Two channels, one event stream
The Meta Pixel sends events from the visitor's browser. The Conversions API (CAPI) sends the same events from your server. Meta receives both and, when they are matched correctly, treats them as a single event rather than two.
The point of the second channel is resilience: browser-side signals can be limited by content blockers, consent choices and platform changes. A server-side event still fires when the browser drops one.
What must match for deduplication
Deduplication depends on two identifiers travelling with both the browser and server event: event_id and event_source_url. If either differs, Meta counts two conversions instead of one — inflating results and confusing optimisation.
The cleanest pattern is to generate one event_id at the moment the event happens and hand it to both the Pixel call and the server payload.
- event_id: identical on browser and server.
- event_source_url: identical on browser and server.
- event name: identical spelling and casing on both sides.
- action_source: set to website on server events.
What the server actually sends
A CAPI payload includes the event name, timestamp, a click ID or browser ID when available, and customer information — email or phone, hashed before transmission — plus custom data such as value and currency.
Hashing matters: customer data should be normalised (trimmed, lowercased for email) before hashing, or the match quality silently drops.
How to verify it works
Open Events Manager and check that events appear with both browser and server sources, that deduplicated counts are stable, and that the event setup tool shows no configuration errors.
Verify the four events that matter to the business — for example ViewContent, AddToCart, InitiateCheckout, Purchase — before scaling spend. Anything else is noise until the core path is healthy.
About the author
T. M. Saiful Islam, full-stack web developer & digital growth specialist.
