One campaign, no ad set controls, and delivery decisions you never made. When performance turns, “trust the system” isn’t an answer you can give a client. Sentrum reads what the automation is actually doing — layer by layer, in your own data — and names the levers you still own.
No tool can see inside Meta’s optimizer — including this one. What Sentrum does instead: read the automation’s observable behavior rigorously, and refuse to over-claim.
The Blueprint scan walks the whole account layer by layer — structure, audiences and overlap, creative mix, delivery mix, budget and pacing, signal quality — and narrates what the automation actually did in each one. Not what Meta says; what your data shows.
The comparison everyone wants is also the easiest to fake. Sentrum requires at least 10 purchases per cohort and a real spend share on both sides before calling a verdict. Below that bar, it says “insufficient data” — out loud — instead of inventing a winner.
Every finding lands on a lever that still exists under automation: creative mix and cadence, signal quality, audience inputs and exclusions, budget level. “The algorithm did it” is a description. The verdict tells you what to do about it.
Delivery mix, audience delivery, creative rotation — the decisions ASC makes for you show up in your data, and the verdict on each one names the lever you still control.
Placement and format allocation over time — and what each lane returns. The black box’s outputs are public to you, even when its reasoning isn’t.
Demo-level delivery and efficiency under automation — checked against who actually converts, so you can catch the system optimizing toward cheap actions instead of your customers.
Automation reallocates across whatever creative you feed it, faster than manual ever did. Fatigue curves and mix gaps under ASC are the highest-leverage read left: creative is the steering wheel you still hold.
Anyone claiming to explain why Meta’s optimizer made a choice is guessing. Sentrum reads observable behavior with statistical methods and gates every verdict by significance — and says “insufficient data” when that’s the truth.
Two reads on what Meta’s automation actually changed — and why the buyers who keep reading their accounts keep winning.
Why Meta’s automation era rewards creative diversity and punishes set-and-forget — the context behind the black box.
READ THE GUIDEThe averages that hide what automation is doing — and the breakdowns that reveal it.
READ THE GUIDEThe optimization layer is — Meta doesn’t expose why the system shifts delivery, and no third-party tool can see inside the algorithm. But the behavior is fully observable in your own data: where spend actually went, which placements and demos it bought, what each lane returned, and how that changed over time. Sentrum reads the behavior rigorously instead of pretending to read Meta’s mind.
The Blueprint scan reads the account layer by layer — structure, audiences, creative mix, delivery mix, budget and pacing, signal quality — and narrates what the automation is actually doing in each one: where ASC moved spend, what it cost or earned, and which of the levers you still own (creative, inputs, exclusions, budget) is the right response.
Only if the data clears a real bar. Sentrum’s ASC-versus-manual comparison requires at least 10 purchases per cohort and a meaningful spend share on both sides before it will call a verdict — below that, it tells you the data is insufficient rather than inventing a winner. Be wary of any tool or thread that’s more confident on less data.
More than the “just trust it” framing suggests: creative mix and refresh cadence (the biggest one — automation optimizes across what you feed it), conversion signal quality, audience inputs and exclusions, and budget level. Sentrum’s findings name which lever applies, because “the algorithm did it” is a description, not a decision.
No. The connection is read-only by architecture with a standing never-touch-budgets rule. Sentrum diagnoses what the automation is doing; what to do about it stays your call.