When web and mobile counts disagree
Why the same purchase can appear once on mobile and twice on web — and how to decide which number to trust for the week.
Product meetings stall when the mobile lead cites 1,240 completions and the web lead cites 1,580 for the same week. Both numbers can be internally consistent and still describe different things.
Common causes include double-firing on redirect flows, session stitching that drops when users move from app to browser, and backend confirmations that never reach the client layer. Distributed services add another layer: a queue consumer may emit an event that the storefront also emits.
Before changing dashboards, write a single definition for the activity you care about — for example, "order confirmed by payment service" — and map every surface to that definition. Numbers that still diverge after that map usually point to an instrumentation bug, not a reporting preference.
Teams in Gumi and elsewhere often discover that one platform was counting cart submissions while another counted successful charges. Aligning those definitions usually removes more noise than any new chart.