Meta Ads with the
signal restored.
Audience overlap controlled, CAPI dedup verified, EU consent handled without losing analytics. Facebook, Instagram, Advantage+ — managed by a senior owner with the measurement layer underneath.
- LAL 1% buyers12M · lookalike
- LAL 3% engagers9M · lookalike
- Interest · stack A7M · broad
- Retargeting · 30d4M · warm
Four failure modes, repeated everywhere.
- 01
Adsets bidding against each other in the auction
Lookalikes overlapping with interest stacks and retargeting. Meta charges you twice to reach the same person. Auction efficiency collapses; reported CPM tells you nothing useful.
- 02
Pixel-only setup quietly losing 30–60% of events
iOS 17, EU consent, third-party cookie loss, in-app browser quirks. No server-side reinforcement means signal degrades silently while bidding strategies starve.
- 03
Advantage+ Shopping treated as a magic box
Default settings, no audience exclusions, no creative discipline. ASC eats brand traffic and post-purchase retargeting, then claims credit for everything. CAC looks great inside Meta; backend disagrees.
- 04
Creative testing without a system
One winning ad runs for nine months. Fatigue invisible in CTR. New variants tested without statistical confidence. Creative team and media buyer not talking — or worse, the same person guessing.
Meta managed as a signal problem first, a creative problem second.
- 01
Audience architecture with overlap control
Lookalikes, interest stacks, broad targeting and retargeting laid out with explicit exclusions. Overlap monitored in Audience Insights, not assumed away.
- 02Server-side
Conversions API with deduplication
Server-side CAPI via GTM SS or Stape. Matched event_id, em, ph, fbp/fbc. Monitored match rate. Cookieless-resilient signal back to bidding.
- 03
Consent Mode v2 for EU traffic
Required for EEA delivery. Configured so analytics survives consent decline via modelled conversions while staying compliant — EU traffic stops disappearing from reports.
- 04
Creative testing framework
Pre-defined variant matrix per cycle. Statistical thresholds documented before launch. Hook · body · CTA · format tested independently, winners logged.
- 05
Advantage+ Shopping with guardrails
ASC where it earns its keep — with brand exclusions, existing-customer audience suppression, and budget caps tied to incremental lift, not Meta-reported CAC.
- 06Reconciled
Backend-reconciled attribution
Meta-reported purchases compared against backend orders weekly. UTM hygiene, GA4 Custom Channel Group routing Meta to Paid Social, source/medium standardised.
- 07
Creative ops handoff
Brief format for your designer or production studio. Asset versioning. Naming convention so you can read the account in six months without us.
- 08
Decision log per campaign change
Same documentation discipline as our Google Ads work. Hypothesis, expected outcome, observed result. Auditable.
From audit to ongoing optimization.
- 01
Free Meta audit
Phase 01 · 5 working daysBM access, written deliverable, no contract. We map the account, score CAPI integrity, surface overlap and ASC behaviour, rank leaks by recoverable spend.
- →BM and account tree health score
- →CAPI dedup health and missing events list
- →Audience overlap map across adsets
- →Backend reconciliation gap (Meta-reported vs orders)
- 02
Signal layer
Phase 02 · Weeks 1–2Server-side CAPI with event_id dedup. Enhanced web events. Consent Mode v2 for EU traffic. Match rate monitored.
- →GTM SS or Stape container with Meta CAPI tag
- →Pixel ↔ CAPI dedup verified end-to-end
- →Consent Mode v2 live without analytics blackout
- →Match rate baseline + alerting threshold
- 03
Audience and campaign architecture
Phase 03 · Weeks 2–5Adsets restructured for overlap control. Advantage+ contained to its honest job. Retargeting separated. Creative testing matrix defined.
- →New audience map with documented exclusions
- →ASC carve-out with brand and existing-customer suppression
- →Creative testing matrix and naming convention
- →Weekly review cadence agreed
- 04
Test, scale, prune
Phase 04 · OngoingWeekly creative refresh, fatigue monitoring, scaling rules tied to incremental lift, not Meta-attributed ROAS alone.
- →Weekly creative test cycle with logged outcomes
- →Monthly attribution reconciliation against backend
- →Quarterly audience and ASC review
- →Documented scaling rules per market
Related case work.
East101
Meta + Google running cleanly together; channel group separating credit honestly.
Mitte
Audience and creative discipline; ASC carved out from retargeting to avoid double-paying.
Ararat
Server-side CAPI rebuilt; match rate stabilised; reported CAC reconciled against orders.
Yaka
Creative testing framework introduced; winner rotation cadence with documented uplift.
What clients ask before signing.
No, not for anyone serious about scaling. iOS 17, EU consent, and third-party cookie loss have shaved 30–60% off pure Pixel signal in the accounts we audit. CAPI with event_id dedup recovers most of it and stabilises bidding.
We buy media and design the testing framework. We do not run a creative studio. We brief and rotate creative against your designer or production partner — and document the system so you don't lose it if we part ways.
ASC where it earns its keep — usually mid-funnel for ecom scale. Always with brand exclusions and existing-customer suppression so it stops claiming credit for repeat buyers. We cap its budget by incremental lift, not by Meta-reported ROAS.
Yes, on every EU-targeted account. Without CMv2 your Meta and GA4 EU traffic gradually disappears from reports as opt-outs accumulate. Configured properly, modelled conversions fill the gap without breaking compliance.
Yes — that's actually our preferred engagement. Channel group separation in GA4 prevents the two from double-claiming credit, and the same senior owner runs both rather than two siloed teams.
Meta running, but the numbers don't add up?
Send BM access. Five working days later you get a written diagnostic — CAPI integrity, overlap map, ASC behaviour, and what we'd touch first.
No long contracts · Free audit before any work