How an attribution discrepancy arises
Every reporting system counts a conversion by its own rules, and the rules genuinely differ. Google Ads and Meta credit a sale back to the date of the ad click; GA4 records it on the day it actually happened. Ad platforms count view-through activity that analytics never sees. Analytics splits a visit into sessions and hands credit to the last non-direct source, while each ad platform only ever knows about its own clicks.
On top of the definitions sit the practical losses. Consent choices, ad blockers, browser limits on how long a cookie survives, and people moving from a phone to a laptop all break the chain between click and purchase in one system but not in another. Where a platform loses that link it may fill the hole with modelling rather than leave it blank, which pushes the totals further apart again.
The result is arithmetic, not error. Two systems asking different questions of the same week will return different answers, and the same pair of reports will usually disagree by a similar margin month after month.
Why attribution discrepancies matter
Budget follows reported performance. If Meta claims more sales than the shop’s own order records show and nobody challenges it, spend keeps climbing against results that were partly counted from views nobody remembers seeing. Cut spending instead because analytics shows less, and you may switch off a channel that was quietly working.
They matter for trust as well. A client who sees a different conversion total in every section of one report, with no explanation attached, stops believing all of them. Naming the gap and saying why it exists is part of honest reporting and a normal part of any analytics and tracking setup.
Where it goes wrong
The usual mistake is treating the difference as a bug to be fixed. Teams spend weeks re-tagging in the hope the totals will meet, and they never will, because the definitions were never the same to begin with. A close relative is comparing figures across different attribution windows without noticing: a platform set to a long click window will always report more than an analytics view that credits only the final click.
The most damaging version is adding platform totals together. Google, Meta and your email tool can each claim the same order, so summing their conversions produces a number larger than the business ever made.
What to do about it
Pick one source of truth for money decisions — normally your own order or enquiry records — and read platform numbers as direction rather than fact. Write down each system’s window, credit rule and timestamp so the same comparison is repeated every month, then watch the size of the gap instead of the raw totals. A stable gap is healthy; a sudden change is worth investigating.
For channel-level decisions, lean on blended cost per acquisition and on holdout tests, which answer what a channel added rather than what it claimed. In a smaller market such as Nepal, where monthly volumes per channel are low, that check matters more rather than less: a handful of double-counted orders can flip a channel from losing money to making it on paper.