (12 Apr 2027). Both computed from the expected case; both re-computed at day 30 from measured rates.
EXECUTABLE The model has exactly one moving part you control today: the touch. Everything in this tab is either arithmetic or a checkpoint's job. Go send.
03 · DAILY OPERATION

The rep, the list, the card, the wall, the shutdown

Everything on this tab is completable. Nothing on this tab is aspirational. The feeling is not consulted anywhere below.

One personalized first-touch to one qualified business in the current test cell, containing one specific claim about their pain point. Sent = complete.

The reply is not part of the rep. The client is not part of the rep. You control sending; you score sending. The filter for fake reps, your own words operationalized: a real rep puts the offer in front of a human who could pay. Audits and follow-ups count as secondary reps under the same standard. Tooling, prompts, and design passes are training equipment, not training.

THE DAY — WEEKDAY (YOUR SCHEDULE, KEPT)
  • 15:30–16:30 — the floor hour. List item 1, then 2, then 3. Five touches live here.
  • 16:30–17:30 — bonus hour when it exists. Never planned on.
  • 17:30 — gym. 19:30 — dinner, Stelli. 21:30 — bed.
  • Shutdown ritual runs before the evening starts, ~16:30–16:40.
WEEKEND / TRAVEL
  • Weekend day: one 5-hour block → 15 touches + audits for any live replies.
  • Travel week (1×/month): quota drops to 3 touches/day. Nothing else changes. The Sunday review still runs.
  • No calls during sleep hours or Stelli time — constraint, not preference (R5).
NIGHT-BEFORE CLOSED LIST — RUNS INSIDE THE SHUTDOWN RITUAL, 5 MIN Tomorrow's list: 3 reps maximum, ordered. Written in the planning state, where your pipeline works. In the morning-after execution state, the list is the selector. If the nudge shows up, welcome — it holds no vote. If the list exists, "what should I do" has an answer before it's asked.
THE SCOREBOARD — ONE PAPER INDEX CARD, NOTHING ELSE Columns: date · touches (tally) · XP · one-word note. Mark the tally the second a touch sends — the mark is the feedback, inside the credit window. This is the gym's logbook, transplanted. Money never appears on this card (Tab 05 ban).
EventXPLevel ladder
Touch sent1L1 start · L2 at 50 · L3 at 150 · L4 at 300 · L5 at 500 · L6 at 750 · L7 at 1,000. Cumulative, never resets, survives niche kills. You asked for levels and score — this is it. The level measures the only thing you fully control: volume of skill-building exposure.
Reply received5
Call booked25
Client signed100
THE 60-SECOND WALL PROTOCOL — FROM YOUR MANUAL, UNCHANGED 1 · Name it: silence (nothing surfacing) or jam (options surfacing, all vetoed) — say which.
2 · Silence → open the closed list, do item 1. Stage 1 is not asked to perform.
3 · Jam → item 1, or coin-flip the top two. 25-minute timer. Evaluation forbidden until it rings.
4 · After: mark the rep. Continue, or run it again.
SHUTDOWN RITUAL — 10 MIN, SAME TIME DAILY, THE DAY'S TERMINAL BOUNDARY 1 · Score: mark the card, look at the number for five seconds — the day is now defined as done.
2 · Capture: every open loop written to one page. Not solved. Written.
3 · Plan: tomorrow's closed list, 3 reps, ordered.
4 · Close: the fixed phrase — "shift over" — spoken or written, identical daily. After it, work thoughts get one line on the pad, never engagement.
EXECUTABLE Tonight, regardless of anything else: run the shutdown ritual once and write tomorrow's 3 reps. That single act makes tomorrow's wall structurally impossible — the answer will already exist.
04 · CHECKPOINTS

Evaluation happens here and nowhere else

The veto isn't deleted — it's rescheduled. Mid-rep evaluation is off, permanently. Every Sunday 20:00, 30 minutes: read the card, check the criteria below, write one word on the log — CONTINUE / ADJUST / KILL. "Observe and adjust" was rejected as a review spec because it is not scheduled behavior; this is.

