Deliverable for

Lesson of the Day PBC (HELD + The Daily Lesson)

The founder asked for a plan to $7,000 a month of profit. The studio verified every intake claim, wrote five independent plans, scored them with three judges, merged them, attacked the ten load-bearing assumptions with twenty refuters, and rewrote the plan to fold every verdict. This is the number that survived.

The number

$1,000^/mo

PROFIT · MONTH 12 (SEP 2027) · P50 · PRE-TAX · FOUNDER HOURS UNPRICED

What was asked

$7,000 / month

On the lanes the company owns today, a $7K month is a 2028 outcome, or earlier only if the gated upside ladder (§7) is observed. It is not a Sep 2027 outcome and not an October 2026 one.

P10

≈ $470

Month-12 contribution ≈ $1,920 against the tool base at observed AI-vendor spend (≈ $1,050/mo), or the base lanes stalling on the week-2 and week-8 gates.

P50

≈ $990

Contribution ≈ $1,920 − today's actual tool base $530 − a $400 buffer for CA tax, refunds, dunning, marketplace fees.

P90

≈ $1,170–2,000

$1,170 after the four cancellable tools are cut (base $350); up to ≈ $2,000 only if two of the seven ladder gates in §7 read true before month 12.

All three are estimates: their inputs are verified (print, shipping, fees, tool bills); their unit counts are targets with gates attached, because the company has one paying human ever. Interim run-rate: Oct 2026 ≈ $0 · Dec ≈ $0–400 contribution · Mar 2027 ≈ $400–1,000 · Sep 2027 ≈ $1,900–2,500 contribution.

The three assumptions the number hangs on

A funeral home, vet/crematory practice or chaplaincy will enroll families at $39 drop-shipped, or take a $625 case handed across a table.Tested 0 of 0. It is $580 of the $1,920 base and the only line that reorders by construction. If two funeral homes say yes, the date moves more than anything the studio can build.
Rulings arrive; clicks do not.Observed: 5 of 5 chat memos ratified within hours; 0 of 7 asynchronous page-actions done in 30–63 days; 6 of 6 print carts paid with the rush surcharge. The plan therefore needs one chat line, one in-session ≈ $180 print-credit load, and 2–4 hand-deliveries a month; nothing else from the founder.
The tool base is what the month says it is.$530 today, $350 after four cancellations, ≈ $1,050 at the AI vendor's observed 25-week average (four spend bursts, each ended by a card decline, not a setting). One burst erases two months of month-12 profit. A hard monthly spend limit is the cheapest hand in the plan.

Disclosure. This work was produced by LOTD Studio, an AI research studio: frontier language models ran a multi-pass adversarial workflow (independent drafts, judges, refuters, verifier). A human — Nicolette Rankin, Lesson of the Day PBC — is accountable for it and reviews what ships. It is business analysis, not legal, financial, tax or investment advice.

1 · Ground truth

Every intake claim, checked

Four ground readers (assets, economics, demand, funnel) read the company's files, ran read-only database queries and fetched live pages. Nothing was sent, bought, deployed or edited. Where the truth is zero it says zero. Personal lines (family, household cash, support, health) are replaced with neutral phrasing in this public sample; every business number is as delivered.

Claim	Result	Source
Lifetime external revenue	Two marketplace orders, both from a family member of the founder: $301.38 gross → $250.45 payout. Direct site: 0 human purchases ever; 1,194+ checkout starts are bots/monitors. CRM: 49 contacts, 0 reviews, 0 paid institutional orders.	D1 held-crm SELECT; Stripe read-only; Etsy seller dashboard, Sep 23
Live SKU prices and autopay	`fivepack` $120, `clinic25` $450, `soul` $48, four sets $35; print autopay off; 15 outreach sends/day cap in code; sends route through a transactional email provider	held-outreach worker src 37–46, 536–572; wrangler config
Print cost, 52/54-card deck (POD)	$15.71 @1 · $12.21 @10 · $11.92 @20 · $9.61 @100 (10-unit linear ramp, identical on all five games); ≥ 10 of one game → bulk queue "several weeks"	printer API `bulk-pricing`, live Sep 23; help 84/94
Print cost, 30-card deck	$12.72 @1 · $10.02 @10 estimate — sheet model reproduces the live 52-card ladder to the cent; never quoted live	HELD-30 memo §2; refuter A3
Single-deck shipping, all-in	USPS Ground Advantage $8.29 (6.59 + 0.59 handling + 1.11 insurance); UPS Ground $9.57; the only 1-deck receipt ever was $8.39. The draft plan used bare postage $6.48. Deployed code picks UPS.	printer API live quote; `tgc.js` 179–190; receipt #548804
Standard print queue	Never measured — 6 of 6 receipts paid the +100% rush; last live regular quote 8 weeks (Sep 14 cart)	receipts; press.draft.js `queue_days_est 56`
Institutional offers ever sent	0 — `offer_sent` NULL on all 49 rows; the 3 sends ever (Aug 20) were paid-review invites to three authors; 12 partner intros queued since Jul 22, never sent, and their bodies claim a print run that does not exist	held-crm SELECT (`contacts`, `emails`, `activity`)
Coordinator acceptance of a disclosed synthetic reader	Tested 0 of 0; the room player autoplays a silent synthetic presence; the door runs a generative chat; the shippable card prints no disclosure; no QR-less variant exists	room-player 5.0.0; held-kelly src; cardgen2/3
"$25 per family" incumbent anchor	One program (Tukios + Aftercare.com), $25/family for 4–5 mailed touches over a year, postage included; OGR $20 for 5–6; hospice paper spend $0.71–11.50 per family-year; nonprofit hospice Medicare margin −1.3% (2023)	tukios.com; aftercare.com; ogr.org; hospicefoundation.org; mourningdiscoveries.com; MedPAC Mar 2026 ch.10
Reorder cadence	Company's own Sep 5 model: `reorder_prob_90d 0.35`, 3 cases/yr → 2.05 cases per account-year, labelled HYPOTHESIS; the draft assumed 12 with 100% retention	held-growth/build_model_xlsx.py 54–57
Renewal report / stand codes	No `accounts` table, no `loc-` code anywhere; room counters are test traffic (61 scans from 5 IPs, Sep 16–22; 2 candles since Jul 26)	harbor D1 SELECT; grep held-api
Keepsake comparables	Personalized card decks sell at $8–20 (Shutterfly $19.99, Zazzle ~$12, Etsy $7.79–14.85); the $70–160 band is candles, chimes, gift boxes with 100+ reviews	shutterfly.com; laurelbox.com; demand report §2a–2b
Founder's hands, observed	Memo ratifications 5/5 within hours since Sep 14; per-recipient sends 3 ever; proof cart unpaid 61 days; 6/6 print receipts rushed; a requested secret paste 7 days absent. Cash on hand: low four figures, with housing due mid-October and a personal window Sep 24–Nov 5 with limited hands.	held-crm; printer receipts; secret list; founder memory (personal lines neutralised)
AI vendor spend	Overage toggle off, but a $200 overage charge attempted and declined twice (Sep 14, Sep 23); Sep 9–11 burst $1,108.87 in 67 charges; four bursts in 25 weeks, each ended by a card decline; one planning day = 18% of a weekly model window	vendor usage card; billing receipts (read-only)
Live promises the base ignores	Reviewer page: "The honorarium is $200"; pricing page: "10% of their subscription — forever … 20% bonus … cash out via bank transfer, USDC"	curl Sep 23; kelly-teaches index.js 5171, 5631
Marketplace shop, what strangers see	Unfulfillable "leather Edition" variation ×6; "Party decor" highlight; $4.95 shipping; a gift-note line the printer cannot honour; category = Greeting Cards; 0 reviews. Sep 1–23: 19 visits, 22 views, 0 orders. Ranks #2 and #4 organically for "grief support cards" — volume, not rank, is the problem	listing 4540373133; seller dashboard; sandbox browser, Sep 23
Free daily-lesson audience	≈ 53 homepage views/day; 25–79 signatures/month go anywhere past `/`; `/pricing`, `/store`, `/api/checkout` have 0 lifetime pageview rows (never instrumented); `?age=kid` returns the adult page	dailylesson D1 `pageviews` SELECT; index.js 17357; curl
Nov/Dec printed-month variety	Adult "learn" track flat (16 distinct openers / 10 themes in November; 20 of 30 begin the same way); kid and elder 29–30/30 → kid and elder print first	D1 `kelly_lessons` SELECT

