The two thresholds nobody warns you about

Every operator we have ever onboarded carries the same internal narrative: "we just need a better agency". The diagnosis is usually wrong. What is actually happening is that they have crossed an infrastructure threshold their current setup cannot survive. There are two of them.

The first lives around €100K monthly media spend. Below it, almost any half-decent buyer with three creative concepts and a basic Meta business manager can keep CPA stable. Above it, creative volume becomes the bottleneck — you need 30+ assets a month, not 5, and the team that produced your first set rarely has the structure to ship the next thirty without quality collapse.

The second lives around €500K monthly spend. This is where almost every operator we meet for the first time has been stuck for two to three quarters. The bottleneck has moved from creative to everything else: account survival, audience hand-off, attribution, supply diversification, compliance review. Hire a new agency at this threshold and you will spend six months rebuilding what your last partner built — and your numbers will not move.

What scaling actually requires

Scaling FTDs past €500K/month requires four things to be true simultaneously. Most operators have one or two and assume the rest will follow. They will not.

1. A creative engine, not a creative team

At scale you need 40+ usable assets per market per month. That is not a brief, a turnaround, a revision cycle, sign-off from legal, then production. That is a pipeline: weekly concept review, parallel production, automated compliance pass, server-side delivery to ad accounts. The output is volume of variations, not polish of individual pieces. Operators who scale ship; operators who polish stall.

2. A buying surface measured in supply, not channels

"We run Meta and Google" is a buying surface of two. Real scale needs nine — Meta, TikTok, Google Search, Performance Max, Taboola, Outbrain, RevContent, MGID, push networks. The reason is not greed. It is distribution risk. When one channel suspends a verified BM (and one always will, every quarter), spend has to keep flowing on the other eight or your numbers go to zero for ten days. Diversification is survival, not strategy.

3. Server-side tracking you actually own

Off-the-shelf tracking — pixels, GTM, the platform's native postback — leaks. Conversion API on Meta, Google's Enhanced Conversions, S2S handshakes with your platform's BI: these are not bonuses. They are how you tell the algorithm what an FTD is so it can find more of them. Without server-side, the algorithm is optimizing toward the registration event, not the deposit. Your CPA looks fine for a week. Then your D7 retention craters and finance asks why.

4. A weekly hand-off cadence between media, creative, and product

The biggest reason operators plateau is not technical. It is organizational. The buying team learns something about the audience on Tuesday, the creative team finds out on Friday standup, by the time the new asset cohort ships it is the following Tuesday — a full week of compounding learning lost. At our spend level the meeting cadence is daily for buying, twice-weekly for creative review, weekly for cross-functional hand-off. Slower than that and the scale ceiling holds.

The five-channel architecture we run past €500K

For operators clearing €500K/month, here is the channel mix that has scaled most reliably across the licensed markets we operate in:

Channel% of spendPrimary role
Meta (FB / IG)30–35%Volume + retargeting
Google (Search + PMax)20–25%Brand defense + capture
Native (Taboola / Outbrain)20–25%Editorial funnel · cold
TikTok10–15%Younger demos · creative R&D
Push + Pop5–10%Re-engagement + low-CPA fill

The proportions move with the market. In DACH, native runs higher (35%) because of strict paid-social compliance review. In LATAM, TikTok and push run higher because Meta inventory is more expensive and quality varies. The framework is the architecture, not the percentages.

What kills CPA when you scale

The four most common reasons CPA collapses when an operator pushes spend past their plateau:

The 21-day rebuild

When we are brought in mid-plateau, we do not propose a six-month "discovery". The first 21 days deliver an audit, a stack rebuild, and a first compliant FTD on the lowest-risk channel cluster. Days 1–3 are an account audit and tracking inspection. Days 4–10 are server-side stack deployment, BM warm-up, and creative cohort pre-stage. Days 11–17 are first spend on a single channel cluster with daily Telegram updates. Days 18–21 are cross-channel hand-off and a written readout — continue or close clean.

The full breakdown of that pilot, day by day, lives on our Process page. We treat the 21 days as an interview that runs both ways.

The honest summary

If your operator is stuck between €300K and €700K and has been there for more than two quarters, the problem is not your agency, your creative, or your channel mix in isolation. It is that the four pillars above are not yet running together. Fix one and the next one breaks. Fix all four in the same 60 days and the ceiling moves.

That is also, incidentally, why we exist. The agency we needed in 2020 did not exist either.