Why your Meta CPL spiked after iOS and the server-side fix that actually works
Server-side Conversions API is not a nice-to-have anymore. Most agencies are still getting it wrong. Here is a practical guide to the migration that actually restores attribution accuracy.
D Venkataramana
Founder, Digi Brand Booster
What actually broke after iOS 14
Apple's App Tracking Transparency framework, launched in April 2021, required apps to ask users for permission before tracking. Most users said no. This broke the Meta pixel's ability to track conversions from iOS devices, which typically account for 40 to 60 percent of mobile traffic for most Indian D2C brands. Meta's attribution window shrank from 28 days to 7 days. Reported ROAS dropped. CPL appeared to spike. In most cases the ads were not actually performing worse, but the data was severely undercounting what was being attributed to Meta.
Why the browser pixel alone is no longer sufficient
The Meta browser pixel fires from the user's device and relies on third-party cookies to match events to ad impressions. Post-iOS, this matching fails on a large portion of traffic. The result is that Meta's algorithm receives incomplete conversion data, which degrades its ability to optimise campaigns for the right users. The algorithm is only as smart as the data it receives. Degraded signals mean the algorithm optimises toward the users it can see, not the users most likely to convert.
What server-side Conversions API actually does
Conversions API (CAPI) sends conversion events directly from your server to Meta's servers, bypassing the browser entirely. It does not rely on cookies or device-level tracking. When a purchase happens on your site, your server fires the event to Meta with the order details, customer email (hashed), and event parameters. Meta matches this against its user data on its side. This restores a significant portion of the attribution that the pixel alone was missing, giving Meta's algorithm better signals to optimise on.
The correct implementation: deduplication is critical
The most common mistake we see when auditing accounts is running both the browser pixel and CAPI without proper deduplication. This results in the same conversion event being reported twice, which inflates reported ROAS and causes over-bidding. Proper deduplication requires passing a unique event ID from both the pixel and the CAPI event for the same conversion. Meta matches on the event ID and counts it only once. Without this, you are solving one problem (undercounting) and creating another (double counting).
What to expect after a proper CAPI implementation
After a correctly implemented server-side CAPI migration with deduplication, most of our clients see reported conversions increase by 25 to 40 percent within the first two weeks. This does not mean more sales happened. It means Meta is now seeing conversions it was previously missing. The algorithm then has better data to optimise campaigns, which typically leads to gradual ROAS improvement over the following 4 to 8 weeks as the campaign learning phase resets with the improved signals.
Want this done for your brand?
Book a free 48-hour audit. We will identify exactly what is holding back your performance and what to fix first.
Get My Free Audit