Consequence carried into every section: no line in the plan rests on an observed conversion. Every unit count is a target with a gate attached, and the gates are what the plan builds.

2 · Five independent plans

Five lenses, written blind to each other

Each planner received the same ground truth and the same target ($7,000/month profit ≈ $7,316/month contribution after a $316 "needed" tool base) and was forbidden to read the others. Full texts ship in files/plans/. The "needs" column is each plan's own count of what the target month requires.

1 · The printed month as a subscription

≈ 458subscribers, month ~14

The daily-lesson corpus holds 48 distinct printed months per age band per language, so one subscriber can receive a new month for four years before anything repeats; the grief product cannot subscribe because a "month two" does not exist and cannot be written by a model. Needs 40 net-new subscribers a month from one buyer ever; ships the flattest (adult) deck first and ships it untracked.

2 · B2B institutional on invoice

≈ 16accounts on a cadence

A Medicare hospice must run a documented 13-month bereavement program per death (42 CFR 418.64(d)); coordinators already pay paper vendors $7–36 per family-year. Sell cases at the incumbents' $25/family with a standing reorder link and a quarterly counters report; the room where the synthetic teacher reads the card is the hook. First dollar 3–6 weeks out; wants ≤ $444 of sample cash.

3 · Classroom, homeschool, family

≈ 490customers · 350 print carts

Kelly is free forever; the school pays for paper: a dated month of cards per age band, a classroom edition, a class set, a school-year box. Found three defects that would have shipped a broken object (a child's QR lands on an adult-titled page; the school widget serves the adult band; 345–362 of 365 non-English kid titles carry translator chatter). $7K from this angle alone is Sep–Oct 2027 at the earliest.

4 · Personalization at a premium

≈ 190orders/mo (95 new + 95 renewals)

The printed object costs the same generic or made for one person, so a $119 "made for one person" keepsake clears $93 against $11.49 for a $35 set: 8× on identical COGS. The bespoke press was dry-run verified end to end in July; what is missing is one live private-game print, a name flag, four SKUs, an interview page and a repeat rail. Volume stays low and every order is prepaid.

5 · Marketplace-first catalog

≈ 385orders/mo (≈ 340 customers)

A gift buyer types the occasion, not the product; the category leader runs 369 re-titled listings of the same objects at $0.20 each. List the same five objects under ~60 honest occasion × recipient titles and let the QR on every card sell the second box direct. Honest frame: ~13 orders a day against a shop that has never sold to a stranger, so written as a ladder with kill criteria at day 45 and day 90.

3 · Three judges

Scorecards (1–10 per dimension, max 60)

Three judges — a skeptical CFO, an operator who would build it, and the customer — re-verified the load-bearing numbers themselves before scoring. They split their winners; they agreed on the same grafts and on the same finding that no plan reaches $7,000 before month 10–14.

CFO · winner: Plan 2 (41)

Plan	Realism	Evidence	Buildability	Time to $	Constraint fit	Repeat	Total
2 · B2B institutional	6	7	6	5	8	9	41
4 · Personalization premium	5	6	7	8	8	6	40
1 · Printed-month subscription	4	5	6	6	6	8	35
3 · Classroom juggernaut	3	5	6	5	6	7	32
5 · Marketplace-first catalog	3	5	7	5	8	4	32

Why: it needs ~53 paying accounts in a month instead of 340–490 paying humans; its buyer has a structural reason to reorder; the customer pays before the case prints. Weaknesses named: first dollar 3–6 weeks out, coordinator acceptance of a synthetic reader untested, 31% of its book is a classroom kit with no comparable. New cross-plan finding: the language axis is over-claimed (non-English kid titles cannot print). Restated the one order the company ever had: −$45.29 cash after the rush surcharge.

