How data latency works
A click, a form submission or an order does not travel straight into a chart. It is collected by a tag, sent to a platform, queued, checked, matched to a user or a campaign, joined with cost and currency data, then written into the tables the reporting interface reads. Each of those stages takes time, and the total is the latency you experience.
Different parts of the same tool move at different speeds. Realtime views are built from a separate, shallow pipeline and appear almost immediately. Standard reports arrive later. Anything involving conversions, cost, modelled figures or currency conversion arrives later still, because the platform is waiting for information that has not been generated yet. A conversion that happens days after the click cannot appear on the click’s day until the conversion exists.
Why data latency matters
Most bad marketing decisions made in a hurry are decisions made on incomplete data. Yesterday’s cost is usually final, but yesterday’s conversions are not, so a campaign checked too early looks unprofitable and gets paused before its results have finished arriving. The campaign was fine; the report was young.
Latency also breaks reconciliation. If your ad platform, your GA4 property and your CRM all settle at different speeds, pulling them into one view on the same morning guarantees a mismatch that nobody can explain, because there is nothing to explain — one source has simply not caught up.
Where data latency goes wrong
The first mistake is treating today’s figures as final. Today is always the least complete row in any table, and it will keep rising for the rest of the day and often beyond.
The second is exporting a report, saving it, and comparing that saved copy with the same report pulled later. The later pull will be higher, and the difference gets blamed on a tracking fault rather than on data that simply finished landing. Automated exports that run too soon after midnight bake this problem into every historical record you keep.
The third is confusing latency with loss. A figure that is late will arrive. A figure blocked by consent, an ad blocker or a broken tag never will. The fixes are completely different, so it is worth waiting the normal settling period before starting an investigation.
How to act on it
Learn how long each of your sources takes to settle by watching the same day’s number over several days, and write that period down. Then set your reporting rhythm behind it: judge campaigns on windows that have finished settling, and leave the most recent days visible but clearly marked as incomplete rather than hiding them.
Schedule automated exports and scheduled reports and dashboards late enough that they capture settled data, and note the pull time on the file. When someone questions a discrepancy, the first question should be when each number was taken, not whether the tracking is broken.