GATE 1 — DAY 7 · SUN 19 JUL 2026, 20:00 Metric: reps completed vs scheduled; true send throughput; trusted-source diligence result.
Pass (green): ≥80% of scheduled reps ran, and one source is named with playbook in hand — fallback: least-bad of top 3, decided tonight, no extension.
KILL CONDITION: <50% of reps ran → the model was never tested. This is an execution or substrate fault, not a strategy fault: re-read manual Tab 04, halve the quota, and book the bloodwork the manual already flagged (B12/folate, ferritin, TSH). No model changes allowed on execution failures.
GATE 2 — DAY 30 · SUN 9 AUG 2026, 20:00 Metric: cumulative sends, replies, measured reply rate; after-tax math resolved.
Pass: ≥1 positive reply exists. Recompute the day-90 and month-9 gates from measured rates.
KILL CONDITION: 150+ sends into the cell with zero positive replies, OR 4 consecutive explicit "no" replies (your rule, given denominators) → the cell is dead. Same day, load the next cell from the queue (Tab 06, R6). A dead cell is data, not a verdict on you.
GATE 3 — DAY 60 · SUN 13 SEP 2026, 20:00 Metric: calls booked, audits delivered to live prospects.
Pass: ≥1 audit delivered to a real prospect.
Adjust: replies exist but no calls book → change the script — one variable, versioned, logged. Not the offer, not the cell, not everything at once.
GATE 4 — DAY 90 · SUN 11 OCT 2026, 20:00 — THE FORK Pass: ≥1 paying client → CONTINUE. Tooling ban status checked (lifts only at client #2).
Fail, branch A: ≥80% of reps ran, 0 clients → the model got a real test and failed it. Change exactly ONE variable — offer, channel, or cell — and run 30 more days.
Fail, branch B: <50% of reps ran → nothing was tested. Substrate protocol: bloodwork if not yet done, manual review, quota halved. Model changes forbidden — changing a model you never ran is how the last three years happened.
Skip-proofing: calendar reminder (your answer) plus the standing rule that the review is a rep — it gets a tally mark. An unmarked Sunday is a visibly broken streak on the card you look at daily.
EXECUTABLE Four calendar entries, created in Day One Step 4. After that, this tab runs itself — you just show up at Sunday 20:00 and read.
05 · DATA

Leading feeds the card. Lagging feeds the checkpoint. Never crossed.

The Supervisor audits in money. Scoring money daily is how it confiscates reward from the only inputs you control. So money — and everything money-shaped — is structurally banned from daily view.

LEADING — DAILY, ON THE CARD, YOU CONTROL 100%
  • Touches sent (the primary metric — you asked me to pick: this is it)
  • Audits shipped
  • Follow-ups sent
  • Objections catalogued (each new one is an asset, +1)
  • Script version number
LAGGING — CHECKPOINT-ONLY, NEVER ON THE CARD
  • Money. All of it. Revenue, MRR, contract value.
  • Reply rate, close rate
  • Churn (your "rate at which clients leave" — real, but a checkpoint metric with zero data until client #2)
  • Client ROI

Your Round 2 picks — close rate, churn, client ROI — all live here. They are real business metrics and terrible daily metrics: other people control them, and they are denominated in the Supervisor's currency.

RETROSPECTIVE LINKING — HOW ORPHANED REWARDS GET ADOPTED When a delayed win lands — reply, booked call, signed client — open the log the same day and trace it backward in writing to the specific reps that caused it: "Client X ← call 12 Sep ← audit 3 Sep ← touch #147, 28 Aug." Five minutes. This manually performs the credit assignment your brain cannot do across weeks. It is the mechanism that eventually regrows the nudge. Non-optional.
EXECUTABLE The card's columns (Tab 03) already encode this split. Draw the card, and this tab is enforced by geometry — there is simply no cell for money.
06 · DECISION RULES

Standing if-thens. No decision is made in the moment.

Read once now. From then on these fire automatically — the whole point is that when a situation matches an IF, the THEN was already chosen by a version of you who wasn't standing at the wall.