Operator · winner: Plan 3 (42)

Plan	Realism	Evidence	Buildability	Time to $	Constraint fit	Repeat	Total
3 · Classroom / daily-lesson month	5	7	8	6	8	8	42
1 · Subscription (month by mail)	5	6	7	6	8	9	41
2 · Institutional B2B	5	8	6	4	8	8	39
4 · Personalization premium	5	6	7	6	8	6	38
5 · Marketplace-first catalog	4	6	7	4	9	5	35

Why: same physical object as Plan 1, sold to a non-crisis buyer on the only indexed domain with the only owned audience; tightest first-dollar slice (28 h in code paths that exist); its diagnostics were re-verified in the database. Graft #1 is Plan 1's subscription rail because it is the only mechanism that takes the founder from one print click per $8–18 to one click per month. December 2026 contribution ≈ $300–1,200 against a household gap that opens Oct 1.

Customer · winner: Plan 1 (39)

Plan	Realism	Evidence	Buildability	Time to $	Constraint fit	Repeat	Total
1 · Printed-month subscription	5	6	6	6	8	8	39
2 · B2B institutional	6	7	5	4	7	9	38
3 · Educational juggernaut	4	7	6	6	8	7	38
4 · Personalization premium	5	6	5	7	8	5	36
5 · Marketplace-first catalog	3	6	6	5	9	4	33

Why: the customer stood where each of four buyers would stand (the griever, the parent, the teacher, the coordinator) and found the same object at the centre of the three top plans; the subscription is the shape that turns one buyer into a calendar. Found that the only month deck ever rendered is the flattest one in the house; that the pricing page "costs the company my trust in one screen" ("Save $-3.12", live classes that do not exist); and that the $119 keepsake copy is the best selling in the company, with zero photos because none has been printed.

4 · Synthesis

The merged plan before refutation: $7,700 contribution

The synthesizer merged the three winners: institutional cases first (sized to deaths), a $119 keepsake second, the printed month third, the marketplace as review farm, the free daily lesson as audience. It reached ≈ $7,700 contribution ≈ $7,000 profit in month 10–12 (Jul–Sep 2027) with ≈ 145 paying customers, 16 of them institutional accounts. This is the book the refuters attacked.

Segment	Offer	Units/mo	Contribution	Basis as written
Hospice / funeral / pet-aftercare case accounts	Case 25 @ $625	8	$2,320	~160 approved sends/mo → 1–3 first orders/mo; a 300-death hospice = a case a month
Annual Care Partner	$2,500/yr	6 ÷ 12	$600	conversions after two reorders
Multi-site Program	$5,900/yr	2 ÷ 12	$467	year-2 in most cases
Therapists, chaplains, grief centres, vets	Practitioner 5 @ $149	12	$720	the likeliest first dollar
Close family, funeral-home arrangement table	Keepsake @ $119	20	$1,444	$70–125 personalized-memorial band; arrangement wholesale $71
Anniversary / first holidays	SOUL @ $48	15	$338	live SKU
Parents, grandparents	Named Month + month @ $35–39	40 active	$628	10% churn → 4 new/mo to hold 40
Teachers, PTAs, co-ops	Classroom kits	9	$553	no comparable
Gift shoppers (Q4-weighted)	Year + singles	32	$628	34–61 listings; QR-loop repeats
Draft book		≈ 145 customers	≈ $7,700	≈ $7,000 profit after a $400 buffer and the $316 base

The ten assumptions the book rested on

id	Assumption (as the draft wrote it)	Weight
A1	A bereavement coordinator will put a disclosed synthetic reader in family materials when the offer is print-first and the room is opt-in.	$3,400 of $7,700
A2	Cold email at ~160/month converts at 5–10% reply → 20–30% sample→order → 1–3 first orders/month from month 3.	the 10–12-month clock
A3	Printer economics hold: 30-card ≈ $12.72; 10-tier on 25 of one game; $6.48 shipping includes handling/insurance; standard queue 2–4 weeks.	every contribution
A4	The founder's hands arrive on cadence: a 45-min day-0 sitting, $300–550 of proofs, 30 min/week of sends, autopay ratified.	everything prints
A5	The $119 keepsake sells at 8× a set's contribution with zero reviews; the printer's private per-order games work.	$1,444
A6	The reprices ($149 / $625 / $2,500 / $5,900) do not kill conversion because institutions anchor on $25/family.	$4,100 institutional book
A7	The $316 tool base holds: no AI-vendor overage, 162 build hours fit the plan, no cash honoraria, no affiliate payouts.	the definition of profit
A8	A program that buys one case reorders roughly monthly and renews annually on a counters report.	segment 1's repeat
A9	34–61 marketplace listings lift visits 5–10× and convert ≥ 0.5% with 0 reviews; first stranger order by Nov 7; 4 reviews per 20 orders.	the consumer third, ≈ $2,700
A10	The printed kid/elder month converts from the free site's 50–150 humans/day at ≥ 0.3%, churns ≤ 10%, private games tolerated at scale.	$628 + the QR loop

5 · Twenty refuters

Two attacks per assumption: one from the company's own data, one from the market

Each refuter was given one assumption, one lens, read-only access, and the instruction to break it. Verdicts: holds weakened refuted. All twenty returned refuted as stated (eleven at high confidence, nine at medium-to-high or mixed); several individual clauses inside an assumption survived and are marked. Full notes with source lists ship in files/refute/.

A1 · The disclosed synthetic reader in family materials

Claim: a bereavement coordinator will accept it when print-first and opt-in. Weight: $3,400 of $7,700.

Data lens

refutedconfidence: high (unproven) · medium (direction)

Zero coordinators have ever been asked: offer_sent NULL on all 49 CRM rows; the 3 sends ever were reviewer invites that never mentioned the reader; 12 disclosure-bearing drafts never left.
The object as built is not what the assumption describes: the room autoplays a silent synthetic presence on load, the door runs a generative chat, the shippable v1.1 card prints no disclosure, and no QR-less variant exists.
The plan's own fallback ("paper only") needs ≈ 8 h of unbudgeted build.

Corrected: "Tested 0 of 0." Segment 1 restated as scenarios: reader accepted $3,387; rejected with a no-room variant ≈ $2,300–2,600.

Market lens

refutedconfidence: medium

No instance anywhere of a hospice, funeral home or aftercare vendor placing a disclosed synthetic voice in materials handed to bereaved families; the category's leading digital product sells on being human-made.
The families served skew 65+ (61.8% of Medicare decedents 85+ used hospice), the cohort least comfortable with AI in emotional contexts; disclosure lowers trust in 13 preregistered experiments.
The door chat is a "companion chatbot" under CA SB 243; a page with a play button is not.

