Guessing why CPA creeps up as you scale
A diagnostic you run yourself — fatigue vs. auction vs. funnel, settled by the numbers
Campaign Lab is a hands-on implementation cohort led by Marcel Sattler. The frameworks behind $100M+ in native spend — applied live to your campaigns, in a room capped at 30 people who are actually running traffic.
It is application-only on purpose: the room is the product, and one wrong seat costs everyone. Founding terms — $1,497, once. This format never runs again.
▸ What this is
Courses hand you videos and wish you luck. Campaign Lab works the other way around: you bring real campaigns — live or planned — and over eight weeks we build, review and fix them together, in session, on screen.
The teaching exists to serve the reviews, not the other way around. Every framework and template you receive gets applied to your account the same week — offer math, campaign structure, advertorials, tracking, scaling decisions.
Thirty seats is not a marketing number. It is the most campaigns one operator can genuinely review in eight weeks without the reviews turning into theater.
Generic examples, watch alone, apply someday
A thousand members, no one accountable for your results
Your campaigns, reviewed live, by the person who spent the $100M
▸ The room
This is for you if
You run a DTC brand or lead-gen offer and you're diversifying beyond Meta — with real budget behind the move.
You're a media buyer already spending on Taboola, Outbrain or MGID and you suspect your funnel leaks more than your dashboard admits.
You're an affiliate with a proven offer, moving into advertorial funnels you own.
You're in-house and accountable for a native line item — and tired of learning by burning it.
This is not for you if
You're collecting content to watch someday. Every seat carries live-review time — an unused seat is a wasted one.
You don't have an offer yet. The Lab sharpens campaigns; it can't invent your product.
You need a guaranteed ROI figure before you commit. Nobody honest can promise that, and we won't start now.
You can't protect roughly four hours a week for eight weeks. The work is the mechanism.
▸ The transformation
Not “inspiration.” Specific capabilities you didn’t have on day one — installed on your own account, not a demo.
Guessing why CPA creeps up as you scale
A diagnostic you run yourself — fatigue vs. auction vs. funnel, settled by the numbers
Judging funnels on dashboard proxy metrics
Tracking that reports the truth, and unit math you can defend to anyone
Learning by burning budget on your own
Every major decision reviewed — by the operator and a room of 30 peers — before the spend
▸ How it works
The core of the Lab. Your campaign on screen, taken apart and rebuilt in session — structure, creative, funnel, numbers. This is the thing a course can never give you.
One focused teaching session per week, following the 8-week arc below. Built to be applied the same week, not archived.
Open Q&A between sessions. Bring the thing that broke on Tuesday instead of saving it for a course module that never comes.
The working documents behind the agency's own campaigns — offer document, launch checklist, advertorial structure, tracking setup. Yours to keep and reuse.
Capped on purpose. Thirty operators at a similar altitude, in a private group, for the duration and beyond — signal stays high because the door stays shut.
All sessions and reviews recorded. Miss a week, keep the material — including everyone else's teardowns, which is where half the learning hides.
▸ The 8 weeks
Most native campaigns fail before the first ad is written. The arc follows the order that decides outcomes — every week pairs one working session with live reviews of your build.
The one-page Offer Document everything else is written from — plus the unit economics that decide whether native fits your margin at all.
Account structure, budgets, testing design. The skeleton that separates a diagnosable campaign from an expensive mystery.
Angle research and the headline-to-hook chain, written from the Offer Document — not from a swipe file.
The asset native lives or dies on: structure, congruence with the ad, and passing platform review the first time.
Event filters, server-side setup, and the proxy-metric traps that make losing campaigns look profitable.
Going live, spend thresholds, and reading early data without panicking — or celebrating — too soon.
The optimization discipline: which numbers trigger which decision, and how to stop negotiating with losers.
Budget ladders that don't break CPA, and the written plan you leave with — for your account, not a generic one.

▸ Who’s running it
Marcel has managed over $100 million in native spend since 2015 — for his own brands and for clients — across Taboola, Outbrain, Teads, MGID, Yahoo Native, NewsBreak, Mediago and RevContent. He runs the agency behind this site and the campaigns behind its case studies.
Campaign Lab is the first time that review process — the same one the agency charges retainers for — is open to operators running their own accounts. It’s capped at 30 because he reviews every campaign personally.
▸ Applying
No payment to apply. No call with a “closer.” You write, Marcel reads, you get a straight answer.
Ten minutes, one form. Honest answers beat polished ones — the form is how we protect the room, not a sales gate.
Marcel reads every application personally. You hear back within three business days — either way.
Accepted: you get the payment link and onboarding. Not a fit: we say so straight, usually with what we'd do instead.
The cohort begins when the room is full. Founding members get the dates and the private group first.
▸ Questions
▸ Founding Cohort · 30 seats
After the Founding Cohort, this format is retired. What follows is a $2,997-per-year membership — founding members keep preferred terms for life. That’s the whole scarcity story: it’s structural, not a countdown timer.
8 weeks · 30 seats · $1,497 · application only
Apply for the Founding CohortApplying costs nothing and commits you to nothing. If it’s not a fit, you’ll hear that plainly — with a pointer to the free material instead.