One of the more thoughtful takes I've seen around the Meta ecosystem lately. Your posts consistently show a systems-level understanding of Meta rather than surface-level dashboard thinking, and that's pretty rare.
Meta Scaling Stages 2026
The four stages of a Meta account, from "$10K a month and nothing sticks" to "$5M a month and boring on purpose." Twenty-six dimensions scored per stage, six of which did not exist in the 2024 version. Andromeda made creative the targeting layer, and that retired ROAS as a maturity signal along with a third of the old advice.
- 01Four stages with real spend bands and an explicit exit test for each: $10-50K/mo (Meta isn't working), $50-150K/mo (first winner), $150-500K/mo (system builder), $500K-5M+/mo (compounding machine). Every stage names the one thing that has to be true before the next is available, so you locate yourself honestly instead of aspirationally.
- 02Six dimensions the 2024 version didn't have, all downstream of Andromeda: concept diversity, AI in the pipeline, approval throughput, creator allowlisting, account structure, and signal quality. These are where most accounts are quietly stuck.
- 03The numbers that replaced ROAS. Incremental breakeven near 2.2x iROAS at 45% contribution margin. CAC payback stretched from roughly 6.2 months in 2025 to 7.8 in 2026. Creative hit rate of 10-20%. Fatigue windows collapsed from 6+ weeks to 2-3, which is why a 30-day forward pipeline is now the minimum just to hold spend flat.
- 04Concept diversity, the row most accounts fail without knowing it. Similarity above roughly 60% triggers retrieval suppression, so 100 ads built off one visual compete as a single entity. Stage 4 runs 8-12 genuinely different concepts live. Stage 1 runs one idea and 15 headline variants and calls it testing.
- 05Approval throughput as a first-class constraint, not an annoyance. Production went cheap, so brief-to-live cycle time is the real ceiling: the same class of concept takes one week at one brand and six at another with identical team size. Includes the pre-cleared claims library that turns review into "does this use approved language" instead of relitigating truth every cycle.
- 06The measurement chain of command (MMM allocates, geo lift calibrates the MMM, MTA decides creative, platform reporting is fast and least authoritative), plus per-stage red flags and common mistakes. Including the failure mode that actually ends mature brands: boredom, and a testing budget trimmed because it looks like waste on a spreadsheet.
What operators are saying
No fluff, no flattery, but your content is really great, amazing. I recently read a couple of your blogs and it blew my mind.
Super helpful, thanks for sharing this. The 1 ad per ~$3k framing really clicked. This gives me a clearer way to help diagnose what's ads.
Lots of alpha in your newsletter. I really enjoy your content on LinkedIn. Keep up the good work.
I read some of your posts and you seem to have a very well rounded understanding of how to monitor and maintain profitability of a Meta account.
Came across your substack and love the content.
One of the more thoughtful takes I've seen around the Meta ecosystem lately. Your posts consistently show a systems-level understanding of Meta rather than surface-level dashboard thinking, and that's pretty rare.
No fluff, no flattery, but your content is really great, amazing. I recently read a couple of your blogs and it blew my mind.
Super helpful, thanks for sharing this. The 1 ad per ~$3k framing really clicked. This gives me a clearer way to help diagnose what's ads.
Lots of alpha in your newsletter. I really enjoy your content on LinkedIn. Keep up the good work.
I read some of your posts and you seem to have a very well rounded understanding of how to monitor and maintain profitability of a Meta account.
Came across your substack and love the content.
Want to reach out?
Question, idea, intro, or just a hello. Drop a note and I'll reply straight from my real inbox.