purchase numbers didn't match the store backend, spent a while chasing down why

YUNJI Agency

Registered Member
Joined
Jul 10, 2026
Messages
55
Reaction score
12
Ran into this one a few weeks back, still not sure i fully nailed down every piece of it.

Had both pixel and capi running on an account, purchase numbers in ads manager started looking off, higher than what the store backend was actually showing. difference wasn't huge at first, maybe 15-20 purchases a week more than actual orders, but it kept showing up consistently.

Checked the obvious stuff first, no test purchases running, no bot traffic that i could find. went into the events manager and started comparing individual purchase events side by side, noticed some orders had two separate purchase events firing within a few seconds of each other, one from pixel, one from capi.

Turned out the event_id being passed wasn't matching between the two, pixel was generating its own id and capi was sending a different one for what should've been the same purchase event, so meta wasn't deduping them.

Fixed the event_id logic so both sides pass the exact same id for the same transaction, watched it for about a week after, numbers lined up with the store backend again, gap closed to basically nothing.

Still not 100% sure this was happening on every order or just a subset, didn't dig into how far back it went. anyone dealt with event_id mismatches like this before, how'd you end up spotting it?
 
I've run into something similar, and one thing I'd add is to be careful with creative testing while the tracking is out of sync. If Meta is over-reporting purchases, it can make a creative look much better than it actually is. You might end up scaling an ad based on conversions that don't exist in the backend. For me, the store/backend numbers are still the source of truth, while Meta is more useful for comparing relative performance between creatives. I'd also keep an eye on this when testing new hooks or formats. If the tracking issue happens at the same time as a creative change, it's very easy to attribute the improvement to the new creative when the difference is actually measurement. Definitely worth checking Pixel + CAPI deduplication before making big creative or budget decisions
that's a good point. the backend is what we use as the source of truth too, but i hadn't really considered how easily this could distort a creative test if both changes happened around the same time. definitely something i'll check earlier next time.
 
Back
Top