Corrected: institutional default = paper + quiet room; the reader is a per-program toggle switched on in writing; ask the acceptance question in week 1, not week 3. Segment 1 −$900 to −$1,100 estimate.

A2 · Cold email as the institutional engine

Claim: ~160 sends/month, 5–10% reply, 1–3 first orders/month from month 3. Weight: the whole 10–12-month clock.

Data lens

refutedconfidence: high

Measured send volume: 1.4/month (3 sends in 63 days) against 160 assumed — a 114× gap; the code caps sends at 15/day.
The list: 2 of 20 emailable rows are actual case buyers; 17 are generic inboxes.
"5–10% reply" has no source in any company file; the 2026 all-industry average is 3.43%.

Corrected: 0.2–0.6 first institutional orders/month once samples exist; first institutional order in month 5–8; 3–5 accounts by month 12, not 16.

Market lens

refutedconfidence: high

The deployed sending rail's acceptable-use policy forbids cold outreach — a risk to every transactional email on the account.
Healthcare role inboxes reply at 0.6–3%; the plan's chain at its own rates yields 0.6–1.8 first orders/month, not 1–3.
Observed throughput: 3 emails in 9 weeks, none to a coordinator.

Corrected: cold email demoted to a trickle (2–5 accounts in 12 months) from the founder's own mailbox, ≤ 15/day, named coordinators only; inbound sample requests become the primary mechanic (the category's own model); hand-delivery converts 20–40% in person.

A3 · Printer economics

Claim: 30-card ≈ $12.72; 10-tier on 25 of one game; $6.48 ship is all-in; case ship ≈ $30; standard queue 2–4 weeks.

Data lens

refuted on shipping4 clauses holdconfidence: high

$6.48 is bare postage. Live all-in: USPS GA $8.29 (6.59 + 0.59 + 1.11), UPS Ground $9.57; the only one-deck receipt ever was $8.39. Insurance is a rate (≈ 5% GA / 2.25% UPS), not a flat.
The deployed code path selects UPS Ground — so $9.57 is the book basis.
The sheet model reproduces the live 52-card ladder to the cent at all five tiers once the residual is read as the per-copy handling fee: 30-card $12.72 / $10.02 holds estimate.

Corrected: every single-deck contribution falls $1.81 (GA) to $3.09 (UPS): a 30-card at $35 direct clears $12.68 / $11.40, not $14.49.

Market lens

refuted as a conjunctionconfidence: high

25 of one game prices at the 20-step ($11.92, better than assumed) — but every ≥ 10-of-one-game order rides the bulk queue "several weeks" by rule, which the plan never says.
The standard queue is unsupported and contradicted by the only live quote on record: 8 weeks ("ships ~Nov 8") on Sep 14; even the rush estimate stretched from 1–3 days in July to 15 days in September.
Case shipping $24–32 survives as an estimate; the "Nov 30 last order for Christmas" does not.

Corrected: queue = 3–8 weeks in Q4 until measured; $39, not $35, is the only 30-card price that clears $15 on the UPS basis; Christmas cutoffs Nov 10–15 (GA) / Nov 20 (UPS).

A4 · The founder's hands arrive on cadence

Claim: a 45-minute day-0 sitting, $300–550 of proofs in weeks 1–3, 30 min/week of sends, autopay ratified.

Data lens

refutedconfidence: high

Two classes of hand-work with opposite records: one-line ratifications of a chat memo land within hours (5 of 5 since Sep 14); asynchronous page-actions are 0 for 7 over 30–63 days.
The one action she does perform — paying a print cart — has been done 6 times, every time with the +100% rush tick; a proof cart has sat unpaid 61 days.
Per-recipient sends: 3 ever ≈ 1% of the plan's cadence; 0 dashboard approvals ever.

Corrected: split day 0 into rulings (one chat message, one line back) and one in-session print-credit load ≤ $180 after Oct 1; drop the secret paste from the critical path (listings are edited through her browser, proven Sep 22).

Market lens

refutedconfidence: high

"Rulings arrive; clicks do not" — the market confirms the shop-credit autopay rail is real and documented, but no vendor removes her payment click while keeping the tuck box.
A Friday half-hour buys 2–6 replies a month, not 8–16, before the follow-ups the hands budget omits.
A personal window Sep 24–Nov 5 with limited hands; weekly hours never stated after two asks.

Corrected: one chat exchange for day 0; one credit load covering every paid cart under a per-cart cap; sends approved as a named list once a week in a session she is already in; kill rule: two weeks with no list approved → hand-delivery only. Default for anything else: 0 minutes until Nov 6.

A5 · The $119 keepsake

Claim: sells at 8× a set's contribution with zero reviews once one proof photo exists; the printer's private per-order games work. Weight: $1,444.

Data lens

refuted, both halvesconfidence: high

The three "unverified" printer steps are already settled in the company's own files: unpublished-SKU cart-add verified Sep 14; the first-order cap is moot after four paid receipts; no packing slip exists (the printer's written answer, Jul 27).
The real blockers are unnamed: an art-stale press, the print-gate law, the cameo approval waiting since Jul 29.
The company's own personalized product ($47.99, live since July) has never sold to a stranger.

Corrected: re-scope the proof from 3 h to 8–10 h with its prerequisites; the keepsake proof rides the same standard-queue cart as everything else so its photo clock starts with the rest.

Market lens

refutedconfidence: medium-high

No personalized card deck sells above ~$20 in any channel checked (Shutterfly $19.99, Zazzle ~$12, Etsy $7.79–14.85); the $70–160 band belongs to candles, wind chimes and gift boxes with 100+ reviews.
Funeral-home keepsake margin norm is 200–500%; a $71 wholesale on $119 gives the home 40% — they buy aftercare, they do not resell.
The marketplace math omits the 15% Offsite-Ads fee that is live on the shop.

Corrected: ≈ $150–400/month, not $1,444 estimate; keep the SKU as a 60-day price test with a kill gate: ≥ 2 stranger orders and ≥ 1 review, else reprice $79–89 or fold into "SOUL + your words" at $59–69.

A6 · Reprices do not kill conversion

Claim: institutions anchor on the incumbents' $25/family, not on the deck. Weight: the ≈ $4,100 institutional book.

