private · UCAT Engine · Jul 2026
How we actually get customers.
Auctions, CPC, and why we don't bid against Goliath.
The acquisition machine — how attention gets bought and sold, what a customer costs, and how a £0 outfit competes with companies that can outspend us a thousand to one. Real numbers throughout, including ours.
1 · The funnel — every business is these four numbers
Someone sees you (impression) → some fraction click (CTR, click-through rate) → some fraction of those buy (conversion rate) → each sale nets revenue. Multiply it out and you get the only number that matters: CAC — customer acquisition cost — what one paying customer costs to find.
Ours: a sale nets ~£77. So the standing rule in our plan is brutal and simple: if a channel's CAC goes above £40, the channel dies. No feelings, no "one more week". Half the sale price is the ceiling, because refunds, tax and time have to live in the other half.
2 · The auction — how a click gets priced
Ad slots aren't bought at a fixed price — every single impression is auctioned in real time. When someone opens TikTok or Googles "ucat resit", the platform runs an instant auction between every advertiser who wants that person. You bid what a click is worth to you; the winner generally pays just enough to beat the runner-up; platforms discount effective prices for ads people actually engage with (higher relevance = cheaper clicks — the platforms sell attention, and boring ads waste their inventory).
That last bit is the poor man's edge: a better ad literally buys cheaper traffic. Which is why we scraped what's winning before filming anything — creative quality is the one auction lever that doesn't need a budget.
3 · The worked example — why CAC decides everything
| Scenario | Cost per click | Landing conversion | CAC (cost ÷ buyers) | Verdict vs £40 rule |
|---|---|---|---|---|
| Good ad, warm audience | £0.80 | 3% | £27 | ALIVE — scale it |
| Average ad | £1.20 | 2% | £60 | DEAD |
| Great ad, cold audience | £1.50 | 4% | £37.50 | alive, barely |
| Head-term Google ("ucat") | £4–6 | 3% | £130–200 | never viable at £79 |
Same product, same price — the spread between profit and bonfire is entirely in two percentages. This is why nobody sensible "just runs ads": you test small, measure CAC, kill fast, scale only what's under the line.
4 · Why we can't outbid the incumbents — and don't need to
Here's the asymmetry that shapes our whole strategy. Medify's subscriber is worth up to £210 a year, and their renewal model means one customer can pay twice. Ours pays £79 once. In an auction for the same click, they can rationally bid more than double what we can — on the head terms, Goliath wins every time, forever.
So we don't fight there. Our plan attacks where bids don't decide:
- Organic content — the auction doesn't exist; the best video wins. Free.
- Long-tail search — "ucat resit", "ucat score for Sheffield": tiny volume the giants don't bother bidding, high intent, £1-ish clicks.
- The wedge — re-sitter mode is a product they don't have, so our ad isn't substitutable by theirs even side-by-side.
- Trust assets — sourced pages, the published before/after, testimonials. These raise conversion %, which cuts CAC without touching bids. (Look at the table again: conversion is half the equation.)
5 · The intelligence loop — we look before we spend
Today's example, real numbers: we pulled the UK's top-performing TikTok ads by click-through rate through a scraping platform — 81 ads of live competitive intelligence, total cost 8.4 cents (from a $5/month free allowance). First finding that changed our plans: the median winning ad is 18 seconds — our drafted 30–40s scripts are getting shorter cuts before a penny is spent making or boosting them.
That's the loop: scrape what wins → shape our creative on evidence → post free → measure our own numbers → only pay to amplify what already proved itself. Amateurs film what they like and pay to find out it doesn't work. The scrape costs pennies; the lesson it replaces costs hundreds.
6 · Spending like an operator — the ladder
- Stage 0 (now): £0. Organic + forums + the free scrape loop. Target: first sales, 200 emails, one post that provably converts.
- Stage 1: unlocked by £250 of revenue. Up to £150 of tests — boost the proven post, small parent-targeted trial. Kill rule live from day one.
- Stage 2: unlocked by £1,000. ~£500 scaling whatever survived, plus long-tail search.
- The principle: the business buys its own growth out of earnings — the downside is capped at zero by construction. Know the number, cap the number, kill what breaches it.
7 · The creative bench — what we'll actually run
Honest provenance: the fifteen scripts were drafted from the product brief and known creative patterns — then the first real scrape (28 Jul) made its first correction: median winning UK ad = 18 seconds, so everything gets a shorter cut. Two more data passes are queued (the education-vertical pull and the organic UCAT scrape); scripts that don't survive contact with real data get culled. That's the section-5 loop working — the bench is a draft the data keeps editing, not a finished thing.
| Archetype | Example hook from the bench | Lane |
|---|---|---|
| Receipts / before-after | "This £70 piece of paper beats every UCAT prep site" — the score report, on camera | Dad — filming first |
| Contrarian myth-bust | "Stop doing UCAT questions" — the volume-moves-nothing evidence | Dad |
| Build-in-public | "My son bombed his UCAT. Rebuilding for the re-sit — week 1" + a Sunday update every week to test day, score published whatever it is | Dad — the spine |
| POV / relatable pain | "POV: you got the TOP band… and your course drills it anyway" · "3,000 questions in and you still can't answer: am I ready?" | Yours, if you want them — post-24 Sep |
| Checklist / urgency | "16 SEPT — 24 SEPT — 15 OCT: the three dates that decide applications" | Dad, doubles for parents |
- Cadence: 3 posts a week (Tue/Thu/Sun), batch-filmed Sunday mornings, phone + window light + auto-captions. Nothing polished — polished reads as advertising.
- Your five (parked till after your test, your call, your cut): the re-sitter unfair-advantage piece, the Band-1 skip, the "am I ready" answer, the pacing-problem line, and the big one — your before/after on results day.
- Every script obeys the house rules: no score promises, no invented numbers, £79/no-subscription is the only hard sell. If you think any hook is cringe, say so — that's literally the consumer critique we want.
The money page is here · the product case is here · back to HQ. Numbers on this page: our real unit economics, our real scrape (28 Jul), platform mechanics as publicly documented; example CPCs/conversions are illustrative ranges, not promises — which, you'll notice, is the house style.