Conversions API: Tracking in a post-cookie world
iOS 14.5 ended the Meta Pixel tracking era. CAPI is not the end of that story — it is the continuation under new rules.
The Conversions API (CAPI) — formerly the Facebook Server-Side API — is Meta's server-side tracking infrastructure. It allows advertisers to transmit conversion events directly from their own server to the Meta Marketing API: completely independent of browser restrictions, cookie blockers, iOS App Tracking Transparency, and ad blockers. This makes CAPI the central tracking technology for e-commerce stores running Meta ads profitably in the current privacy environment.
What iOS 14.5 actually changed — and what it means
With iOS 14.5 (April 2021), Apple introduced the App Tracking Transparency framework: users must actively permit apps to track them across apps and websites. Over 85% of iOS users deny this permission. The consequences for Meta advertisers were immediate and structural:
- Event loss: The Meta Pixel runs in the browser and depends on cookies and JavaScript execution. For iOS users who deny tracking, Pixel events are either not captured at all or recorded with significant delay. Depending on audience composition, 30–60% of all Pixel events are lost.
- ROAS underreporting: Fewer captured conversions means lower measured ROAS. Campaigns that are genuinely profitable appear loss-making in reporting — and are incorrectly paused or reduced, cutting off what was actually working.
- Algorithm degradation: Meta's campaign algorithm optimizes based on received conversion signals. Fewer signals mean worse algorithm quality, higher CPMs, and lower efficiency — a self-reinforcing downward cycle that compounds with every optimization period.
- Audience quality degradation: Retargeting audiences and lookalike audiences are built from Pixel data. Missing events degrade the quality of both audience types — with direct impact on the conversion rate of retargeting campaigns.
Benchmark to assess your own tracking damage: compare Meta-reported purchase events with actual orders in Shopify or your order management system for the same period. A discrepancy of more than 20% is a clear signal for tracking gaps that CAPI can close.
How CAPI works technically
The technical architecture of the Conversions API differs fundamentally from the Pixel:
- 1 Pixel (client-side): JavaScript code in the user's browser. Fires events directly from the user's device to Meta. Dependent on: JavaScript execution, cookie consent, ad blocker status, browser privacy settings, app tracking permission. Susceptible to all browser-side privacy restrictions.
- 2 CAPI (server-side): Server-to-server communication. Your own server (Shopify backend, custom infrastructure, CDP) receives order data and sends it directly to the Meta Marketing API. Independent of browser restrictions, fully under your own control and operational visibility.
- 3 Deduplication: Since both systems run in parallel, Meta must be instructed to remove duplicates. This is done through identical `event_id` parameters in both Pixel and server events. Without correct deduplication, conversions are double-counted — with serious consequences for campaign optimization and budget allocation.
- 4 Hashing of personal data: Matching parameters (email, phone, name) are SHA-256 hashed before transmission. Meta dehashes on its side and matches against its own user profiles. The hashed data is usable by Meta but cannot be decrypted by third parties without the original data.
Event Match Quality: The critical optimization lever
Event Match Quality (EMQ) is Meta's scoring scale (0–10) for the quality of user attribution of a CAPI event. The higher the EMQ, the more conversions can be attributed to a Meta user profile — and the more precisely the algorithm can optimize:
- EMQ 8–10 (excellent): Multiple high-quality matching parameters present (email + phone + name + address). Nearly all events can be attributed to a Meta user. Maximum algorithm quality and lookalike audience precision.
- EMQ 5–7 (good): 1–2 matching parameters. Most events are attributed. Sufficient for effective campaign optimization and solid audience building.
- EMQ 0–4 (weak): Only IP address or user agent. Many events remain unattributed. Algorithm quality suffers, lookalike audiences become less precise.
- Most important lever for high EMQ: Email address is the single strongest matching parameter because Meta has a valid email address for most active users. Every validated customer email sent with the purchase event significantly increases attribution probability — this alone can move EMQ from 4 to 7+.
CAPI implementation: Options for e-commerce stores
CAPI implementation has several technical paths with different levels of effort and completeness:
- Native shop integrations (recommended for getting started): Shopify offers a native CAPI integration through the Meta app that activates basic server-side tracking without development effort. Limitation: limited control over matching parameters and event configuration compared to custom implementations.
- Partner integrations (e.g., Elevar, Littledata, Conversios): Third-party tools that manage CAPI integration for Shopify and other platforms. Higher EMQ through better data mapping, simpler configuration than custom development. Recommended for stores without dedicated engineering resources.
- Server-side Tag Manager (Google Tag Manager Server-Side): GTM running on your own server manages Pixel and CAPI events centrally. Enables full control and customizability. Recommended for advanced tracking architectures with multiple ad platforms.
- Direct API integration: Custom development via the Meta Marketing API. Maximum flexibility and data quality, highest development investment. Appropriate for enterprise stores with dedicated engineering capacity and complex data requirements.
For most Shopify stores, the fastest path to improved tracking is: activate Shopify's native Meta integration + Elevar or Littledata for higher EMQ. This combination delivers 80% of the benefit of a custom implementation in a fraction of the time and cost.