Data lens

refutedconfidence: high (rationale) · unmeasured (effect)

No institutional buyer has ever been shown a price: 0 of 49 offers, 0 sample requests, 0 human B2B checkout starts.
The company's own Sep 5 research already concluded "$24/family is a pilot price, not a program price" and that the case "will be compared to $15 books and sub-$1 mailers."
Hospice paper spend is $5–11.50 per family-year; pet aftercare ≈ $1; two of the three target buyers have no $25 anchor at all.

Corrected: keep the reprice (per-deck margin $4.06 → $10.85; break-even at ~50% account loss) but strike "does not kill conversion" and carry institutional lines at ≤ 50% of counts (≈ $2,050) until three accounts have paid $625.

Market lens

refutedconfidence: high

Every incumbent price is per family per year, all-in with postage, for 4–6 scheduled mailings; a $625 case is $25 for one object delivered to the institution, postage and labour excluded — like-for-like $31–35 per family per touch, above every incumbent.
Tukios $25 and Aftercare.com $25 are one program, not two anchors; HFA Journeys "$36.25" is single-copy retail, bulk $6.45–11.50.
Nonprofit hospices ran a −1.3% Medicare margin in 2023.

Corrected anchor: "Aftercare buyers anchor on $20–35 per family-year, all-in, for 4–6 touches; hospices' own mailed-paper spend is $9–21." Copy on every B2B surface changes; if ≥ 10 samples land and every conversation stalls on price, test $549 before concluding the lane is dead.

A7 · The $316 tool base holds

Claim: no AI-vendor overage; 162 build hours fit the plan; no cash honoraria; no affiliate payouts. Weight: the definition of "profit".

Data lens

refuted2 clauses holdconfidence: high

The overage toggle is verifiably off — and a $200 overage charge was still attempted and declined twice (Sep 14, Sep 23).
The actual base is ≈ $530, not $316: a second AI vendor at $93 (was $20) and four cancellable tools ($67.64/mo) still charging.
One planning day consumed 18% of a weekly model window; 162 hours in 13 weeks overflows it. Live pages still promise a $200 reviewer honorarium and a "10% forever" affiliate payout.

Corrected: split money from throughput; set a hard monthly spend limit (the toggle has been "off" three times before a charge); plan 16–18 calendar weeks, not 13; base $530 until the cancellations happen.

Market lens

refutedconfidence: high

Trailing 25 weeks of overage: ≈ $213 + ≈ $535 + ≈ $2,370 + $1,224 ≈ $4,340 ≈ $750/month average, peak month $2,120 estimate.
Four bursts, each ended by a card decline, not a setting.
A 13-week build quarter at the observed cadence carries ≈ $2,000–3,500 of overage — larger than the plan's entire projected Oct–Dec 2026 contribution.

Corrected: tool base ≈ $1,050–1,100/month at observed behaviour; the contribution target for a $7K month is ≈ $8,050 until three consecutive clean months are observed. Profit is reported against whichever base is true that month.

A8 · A program reorders monthly and renews annually

Claim: sized to ~300 deaths/yr = a case a month; renewal rides a counters report. Weight: $3,387 — segment 1's structural repeat.

Data lens

refutedconfidence: high (no evidence) · medium (corrected numbers)

Every internal record is zero: no account has ever received a case, so nobody has ever reordered anything.
The company's own Sep 5 model carried reorders at 0.35 × 3 ≈ 2.05 cases per account-year and labelled it HYPOTHESIS — one-sixth of the draft's cadence.
The renewal artefact (a per-account counters report) has no schema, no code, no template, and no family has ever scanned a card.

Corrected: model each win as one case plus ≈ 1.05 expected follow-ons in 12 months ≈ $50/month per account at the $290 blend, not $290/month.

Market lens

refutedconfidence: high (cadence) · medium (renewal share)

Nobody in aftercare sells cases: every physical vendor fulfils per family, on demand, no inventory, no minimum, cancel anytime.
Corrected cadence: hospice ≈ 2–3 cases/yr (tiered use), funeral home ≈ 4/yr (113 calls), blended ≈ 3/yr, not 12; first-year renewal norm 56–75%.
At $25/family a drop-shipped deck clears ≈ $1.78 — the incumbent price point is unreachable with per-family shipping.

Corrected: sell per-family enrollment, drop-shipped at $39–45, invoiced monthly — each death is the order; keep the case only as an in-person add-on modelled at 3/yr; segment 1 at 16 accounts ≈ $1,500–2,500/month estimate.

A9 · The marketplace as review farm

Claim: 34–61 listings lift visits 5–10×, convert ≥ 0.5% with 0 reviews, first stranger order by Nov 7, 4 reviews per 20 orders. Weight: the consumer third ≈ $2,700.

Data lens

refutedconfidence: high (arithmetic) · medium (ranges)

The draft's "100–300 visits/week" is a 4–5× arithmetic slip propagated from a ground report (5/week × 5–10 is 25–50/week); granting the full lift gives 125–250 visits/month → 0.6–5 orders.
Lifetime stranger conversion: 0 on ~50–80 visits; the 0.5% figure is one vendor blog.
The live listing is defective today: an unfulfillable "leather Edition" variation, a "Party decor" highlight, $4.95 shipping, a gift-note line the printer cannot honour, a 3–5-day promise.

Corrected: 34 listings → 100–150 visits/month; 0.5–3 orders/month ≈ $5–60/month contribution. Not a lane; a measurement. Fix the defects before any clone.

Market lens

refuted as a compositeconfidence: medium-high

5–10× is the linear ceiling of adding listings; the marketplace's own search-diversity changes cap it below linear.
≥ 0.5% is the floor of the new-shop band, conditioned on "few reviews", not zero; 1–2% arrives after ~10 reviews.
At 0.5% of 125–250 visits the shop produces 0.6–1.25 orders a month, so 4 reviews per 20 orders is a Q2–Q3 2027 reading.

Corrected: first stranger marketplace order ≈ 40% by Nov 7, 75–90% by Dec 31; 0.2–1.5 orders/month through December; the first-dollar gate becomes "first stranger order anywhere", not on the marketplace.

A10 · The printed month from the free site's own audience

Claim: converts from 50–150 humans/day at ≥ 0.3%, churns ≤ 10%, private games tolerated at scale. Weight: $628 + the QR loop.

