Measurement & CAPI
This is the service that makes the other one honest. We implement and verify Meta's tracking stack: the Pixel, the Conversions API, Event Match Quality, deduplication, and incremental attribution. It ships as a fixed $3,400 build, standalone or folded into management.
The premise is simple. Delivery optimizes on the events you send, and reporting is only as real as the pipeline behind it. Fix the pipeline and both improve at once.
What broke, and when
Apple's App Tracking Transparency cut the ground out of browser-only tracking in 2021; Meta told investors the change would cost it around $10 billion in 2022 alone. Opt-in rates have recovered only partway, sitting near 25 to 37% globally in 2025 depending on the panel. Meta retired Aggregated Event Measurement in May 2023, and since January 2025 it blocks Purchase and Add-to-Cart events entirely for health and financial advertisers.
Each change pushed more weight onto server-side data. An account still running a lone browser Pixel in 2026 is optimizing on the thinnest signal Meta has ever had.
What the build includes
| Step | What we do | How it is verified |
|---|---|---|
| Event audit | Map every event firing today, browser and server | Events Manager diagnostics + raw payload review |
| CAPI implementation | Server-side events from your stack (Shopify, GTM server, or custom) | Test-events tool, then 7-day live comparison |
| Deduplication | Shared event IDs across Pixel and CAPI | Overlap report: no double-counted purchases |
| Match quality | Hashed customer parameters added per event | EMQ scored before/after; target 8+ |
| Attribution settings | Window review, incremental attribution on | Side-by-side report for one full cycle |
The pet-supplements brand in our CAPI write-up moved Purchase match quality from 4.9 to 8.2 in three weeks; cost per acquisition followed, $47 down to $40 over six weeks. That trajectory matches Meta's own published expectation of roughly 13% lower cost per action when CAPI joins the Pixel.
Why do we report incremental attribution?
Since April 2025, Ads Manager offers incremental attribution: a holdout-based model that reports only conversions Meta estimates the ads caused, rather than every conversion an ad touched. It nearly always shows a lower number than standard attribution. We turn it on anyway, because a flattering number you cannot defend to a CFO is a liability, and the deflated one tends to survive contact with your revenue ledger.
The weekly report shows three columns: Meta's standard attribution, the incremental view, and your actual revenue. Divergences get investigated, not narrated. What a healthy gap looks like is covered in our ROAS answer.
Attribution windows, set deliberately
Meta currently offers 1-day click, 7-day click, and 1-day view windows, with 7-day click plus 1-day view as the default. Longer windows report bigger numbers; that does not make them wrong, it makes them a choice. We match the window to your consideration cycle, document the choice, and never change it silently, because a moved goalpost invalidates every report before it.
What it costs, and what it touches
The measurement build is $3,400 fixed, most builds landing inside three weeks. It covers Shopify and GTM server-side stacks natively; custom backends add scope we price in the audit. Bundled with management, the build fee stands and the $1,400 audit is waived. Full details on pricing.
Questions we hear on calls
Do I still need the Pixel if I run the Conversions API?
Yes, run both. The browser Pixel catches what the server misses and vice versa, and Meta deduplicates the overlap through event IDs. Meta's published figure for the combination is about 13% lower cost per action versus Pixel alone, driven mostly by better event match quality.
What is a good Event Match Quality score?
Meta scores each event out of 10 based on how many customer parameters arrive with it. We treat 8 or higher as the target on the events that gate delivery, usually Purchase or Lead. Accounts routinely arrive at 4 to 6, which means delivery optimizes on a blurry picture of your buyer.
How is this different from what my developer already set up?
Most setups we audit fire events, which is not the same as firing them well. Typical findings: duplicate purchases inflating reported ROAS, missing event IDs breaking deduplication, hashed fields absent so match quality sits low, and test events left live. The build is verification work as much as implementation.
My brand is in health or finance. Can you even track conversions?
Partly, and honestly. Since January 2025 Meta blocks mid- and bottom-funnel events like Purchase for health, wellness, and financial advertisers, through both Pixel and CAPI. We work with upper-funnel optimization events, modeled conversions in your analytics stack, and the two-ledgers report so you still know what the spend produced.
Suspect the numbers? The audit scores your event pipeline in 14 days and costs $1,400.
Book a working session