private · UCAT Engine · Jul 2026
The money.
P&L strategy for a business that can't lose cash.
The accounts view of the Engine: what it costs, what it keeps, what it could make — and the construction trick that makes the downside zero. Real figures where they exist; anything projected is labelled a scenario, never a forecast. House rule.
1 · The cost side — read it and laugh
| Line | Cost | Note |
|---|---|---|
| Hosting (site + app + these pages) | £0/mo | Cloudflare free tier |
| Payments (Stripe) | £0/mo | fees only when money moves: ~1.5% + 20p a sale |
| Email platform | £0/mo | free to 250 contacts — paying is a success problem |
| Ad intelligence (the scraper) | ~£0 | 8.4 cents spent to date, $5/month free allowance |
| Content production | £0 | a phone, a window, £20 mic at most |
| Build cost | sunk, £0 cash | already done |
| Total to reach first revenue | £0 | optionally £10 domain + £20 mic |
Most businesses carry a fixed-cost stack — premises, staff, stock — that bills them whether customers show up or not. Here, the entire fixed-cost base of a national product is a rounding error. That single fact shapes every other decision on this page.
2 · Unit economics — what each sale actually keeps
- £79 sale → Stripe takes ~£1.39 → £77.60 net. £149 tier → ~£146.56 net.
- We budget refunds at 8% (the 14-day no-quibble promise is a trust asset — it costs a little and converts a lot), so call it ~£71 kept per £79 sale, ~£80 blended across tiers.
- Marginal cost of one more customer: zero. No stock, no seats, no inventory. Gross margin ~98% — for orientation, supermarkets run ~3% and most bricks-and-mortar businesses live between 5% and 20%.
- Tax set-aside ~25–30% of net (sole trader). Your share, when agreed, comes off the top of the cycles you front.
3 · The spend strategy — profit buys growth, never hope
The whole P&L strategy in one rule: the business only spends money it has already earned.
- Stage 0 (now): £0 — organic content, forums, the 8-cent scrape loop. Job: first sales + proof of which post converts.
- Stage 1: unlocked at £250 banked → up to £150 of paid tests. Kill rule from Module 2: CAC over £40, channel dies.
- Stage 2: unlocked at £1,000 → ~£500 scaling only what survived, plus long-tail search.
- January onward: the season push, sized by what the autumn proved — never by optimism.
The opposite of the classic startup failure mode — capital first, demand assumed. Downside capped at zero by construction; the only thing genuinely at risk is time.
4 · Revenue — the honest scenario table
Scenarios, not forecasts. ~£80 blended net per sale. The market is seasonal by design: bank the summer, run lean in autumn.
| Phase | Quiet | Working | Strong |
|---|---|---|---|
| A · This window's tail (now → 24 Sep) | ~£400 | ~£1,200–2,400 | ~£4,000+ |
| B · Early cohort (Oct–Dec: next year's sitters starting early + school licences) | ~£1,200 | ~£4,000–8,500 | ~£13,000+ |
| C · The season (Jan–Sep '27: ~41k fresh sitters, proof assets live) | ~£4,000 | ~£17,000–35,000 | ~£60,000–90,000+ |
- Phase A's real job isn't the money — it's funding Stage 1 and proving the rails work.
- Phase C is the main event, and it repeats every year: 2028's cohort, then 2029's…
- What decides the row: the 24 Sep result and its story · whether the 3-posts-a-week lane actually runs · the January assembly landing on time · schools. Not the product — the product's built.
5 · The P&L lesson under all of it
Every business is an answer to one question: what do you have to spend before you're allowed to earn? This one answered: nothing — so its only real costs are attention and discipline, and its only way to fail is to stop showing up. That's why the plan protects two things above everything: your 90 minutes a day (the proof engine), and the weekly content cadence (the distribution engine). Both are time, not money. Which, it turns out, is the actual currency of this business.
6 · Show your workings — where every number comes from
The audit trail. Three kinds of number on this page: sourced facts (linked or named), arithmetic (shown), and assumptions (labelled, arguable — challenge them).
| Claim | Working / source | Kind |
|---|---|---|
| ~41,000 market, resets yearly | 41,354 UCAT tests in 2025 — official statistics, ucat.ac.uk/results/test-statistics-2025. Resets because nearly all candidates leave the market after one cycle (they get in or move on) | Sourced |
| £77.60 kept per £79 sale | Stripe's published standard UK online rate ~1.5% + 20p: 79 − (79 × 0.015 + 0.20) = 79 − 1.39 = £77.61 | Sourced + arithmetic |
| ~£71 after refunds; ~£80 blended | 8% refund allowance is an assumption (digital-product norm 5–10%; we run a no-quibble policy so we budget the top half). 77.60 × 0.92 = £71.39. Blend assumes ~85% buy £79 / 15% buy £149 → (0.85 × 71) + (0.15 × 135) ≈ £80.6 | Assumption + arithmetic |
| £0 standing costs | Cloudflare Pages free tier (current pricing page) · Stripe no monthly fee (their pricing page) · Klaviyo free to 250 contacts (their pricing page) · Apify $5/mo free credits — our own usage meter read $0.084 today | Sourced, partly observed |
| £40 kill rule | A rule we chose, not a law: ~half of net-per-sale, so a customer never costs more than half of what they bring, leaving margin for refunds, tax and time. Judgement call — argue with it | Assumption (policy) |
| Medify can outbid us (~£210 LTV) | Their published pricing at the July 2026 competitor teardown: £15/wk · £30/mo · £210/yr, auto-renewing. A bidder's rational max bid scales with customer lifetime value; £210 > £79 | Sourced + logic |
| Scenario table (Phases A–C) | Pure arithmetic on assumed sale counts — e.g. Phase A "working" = 15–30 sales × £80 ≈ £1,200–2,400; Phase C "base" = 200–400 sales + 2–5 school licences (£500–1,000 each) ≈ £17,000–35,000. The sale counts are the assumption — nobody can source a conversion rate for a product with no sales history, which is why they're labelled scenarios and the page refuses to call them forecasts | Assumption + arithmetic |
| Sector margin comparisons (~3% supermarkets; 5–20% typical bricks-and-mortar) | Typical published industry figures (e.g. Tesco's operating margin runs ~3–4% in its annual reports) — used as orientation, not precision | Sourced, approximate |
| 18-second median winning ad | Our own scrape, 28 Jul 2026: 41 GB top-CTR TikTok ads via Creative Center, median duration 18s. Cost of knowing this: $0.084 | Observed (our data) |
| Refund policy as a conversion asset | Direction (guarantees raise conversion) is standard commerce evidence; the size of the effect for us is unknown until we have data — flagged honestly | Assumption (direction sourced) |
Method note: this is the same discipline the product itself runs on — official statistics before opinions, arithmetic in the open, assumptions labelled and open to attack. If you can break one of these workings, the page gets corrected, not defended. That's the family house style now.
How we get customers: here · the product case: here · back to HQ.