Data lens

refutedconfidence: high

By the company's own product law the purchase line may appear only after a reading: of ~740 homepage signatures in 30 days, 79 went anywhere else and 25 reached a lesson — the audience that can lawfully see the offer is 1–3 people a day.
Two sub-claims are unmeasurable on today's site: /pricing, /store and /api/checkout have 0 lifetime pageview rows; ?age=kid returns the adult page.
The founder's own ratified memo prices the month at $29 and says "not a subscription"; the draft overrode both without a ruling.

Corrected: $0–60/month from the site's own audience estimate; the month is a one-off object until 20 stranger months are paid; instrument before estimating.

Market lens

refuted on conversionchurn & printer clauses holdconfidence: high (conversion) · medium (price)

0.3% is applied to the wrong population; free-to-paid content converts at a 0.62% median, e-commerce 1.4–3% — on 25–79 people a month that is 0–2 orders.
$35/$39 sits above the entire kids' mail band ($10–29) and inside the $30–50 elder-gift band.
≤ 10%/month churn is inside published bands (curated boxes 8–15%; 44% of box cancellations in the first 90 days); the printer states no limit on private games.

Corrected: kid $29 / kid named $35 / elder $39; plan for a 90-day cliff and gift cohorts that end rather than churn; the named cover as switching cost is plausible but unmeasured.

6 · The final plan

Rewritten to fold every verdict

All ten assumptions were refuted by two independent lenses. The final plan does not argue with them; it folds them. The draft's own sensitivity line — "at half that rate the target month is 2028" — turns out to be the base case, not the downside.

What survives, in priority order: (1) institutional, sold the way the category actually buys — per-family enrollment drop-shipped at $39/family invoiced monthly (reorders by construction), the $625 case only for buyers who hand the object across a table, and the $149 Practitioner 5, which every refuter left standing; the institutional default is paper plus a quiet room (card text, practice, sources, candle, crisis line) with the synthetic reader as a per-program toggle; (2) personalization as a 60-day price test with a kill gate; (3) the printed month as a one-off object at band prices, the subscription rail unbuilt until 20 stranger months are paid. The marketplace is $10–60/month and a review clock; cold email is a trickle from the founder's own mailbox; every consumer contribution is re-run at $9.57 all-in shipping; the tool base is $530 today; day 0 is one chat message and the only money hand is one ≈ $180 print-credit load.

The base book — only what the verdicts left standing (month 12, Sep 2027)

Pri.	Segment	Offer	Price	Contrib./unit	Units/mo	Revenue	Contribution	Basis (corrected)
1	Therapists, chaplains, grief centres, vet practices	Practitioner 5	$149	$58	4	$596	$232	hand-delivery 0.4–1.6 first orders/mo + inbound form + ≤ 60 named sends/mo at 3–6% reply
1	Funeral / vet / chaplaincy case accounts (in-person hand-off)	Case 25	$625	$300	1	$625	$300	3–5 accounts by month 12 × ≈ 3 cases/yr
1	Funeral / vet per-family enrollment, drop-shipped	Enrollment	$39	$14	20 families	$780	$280	the category's own model; a home enrolling every family ≈ 9/mo; untested 0 of 0
1	Care Partner / Program (annual)	$2,500 / $5,900	—	—	0	$0	$0	on the shelf until one renewal is observed
2	Close family	Keepsake	$119	$85	3	$357	$255	2–5/mo at 0.5–1% of keepsake-listing visits; 0 via homes
2	Anniversary / first holidays	SOUL	$48	$21	4	$192	$84	live SKU; marketplace total 0.5–3 orders/mo
3	Parents, grandparents (one-off months)	Month / named	$29–39	$10.75	12	$412	$129	post-reading pool 25–79/mo × 0.6–3% + Q4 gifts
4	Teachers, PTAs, co-ops (test)	Class kits	$39 / $599 / $249	—	6	$1,004	$467	not refuted, never tested; needs one school conversation
5	Gift shoppers (Q4-weighted)	Year + singles	$299 / $35–45	$115 / $10	7	$539	$175	marketplace 1–3 orders/mo + QR loop (unmeasured)
	Base book				≈ 57 customers · 5 accounts	≈ $4,505	≈ $1,922	vs the draft's 145 / $7,700

Profit is defined against the base that is true that month

Against actual · $530

≈ $990

$1,920 − $530 − $400 buffer. Today's bills: AI vendor plan, a second AI vendor at $93, four cancellables, hosting, domains, email.

Against needed · $350

≈ $1,170

After the founder cancels four tools and caps the second vendor at $20. Her hand, once, ≈ 30 min.

Against observed · $1,050

≈ $470

At the AI vendor's trailing 25-week average. Holds until three consecutive months show no overage receipt.

The $7K month needs contribution ≈ $7,930 against today's base. The base book is $1,920; the ladder below is the only path on these lanes.

The gated upside ladder — what must be observed to re-add each line

Gate that must read true	Line re-added	+ contribution / mo
≥ 2 of 3 coordinators answer yes / yes-if on the reader and ≥ 1 states a per-family budget with a physical-item line; 15–25 enrolling accounts by 2028	Enrollment at scale	+$1,200–2,200
One Care Partner renewal observed	Care Partner / Program	+$500–1,000
Keepsake test passes (≥ 2 stranger orders + ≥ 1 review in 60 days) and one home sells one	Keepsake to 10–12/mo	+$500–1,000
≥ 200 reading-reaching humans/month on the free site and ≥ 20 paid stranger months; churn measured < 12.5%	Month by mail	+$400–600
≥ 10 marketplace reviews; per-listing yield ≥ 0.05 search visits/day holds at 34 listings	Marketplace singles + SOUL	+$300–800
One PTA / school purchase order paid	Classroom to 9+/mo	+$300–600
The founder buys the phone number and the first program signs the line	"Call Kelly" ladder ($450 → $1,800 → $4,000/yr)	+$500–1,500
Base + ladder		≈ $5,600–9,600

Read: the $7K month is reachable only when most of the ladder has been observed. On measured channel throughput that is 2028, not Sep 2027. The two things that move the date most are the founder's 2–4 hand-deliveries a month (3–4× cold email's yield) and the enrollment offer being accepted by the first two funeral homes.

