Every platform overreports itself; that is not a scandal, it is a settings default. The fix is a measurement stack you control: server-side signal that feeds the algorithm properly, blended metrics that catch dashboard drift, and incrementality tests for the decisions big enough to deserve proof.
Four patterns from the tracking audits we run.
Browser-only tracking loses events to privacy tools, blockers, and redirects. Meta then optimizes on the fraction it sees. Server-side CAPI with deduplication is table stakes, and most accounts still don't have it wired correctly. Our free tracking audit shows you in 20 seconds.
Ads Manager, GA4, and your revenue never agree; each has different windows and biases by design. Without a declared source of truth, budget meetings become theology debates.
Some conversions would have happened without the ad. Where spend is big enough to matter, geo holdouts and lift tests separate caused revenue from claimed revenue.
A measurement stack earns its keep when it changes budget allocation. We wire every metric to a decision rule: what number, at what threshold, moves what dollars.
Built in week one, verified continuously, escalated to tests when spend justifies.
Conversions API with event deduplication, event match quality pushed to 8.0+, consent-stack compatibility, and monitoring that catches breakage before Meta does.
Platform numbers reconciled against revenue truth: MER, blended CAC, cohort views. One agreed source of truth for budget decisions.
Geo lift tests and holdouts for the big questions: does retargeting pay, is brand spend real, what happens at 2x budget. Proof priced into the decision it settles.
Offline conversion uploads, LTV modeling by cohort and creative, lead-quality feedback for lead-gen accounts. The algorithm learns what a good customer looks like.
Run the free tracking audit. It scans your pixel, GTM container, Meta's own pixel config, and your consent stack in about 20 seconds, no login, and hands you a prioritized fix list with evidence.
The Conversions API sends conversion events from your server to Meta directly, alongside the browser pixel, with deduplication so nothing counts twice. Browser-only tracking routinely misses 20 to 40% of conversions, and Meta optimizes on what it receives. So yes, if you spend real money, you need it.
The full working stack is: CAPI, dedup, blended reporting, and offline uploads are part of every engagement on the pricing page. Formal incrementality studies get scoped when spend is large enough for the answer to pay for the test.
Yes, that is the most common starting state: duplicate pixels, missing dedup, GTM tags firing twice, consent blockers eating events. The audit finds it; week one fixes it.
We hold accounts to 8.0 or better before scaling spend. Below that, Meta is matching too few events to real people, and every downstream optimization inherits the blur.
Run the free tracking audit first if you want receipts before the call. Either way, thirty minutes and you’ll know exactly what your measurement is hiding.