R1
IF the urge fires to build anything — tool, site, skill file, system, document, dashboard — THEN one line on the capture pad, return to list item 1. The tooling ban lifts at client #2 signed, not before. Exempt as delivery, not tooling: the outreach script and per-prospect audits. HARD GATE
R2
IF a completed rep gets re-audited as "cheap, doesn't matter, only money matters" — THEN the 7-day reclassification ban applies: done is done. The objection gets one notebook line and zero behavioral authority. The urge to re-grade is the loop defending itself.
R3
IF "how do I know this works" fires mid-rep about the source's playbook — THEN the answer is scheduled: next Sunday, 20:00, with rep data as evidence. You cannot know from the outside; executed reps are the only instrument that measures it. Finish the rep.
R4
IF pressure arrives without a named, completable task ("you could be doing so much") — THEN the standing test: "Like what?" Silence = dismissed, zero authority, back to item 1.
R5
IF a path requires cold calls below a €10k/month baseline, or calls during sleep hours or Stelli time — THEN refused by constitution. Refusals are design constraints, not character flaws. Route around, never through.
R6
IF a cell hits 4 consecutive explicit negative replies OR 150 sends with zero positive replies — THEN the cell dies same day and the next cell loads from the queue: Cell 1 HVAC · Tampa → Cell 2 Plumbers · UK → Cell 3 Electricians · Tampa. Post-kill action pre-decided; no wall possible. KILL RULE
R7
IF a gate fails on real data — THEN change exactly one variable: offer, channel, or cell. Never two. A two-variable change destroys the information the failure just paid for.
R8
IF it's a travel week — THEN quota drops to 3 touches/day, the Sunday review runs wherever you are, nothing else changes. The business lives in the laptop; prove it monthly.
R9
IF the blank hits at the PC — THEN the 60-second wall protocol (Tab 03), never live generation. Silence → item 1. Jam → timer.

The Laws — portable beyond FJScaling

Requested addition. Same format on purpose: a law that only generates pressure without terminating in a rep would reproduce the Supervisor, so every one of these ends in something you can do in under a minute. These outlive this business.

L1
PULL IS DOWNSTREAM OF REPS. The Witcher law — your own precedent, no exceptions in your dataset. IF no motivation exists toward a domain → run the next scheduled rep anyway. The feeling arrives after competence, never before it, and it holds no vote either way.
L2
THE FEELING IS A GAUGE, NEVER A SELECTOR. It reports cache state; it does not choose actions. IF deciding what to do next → read the list, not the feeling. An empty gauge in a new domain is normal, not a verdict.
L3
DECISIONS AT PLANNING TIME, EXECUTION AT EXECUTION TIME. Your planner works; your live selector doesn't. IF a decision surfaces mid-execution → park it on the pad, decide it tonight at shutdown. The two states never mix.
L4
SCORE ONLY WHAT YOU CONTROL. IF a metric depends on another person's behavior → it goes to a checkpoint, never the daily card. Daily scores exist to write value tags; only controlled actions can carry them.
L5
DONE IS DONE FOR SEVEN DAYS. IF tempted to re-grade finished work → banned; log the urge as one line. Completed work keeps its reward long enough to accumulate into tags. This law is the direct counter to your documented devaluation loop.
L6
A LOGGED FAILURE IS A COMPLETED REP. IF an attempt fails → mark it anyway. Skill runs on volume; quality is a lagging output of volume. Same law as year one under the barbell.
L7
ONE VARIABLE AT A TIME. IF something isn't working → change one thing, measure, then the next. Anything else is superstition with extra steps.
L8
BUILD ON THE FLOOR, NOT THE CEILING. IF planning capacity → use the worst honest week (1h/day survived sickness, extra work, and life; 2h didn't). Systems built on ceilings collapse on contact and hand the Supervisor evidence.
L9
ENTRY INSPECTION BEFORE ANY NEW DOMAIN. IF entering anything new — venture, field, project → run the four-input check first (rep? feedback <48h? seeded source? mastery frame?). No rep, no entry. The wall is predictable, therefore pre-emptable.
L10
PRESSURE WITHOUT A TASK HAS NO AUTHORITY. Any voice — internal or external — that demands more without naming a completable action is an alarm without a map. IF it fires → "Like what?"; no answer → one line, pad, dropped, back to item 1.
L11
EVALUATION IS AN APPOINTMENT. IF the urge to judge progress fires → it waits for Sunday 20:00, where it gets full skeptical authority and real data. Between appointments, the auditor is off duty.
L12
THE ALARM NEEDS A BOUNDARY IT CAN LEARN. IF work thoughts intrude after "shift over" → one written line, dropped. The boundary is taught by repetition, not by winning arguments with the alarm.
EXECUTABLE One read, now. Then the rules run the moments so you don't have to.
07 · ASSUMPTION REGISTER

What this document does not know

Every INFERENCE and UNKNOWN, with the cheapest test that resolves it and a deadline. This tab exists so the document never hardens into dogma — when a test resolves, the register updates and the document bends to the data.

