Scribble Works business opportunity

Consolidates the September 11 founder conversation and read-only Cloudflare checks. Recommendations are proposals, not adopted prices or authorization to spend, launch billing, or purchase inventory.

Assessment

7/10 bootstrap opportunity: promising economics and a plausible small profitable business, with demand and retention still to prove. This rating is qualitative judgment, not a scored investment model. A small team fits the reusable-content economics. The main risk is whether parents repeatedly print and use delivered playsets, rather than enjoying a few free downloads and leaving.

The founder intends to keep the team small and bootstrap. The product includes a supplies store: affiliate links or products sold directly, such as crayons and safety scissors. The founder expects corpus growth to follow a sigmoid curve and eventually require less incremental creation. The analysis below incorporates those points.

Market scope and sizing

Initial scope: US households with children ages 3–10 seeking supplemental printable activities. School licenses and international expansion are separate opportunities. Household, not child, is the subscription unit.

The 2024 ACS five-year S0101 table supplied these indexed Census counts: under 5 = 18,779,951; ages 5–9 = 20,081,975; ages 10–14 = 21,463,610. The direct table required JavaScript; the newer Vintage 2025 single-age download was not obtained. These are not September 2026 population measurements.

Approximation: ages 3–10 = 2/5 × under-5 population + all ages 5–9 + 1/5 × ages 10–14 = 31.9 million children. Uniform ages within each band and 1.5 eligible children per household are assumptions, yielding 21.3 million households.

The narrowing inputs are unmeasured scenarios: convenient printing access 40–60%; recurring need among those households 10–20%; willingness to subscribe among that group 15–40%. Middle case: 50% × 15% × 30% gives 478,300 households and $51.7m annualized potential. These are neither confidence intervals nor measured category spending. Twelve paid months is a run-rate convention, not established retention. Supplies revenue is excluded.

Measured model-cost evidence

Read-only queries reached the Scribble Works Cloudflare gateway and production UGC D1 database on September 11.

Recent settled calls were created 20:06:59–20:42:40 UTC, with costs $0.007216–$0.030340. Excluded: two unresolved reservations totaling $0.123923 and a $1 opening allowance. These exclusions prevent reserved budget from masquerading as completed-call spend.

The gateway returned 14 retained logs total: the 12 uncached Sonnet calls, one cached zero-cost Sonnet response, and one tiny Haiku call. Current code sends cf-aig-collect-log=false, so retained gateway logs are not a complete spend history. The newer ledger records response-usage costs using configured rates; neither source was reconciled to invoices. No image calls appear in this ledger snapshot. Calls are not tagged sufficiently to separate authoring tests, customer features, or finished games.

The aggregate query was:

SELECT model, outcome, COUNT(*) AS calls,
       SUM(charged_micros)/1000000.0 AS charged_usd,
       AVG(charged_micros)/1000000.0 AS avg_usd,
       MIN(charged_micros)/1000000.0 AS min_usd,
       MAX(charged_micros)/1000000.0 AS max_usd,
       datetime(MIN(created_at)/1000, 'unixepoch') AS first_utc,
       datetime(MAX(created_at)/1000, 'unixepoch') AS last_utc,
       SUM(CASE WHEN settled_at IS NULL THEN 1 ELSE 0 END) AS unsettled
FROM generation_calls
GROUP BY model, outcome;

Cloudflare cost documentation describes estimates; gateway pricing describes free core features. Model cost is distinct from gateway/platform fees. No additional paid test generations were run for this analysis.

Corpus growth and margin

Local source inspection established that workers/daily-playset/src/deliver.js calls deterministic curateHousehold, packages existing PDFs, and sends email. src/lib/playset/curate.js makes no model call. This pass did not independently reconcile every deployed Worker version.

Separate shared library creation/review/maintenance from per-household customization, packaging, delivery and support. The earlier $0.71-per-credit model was a legacy unmeasured pack estimate and should not anchor routine library-delivery margins. The measured 1.56-cent call is not a complete game cost either: art, evaluation, and retries may add calls.

A sigmoid corpus curve is plausible but unverified. Measure demand-weighted coverage and acceptable variety, not simply game count. Additional ages, languages, themes, and seasonal needs may keep expanding the useful library.

Illustration: 180 game-page uses per child-month × fresh-generation share × assumed $0.10 fully loaded generation cost gives $3.60 at 20% fresh, $0.90 at 5%, and $0.18 at 1%. Both generation share and finished-page cost are hypothetical. Current deterministic delivery itself has no fresh-generation step.