The first 90 days, week by week (≈ 150 studio build hours; her hands ≈ 1 chat line + 10 minutes of billing + one money click)

DAY 0
One chat message with defaults, one line back. Honoraria = cards; kid and elder print first; prices ($149 / $625 / $39 enrollment / kid $29 / named $35 / elder $39); scoped print autopay under a per-cart cap; institutional default = paper + quiet room; outreach from her own mailbox ≤ 15/day. Plus ≈ 10 minutes of her own login: set a hard monthly AI-vendor spend limit and confirm the renewal card.
WEEK 1
Measurement first, then the print clocks. Pageview instrumentation on every purchase surface; marketplace stats table; Monday digest gains sends/week, days-since-send, first vs second cases, rush flag, vendor-window %. Day page honours the age band server-side. Kid/elder November and December renders. Printer games by API with live bulk-pricing reads; the standard-queue proof cart ≤ $180, built, never rushed. Reprice constants; inbound sample form replacing three mailto: links; professional-page copy truth (4–8-week lead time, no packing slip, quiet-room default). Marketplace defects fixed before any clone. Honoraria and affiliate promises taken down.
WEEK 1 (money)
The only money hand: after the Oct 1 household inflow and only if mid-October housing clears, one in-session print-credit load ≈ $180 (budget 2× until a receipt reads urgent=0). Until it exists, every physical line is $0 throughput.
WEEK 2
Money can arrive by invoice or enrollment. Discard the 12 stale partner intros; stage v4 bodies (paper + quiet room, two coordinator questions: "what do you pay per family today, for how many touches?" and "how many families a year get a physical item?"); first 10 named coordinators as drafts in her mailbox. No-room card variant (no QR, plain tuck). accounts table and per-program stand codes before the first sample ships. Invoice/PO rail → per-family "resource provided" record. Per-family enrollment rail (form or monthly CSV → one print cart per family → monthly invoice). Scoped autopay as ruled.
WEEK 3
Prospect list of named coordinators, 60 → national; sample kit that tells the truth (silent presence on scan, reads on tap, crisis line first, no data); bespoke live proof re-scoped with its prerequisites, into the same standard-queue cart. Proofs land ≈ week 4–8 (queue unmeasured; last quote 8 weeks).
WEEK 4
Depth, not conversion, on the free site: route / to today's reading in the kid band as the depth test; cards page; quiet post-lesson line; homeschool and classroom pages with a PO form. Marketplace wave 1: the four written clones → 10 listings, measured 30 days; photo 1 = the founder's photograph of the proof. A single "first dollar" folder for every hand she has.
WEEKS 5–8
The renewal artefacts the category actually uses: a name on the cover and gate zones; a per-order greeting template in her voice staged for her ear; the door's third-card and day-25 lines; review ask pointed at direct orders and, on funeral-home stand codes, at the home's own listing; class-set one-pagers; reorder link per account with an approval-gated nudge and a standing-order option; Year of Lessons SKU; marketplace wave 1b (20–25 distinct occasion pages) only if the 30-day gate passes.
GATED
Not scheduled, only earned: subscription rail (19 h) after ≥ 20 paid stranger months and two consecutive monthly in-session cart payments; quarterly counters report after one account has a non-zero counter; phone-line go-live when she buys the number; marketplace wave 2 after the first stranger order.

Kill criteria — read from data, not from asking

Week (date)	Gate	If missed
1 (Sep 30)	Day-0 chat line returned; AI-vendor spend limit set	Nothing prints; every physical line's December floor = $0
2 (Oct 7)	≥ 1 print receipt with `urgent=0` and its ship date read; ≥ 15 emails sent from her mailbox on ≥ 1 day or ≥ 1 hand-delivery; marketplace stats has a first row; 12 stale intros discarded	No receipt → printed lanes dormant until Nov 6, December physical = $0; queue ≥ 6 weeks → order December objects by Oct 10–15; < 15 sends → institutional is a 2027–28 line
4 (Oct 21)	≥ 3 recorded coordinator answers; ≥ 2 hand-deliveries; 10 listings live with 0 defects; ≥ 100 marketplace visits in 30 days with ≥ 1 add-to-cart; invoice + enrollment rails live	0 answers → the block is samples-in-hand, not copy; < 100 visits → titles / photo 1 / price, not count
6 (Nov 7)	Proofs landed and photographed; first stranger order anywhere is a signal (≈ 40%), not a gate	No stranger order → the object, photo or price is wrong, not the listing count
8 (Nov 21)	≥ 2 of 3 written coordinator answers yes / yes-if on the reader; ≥ 1 states a per-family spend; ≥ 60 named sends with follow-ups; free site ≥ 200 reading-reaching humans/month and ≥ 1 site-sourced order	Reader "no" → paper + quiet room permanent; every answer "no budget line" → hospice to quarterly cadence, drop-ship retired for hospices (kept for funeral/vet); site miss → parents row stays ≤ $250
10 (Dec 5)	≥ 1 paid practitioner pack, case or enrollment; keepsake listing live with its 60-day clock running	0 institutional dollars after 10 samples → hospice to maintenance; the book becomes funeral/vet enrollment + classroom + Q4 marketplace
13 (Jan 31)	≥ 3 stranger orders total; ≥ 1 review from the first 5; ≥ 1 institutional order; keepsake gate	< 3 orders → consumer lanes to maintenance; keepsake miss → reprice $79–89
Month 6 (Mar 31, 2027)	≥ $800/month contribution; ≥ 5 first cases/enrollments and ≥ 2 second orders; ≥ 2 reviews from the first 10 orders	Re-plan from measured conversion, not from this table
Month 10 (Jul 31, 2027)	≥ $1,500/month contribution	< $1,000 → grants (Echoing Green Oct 2026, DoE SBIR Jan 2027) are the household line; this plan is the product line; $7K is 2028+ or never on these lanes

Standing kills at any week: any print cart rushed · any AI-vendor overage receipt (restate that month's profit against the observed base) · any auto-sent external email · any cold send through the transactional provider · any listing that claims a review before one exists · any "3–5 days" processing promise before the queue is measured · any reader-forward institutional copy before the week-8 gate reads.

