What a cross-device conversion is
Someone sees an ad on their phone during a commute, taps it, reads for a minute and moves on. That evening they open a laptop, search the brand name and buy. The click happened on one device and the purchase on another, so nothing on the laptop connects the sale back to the phone unless the platform can join the two.
Ad platforms recognise these journeys in two ways. Where a person is signed in to the same account on both devices, the link can be observed directly. Where they are not, the platform estimates it from patterns across many similar journeys, which means part of the reported figure is modelled rather than counted.
Why cross-device conversions matter
Without them, mobile advertising looks far weaker than it is. Phones are where discovery happens; larger screens are where forms get filled in, cards get typed and comparisons get made. Judge a mobile campaign on same-device conversions alone and you will cut the activity that starts most of your journeys.
This matters especially in Nepal and similar markets, where browsing is overwhelmingly mobile but the final step often is not — a bank transfer completed on a desktop, an office computer used for anything involving documents, a family member’s laptop for a larger purchase. The same pattern shows up in high-consideration categories everywhere: property, education abroad, medical treatment, B2B services.
Where cross-device measurement goes wrong
The first mistake is expecting completeness. Not every journey can be joined, so the reported figure is a floor rather than a full count, and the gap grows as more people browse signed out or in private windows.
The second is double counting across platforms. Each ad platform resolves these journeys with its own method, so adding one platform’s cross-device conversions to another’s produces a total that never happened. The third is treating the estimate as a person-level fact. These figures describe patterns, not identified individuals, and using them as if they name a customer is both wrong and a privacy problem. Where the estimate is a large part of your reported result, say so in the report rather than presenting one confident number.
What to do about it
Improve what can actually be joined before worrying about what cannot. Capturing an email address or phone number early in the journey — a saved basket, a brochure request, an account — gives you a link that survives a change of device, which is why lead magnets earn their keep beyond the leads themselves. Sending outcomes back from your CRM does the same job at the other end.
In reporting, separate what was observed from what was estimated, and make device decisions on the whole journey rather than on last-device results. If mobile shows heavy engagement and few completed sales, the honest reading is usually that mobile is doing its job and the desktop step is taking the credit — check that before shifting budget. For anything where the split really drives spending decisions, read the modelled conversions definition first, and make sure the basics of your analytics and tracking setup are sound, because a broken tag on one device looks exactly like a cross-device journey.