Pricing and acquisition proposal

Preserve the founder's free manual-planning versus paid curation/delivery boundary in [[2026-08-31-product-model-rulings]], especially rulings 31 and 33. Paid value: suitable playsets, packaged and delivered on schedule. Lead with convenience and quality; show customization limits clearly without making credits the main promise.

Test $9/month household pricing against $12 for the same offer, holding retention and costs in view. Defer annual pricing until renewal behavior and costs are better understood. These prices remain proposals.

Published offers checked September 11: NewPath $29.95/year; Twinkl $7.49/month headline conflicting with a $10 starting-monthly FAQ; ABCmouse $14.99/month or advertised $45/year. Promotions and renewal terms vary. Pure resource access faces inexpensive competition.

Using Stripe's published pricing, assume 2.9% + $0.30 domestic card processing plus 0.7% Billing, and a further unmeasured $1–$3 variable cost per subscriber-month:

Formula: price × 0.964 − $0.30 − variable cost. Allowance must eventually include actual generation, images, retries, delivery and variable support. Shared library investment, acquisition, fixed overhead and founder time remain outside this contribution figure. Free-user costs need a separate measured budget. These margins are plausible scenarios, not observed business performance.

Proposed bootstrap discipline: recover CAC within roughly three months of contribution. At an assumed 8% monthly churn, three-month contribution = monthly contribution × (1 + 0.92 + 0.92²): about $15–$20 at $9, or $23–$28 at $12. These pools leave no allowance for fixed overhead or profit. Initial test targets: CAC below $10 at $9 pricing or below $15 at $12. Track creative costs and founder acquisition time separately; organic distribution is not costless. Exclude unproven store income from CAC budgets.

At 500 subscribers × $12, MRR is $6,000 and illustrative contribution is $4,100–$5,100 before fixed overhead, acquisition and owner compensation. At 1,000, annualized subscription revenue is $144,000. These are run rates, not forecasts or take-home income.

Supplies store

The store can monetize free and paid households and remove the materials barrier to using playsets. Start with materials links attached to relevant games and an easy-to-understand starter basket. Affiliate transactions can establish item demand before choosing to hold inventory.

Affiliate revenue = qualifying purchases × basket × commission. Owned-store contribution = sales − merchandise − fulfillment − shipping subsidy − payment fees − returns. Affiliate merchandise sales are not Scribble Works revenue.

Illustration only: 1,000 active households × 10% purchasing monthly × $25 basket = $2,500 merchandise volume. At assumed 4% commission, revenue is $100/month. At assumed $5 net contribution per directly fulfilled order, contribution is $500/month before fixed store overhead. No partner agreement, supplier quote or conversion evidence supports these assumptions yet. Durable scissors and replenishable crayons/paper need different purchase-frequency assumptions.

Stock inventory only when basket demand and fully loaded fulfillment contribution justify a second operational workload. Separately test whether easy access to supplies improves activity use and subscription retention.

Marketing and validation

Start with parents already printing activities and spending time choosing and assembling them. Ages 3–6 are an initial acquisition focus, not the product ceiling. Test supplemental-school and homeschool segments alongside them.

Proposed promise: Their next playset, ready to print. Games chosen around their interests and what they're practicing—delivered on your schedule. Demonstrate email → one PDF → physical games. Specific printable landing pages and Pinterest are channel experiments; success means activated and retained paying households, not impressions or saves. See [[2026-09-06-content-outreach-research-scribble-works]].

Milestones:

1. First 30 paying households beyond friendly testers: establish actual willingness to pay.
2. Three monthly cohorts: track renewals, repeated use, support burden and complete household costs.
3. Small acquisition tests: scale only channels whose customers retain and repay CAC.
4. 300–500 subscribers: verify content maintenance and service remain manageable for a small team.

Next measurement work: link every call and retry to a completed generation; distinguish authoring from customer requests; measure cost per completed customization and household-month; track library coverage, repeat tolerance, paid renewals, and store contribution. No new collection or instrumentation was implemented in this analysis.

Related work and evidence

• GitHub #149 — paid-pilot decision: existing backlog item; this note supplies preliminary evidence but does not complete it.
• [[2026-09-02-tier-line-plan]] — historical proposal; read its appended ruling 31 correction.
• [[2026-08-31-studio-charter]] — historical cost assumptions and founder-time allowance.
• Local source receipts: /Users/ray/Documents/Codex/2026-09-11/scribble-tam/sources.json and cost-store-sources.json, with rendered HTML companions. They retain calculations and bounded evidence, not a complete billing export.

Overall recommendation: continue developing a profitable niche business, preserve operational simplicity, and tie acquisition spending to observed retained contribution. Library reuse and the supplies store improve the opportunity; neither substitutes for paid retention evidence.