ClaimStatusCheapest testResolve by
HVAC is the right first trade (your stated doubt is logged)INF150 sends into Cell 1; R6 decides9 Aug
Tampa as the metroINFSame test. Free to overwrite any time before the first send — after that, the data owns it9 Aug
True send throughput ≈ 55/weekINFCount the card19 Jul
Reply rate 1–3%INFMeasured at 150 sends9 Aug
Close rate 10–20%INFFirst 10 sales conversations11 Oct
"3 focused months of SEO produces Top 3" (delivery confidence)INFClient #1's 90-day result+90d after signing
Capacity ceiling = 4 clientsINFReal delivery hours logged at client #2post client #2
A trustworthy operator with verifiable results exists and is selectableUNKDiligence week 13–19 Jul; fallback = least-bad of top 319 Jul
Demand exists in Cell 1 at these pricesUNK150 sends; you have zero witnessed demand and refused to speculate — correct9 Aug
After-tax math on €6k/mo (OSVČ)UNKOne calculation, 30 minutes9 Aug
Churn rateUNKNo data possible until client #2; parked
Deadline realism (12 Apr 2027)INFRecomputed from measured rates at day 309 Aug
EXECUTABLE Next resolution due: 19 Jul (throughput + trusted source). It happens inside a review that's already on the calendar. Nothing to do before then except send.
08 · DAY ONE

Executable by Sunday 12 Jul, 23:59. Total: under one hour.

No step below builds anything. Step 0 is the scoreboard (paper, two minutes — feedback input, not tooling). Steps 1–3 are the first rep. You confirmed this rep in Round 3, verbatim: "Yes I confirm."

STEP 0 · 2 MIN — THE CARD Take one index card or torn piece of paper. Write across the top: DATE | TOUCHES | XP | NOTE. Put it where the laptop lives. Input 2 just flipped LIVE.
STEP 1 · 30 MIN — THE LIST OF TEN Open Google Maps. Search HVAC repair Tampa FL. Open the local results list. Skip the top 3. For the next 10 businesses, write on paper or in Notes — per business: name · position you found them at · their review count vs the #1's review count · one visible flaw (missing categories, no photos this year, owner not replying to reviews, no posts). That flaw is the "specific claim about their pain point" the rep requires. This list is prospecting, part of the sitting — not a project.
STEP 2 · 15 MIN — WRITE TOUCH #1 Open email (or their contact form / Instagram DM if no address exists). Write to business #1 using this skeleton — your words, their numbers, four lines, no template smell:
Subject: [Name] — you're #[N] on Google for "AC repair Tampa" Searched "AC repair" in Tampa — you're sitting at #[N] while [competitor] holds the top spot with [X] reviews to your [Y]. Noticed one thing on your profile that's costing you calls: [the flaw, one sentence]. I put together short breakdowns showing exactly what's keeping companies like yours out of the top 3 — takes you 3 minutes to read, costs nothing, no call needed. Want me to send yours? — Filip
STEP 3 · 1 MIN — SEND Send it. Mark one tally and 1 XP on the card. Rep #1 of the business is complete. Whatever happens next is not part of the rep — a reply is tomorrow's list item, handled per the funnel in Tab 02.
STEP 4 · 5 MIN — CALENDAR Create five entries: (1) recurring — Sunday 20:00, "Review — read Tab 04, write verdict"; (2) Mon 13 – Sun 19 Jul, daily 20 min, "Source diligence"; (3) 9 Aug "Gate 2"; (4) 13 Sep "Gate 3"; (5) 11 Oct "Gate 4 — client #1 fork."
STEP 5 · TONIGHT — FIRST SHUTDOWN Run the ritual (Tab 03): score the card, capture the loops, write tomorrow's 3 reps (suggested list for Monday: touch #2, touch #3, one diligence session), say "shift over." Day one ends at the boundary, like every day after it.
✓ FOUR INPUTS INSTALLED & VISIBLE ✓ FIRST REP <24H, NON-INFRASTRUCTURAL ✓ KILL CRITERIA CONCRETE ✓ ALL CLAIMS TAGGED ✓ ZERO LIVE DECISIONS REQUIRED ✓ TOOLING GATED AT CLIENT #2