The Engine.
The hardest media, made to pay.
Most ad money buys reach and hopes. This buys nothing until a stranger hands over a card. I built both halves of the client’s machine and run them: the engine that bids on the dearest attention there is, and the store it points at. Reach gets no vote. The click is the only score, and the trust built into it before it costs a cent is what makes it pay. Built to clear $0.90 a click against a $0.45 field.
Paid nothing until a stranger hands over a card.
The brief was one line: build the affiliate operation from zero and run it. So I did, and I still do. Affiliate is paid per result, not per post. You don’t own the product, you don’t hold stock; you send interested people to someone else’s, and you earn only when one buys. The client sees every number, the good days and the bad. Most people in this business would rather you didn’t look that closely.
Paid for attention, whatever happens next. The post runs, the fee is owed, the sale is someone else’s problem.
Paid only on the conversion. No sale, no fee. The whole job is to turn attention into a card handed over.
Earnings per click: commissions divided by clicks. Over $0.50 is strong, under $0.10 is weak. This operation is built to clear EPC near $0.90 against a $0.45 field.
EPC benchmarks directional · aggregator field data 2025. The $0.90 figure is representative of the operation, the bar campaigns are built to clear, not an audited account.
The instrumented US model doesn’t port to Greece. So I re-price the whole playbook before a cent goes out.
Affiliate is a mature, measured American industry built on one thing the Greek and European ground is built to resist: a clean, prepaid, instantly-attributed path from click to sale. Cash on delivery, a statutory return right, consent law that clips the signal, VAT on every import. Copy the US playbook here and it quietly falls apart. The edge is reading which market you’re bidding into before you spend a cent.
Greece leads Europe on cash-on-delivery: around 85.6% of stores offer it. The money isn’t captured until the box is in hand, which breaks the prepaid, card-on-file, instant-attribution US model.
A statutory 14-day right of withdrawal, no reason needed, clock starting on delivery, extending to twelve months if the seller fails to inform. Returns aren’t a CX choice; they’re the law.
GDPR consent clips the click-to-sale signal. You optimise on the path the law leaves you, not the whole path. The map arrives partly torn.
Bid into the most expensive attention there is, and close every leak before the budget scales.
Four moving parts run the read: the model, the always-on paid engine, the creative tested in volume, and the daily read that moves money to what converts. On pay-on-results a leak isn’t a soft metric, it’s spend that never becomes a payout. So every stage that can leak is named and sealed before the budget scales.
Products promoted to convert, paid on the sale, scaled across four markets. The income is the proof.
Run as one system across markets. Audiences built per region, budget shifted in real time to what converts. Disclosure done to the letter, so the spend is never one complaint from being shut off.
Offers turned into motion and short-form reels. Hooks tested in volume, winners scaled, the rest killed fast.
Attribution, ROAS and the funnel, read daily. Spend moved to what worked, not to what looked good.
Funnel proportions are illustrative; the checkout and return leaks are real, measured US behaviour (Baymard, NRF). Every stage the operation was designed against.
In Europe you optimise on ~70% of the signal and know which 30% is missing: consent law clips the click-to-sale path, and you bid into the part you can’t see.
There’s never one ad. The job is a loop fast enough that the winners surface before the budget does.
A content shop ships a handful of assets a month and calls it a campaign. A performance operation ships ten times that, on purpose, because most are supposed to lose. Forty to fifty hooks a week, the winners scaled, the rest killed fast, refreshed on a calendar before each one fatigues, never after.
A 10× volume gap is the spine of it. Content-shop output vs performance-creative output for one client (Darkroom 2025); 40–50 hooks a week, 3–5 expected to win (Take Flight 2025); roughly 1–3 of every 10 creatives becomes a true winner (AdManage 2026).
Refreshed on a calendar, never on a whim: a top-of-funnel hook wears out in three to four weeks, so fresh creative ships every seven to ten days, on the numbers (CTR down 20%, frequency past 4, CPA up 15–25%), not taste. Field benchmarks, the tempo this engine runs to, not a count of my own output.
Every paid click, made to pay twice the field.
One line decides everything on pay-on-results: what it costs to win a buyer against what the sale pays back, read daily. The field earns about forty-five cents a click. This engine is built to clear double, and it’s checked every morning.
Non-Amazon offers only, priced over a hundred dollars a sale on a thirty-day cookie. The math that lets a bought click pay back. The field has moved with it: recurring commission now beats one-time for 62% of affiliates.
US Meta CPM runs near $20, up to $32 on strict audiences, the dearest attention on earth. The click has to convert hard enough to clear that, every time.
The average affiliate click converts around 2.1 to 2.5 percent. The trust built into the click before the spend is what doubles what it pays.
Representative figures, not audited accounts. The bar the campaigns are built to clear. Field EPC ~$0.45, conversion ~2.1–2.5% (Scaleo, wecantrack 2025).
The money doesn’t spread, it stacks. Being in the ten percent is the whole job.
Affiliate looks like an open field. It isn’t. The returns follow a power law: a thin top decile earns almost everything, the long tail next to nothing. This isn’t built to play the field. It’s built to be the exception the field pays for.
Ninety percent of the money goes to ten percent of the operators. The whole case is an argument for which ten.
Reach is free. The only thing worth paying for is the click that pays back.
The trust is built into the click before it costs the account a cent, and the return is priced in before this side spends a euro. One plan runs both sides of that seam, where everyone else leaks. The store it all lands on is the Dropshipping case .
No pitch. No deck. Just a straight read on what you’ve got.