What is not known (carried, not hidden)

Demand: every unit count is a target. One paying human ever; 0 reviews; 19 marketplace visits / 0 orders in September; 0 samples ever shipped; 0 partner-intro sends ever; ~340 enterprise cold sends in Mar–Apr → $0.
Whether any program will enroll families for drop-ship — no hospice-specific per-family price for a physical object is published anywhere; per-family bereavement budgets have never been surveyed.
Coordinator acceptance of a synthetic reader — 0 of 0; the external record has caution from named coordinators and no counter-example.
Printer: standard queue time (last live quote 8 weeks); 25 of one game in one cart (mechanism verified at 10, ladder read at 20); whether an unbranded, slip-less parcel is acceptable in a bereaved family's mailbox.
The founder's rulings not yet given: all ten day-0 defaults, autopay scope, honoraria, Offsite Ads on/off, keepsake price parity.
Regulatory: CA SB 243 / NY GBL Art. 47 exposure of the door chat if a program card ever links to it (the quiet-room default removes it from the case product). Tax: CA sales tax on direct tangible sales (none collected); printer tax on CA-bound cases until a resale certificate exists.
Whether $7K profit is the household number: it is pre-tax; the corrected month-12 figure (≈ $1,000) is far below it, and the ladder is the only path on these lanes.

7 · Verifier's certificate

What "verified" means on this deliverable

HN-SAMPLE-001 · adversarial verification

PASS · 2 NOTES

Pass
Numbers reconcile. Base-book revenue re-summed: 596 + 625 + 780 + 0 + 357 + 192 + 412 + 1,004 + 539 = $4,505 (stated ≈ $4,500). Contribution re-summed: 232 + 300 + 280 + 0 + 255 + 84 + 129 + 467 + 175 = $1,922 (stated ≈ $1,920). Profit columns: 1,920 − 530 − 400 = 990 · 1,920 − 350 − 400 = 1,170 · 1,920 − 1,050 − 400 = 470. Draft book: 2,320 + 600 + 467 + 720 + 1,444 + 338 + 628 + 553 + 628 = $7,698 (stated ≈ $7,700). Judges' rows re-summed: all fifteen totals correct.
Pass
Every claim sourced or labelled. Each ground-truth row names a file, a read-only query or a live fetch with a date; every unit count and every 30-card print cost carries estimate; third-party vendor statements not independently checked are labelled as claims in the underlying notes.
Pass
No invented facts. Each of the twenty refuter verdicts, confidences and corrected values in §5 was matched line by line against the twenty notes on file (files/refute/A1–A10-{data,market}.md); no figure appears in this deliverable that is absent from the underlying notes or the ground reports.
Pass
Disclosure present on the first page, every section and the running footer of the PDF; the accountable human is named; the "not advice" line is intact.
Pass
Third-party PII absent. Family names, household cash, support and health lines are replaced with neutral phrasing; no customer, contact or prospect is named; no email identifiers or card digits are reproduced.
Note
Timing. "82 minutes" is the orchestrator's wall clock from first ground read to final file; file timestamps span 12:54–14:14 (80 minutes). Both are reported.
Note
Sample provenance. For this public sample the verification checks above were run by the publishing session on Sep 23, 2026 and are reproducible from the shipped files. For a paid job the verifier runs in an independent session that did not write the plan, and a FAIL blocks release and triggers a full refund.

verifiedRead in a named file, returned by a read-only database query, or fetched live on the date shown — and the source is named next to the number.

estimateArithmetic on verified inputs with its basis stated (a sheet model, a benchmark band, a channel's measured yield). Never a round number without a basis.

refutedA load-bearing assumption that two independent refuters could not find support for — or found contradicted — in the company's own records and in the market. The plan folds the corrected value; it does not argue.

8 · How this was made

34 agents, 82 minutes, one accountable human

Every agent ran read-only: files, read-only database queries, live page fetches, the printer's and payment provider's APIs in read mode. Nothing was sent, bought, deployed or edited. The phases ran in the order below; each phase read only the phase before it, and the refuters were forbidden to read each other.

12:541orchestratorGround truth compiled inline from the founder's intake: money today, live products and prices, standing constraints, where the evidence lives, live marketplace stats read from the seller dashboard.
12:54–13:094ground readersAssets (what the company owns that a customer could pay for), economics (verified unit costs and the fixed base), demand (who pays repeatedly for this shape of product, at what price), funnel (traffic, conversion, idle distribution assets). Every state tag verified that day or labelled.
13:20–13:235plannersFive independent plans from five lenses — subscription, institutional, classroom, personalization, marketplace — each reaching the target its own way, each forbidden to read the others.
13:32–13:363judgesA skeptical CFO, an operator, the customer. Each re-verified the load-bearing numbers before scoring six dimensions; each named a winner and the grafts.
13:441synthesizerOne merged plan with its own re-verification pass, an offer ladder, a 90-day sequence in hours, and — the part that matters — ten named load-bearing assumptions.
13:47–14:0320refutersTwo per assumption: a data lens (the company's own records, code and pipelines) and a market lens (external evidence, incumbents, benchmarks, policy). Each returned a verdict, a confidence, the evidence, and the corrected value or plan change that survives.
14:141finalizerThe plan rewritten to fold every verdict: corrected book, three-column base, the gated ladder, week-by-week sequence, kill criteria, what is not known.

The refutation record

Ten assumptions × two lenses = twenty notes. Twenty of twenty returned refuted as stated; eleven at high confidence, nine at medium-to-high or mixed (high that the assumption is unsupported, medium on the corrected number). Nine individual clauses inside those assumptions survived and are marked holds in §5: the 30-card sheet model, the 20-step bulk price, case shipping as an estimate, the churn band, the printer's tolerance of private games, honoraria-in-cards, no affiliates, and the reprice's per-deck margin. The draft's book fell from ≈ $7,700 to ≈ $1,920 contribution; nothing was added back without a gate.

What the customer receives

This page (private, tokenized, no-index), the PDF, and a files/ archive: four ground reports, five plans, three scorecards, the synthesis, twenty refuter notes with their source lists, and the final plan — every number traceable to a named file, query or fetch. One revision pass within 14 days: answer the refuters, and the final is re-run.

LOTD Studio · Lesson of the Day PBC · California · studio.lotdpbc.com · This sample is the founder's own company, published by the founder; personal lines neutralised, every business number as delivered.