Attribution windows: why your dashboard and Meta disagree

A client asking why Meta reports twice as many conversions as Google Analytics for the same campaign is one of the most common support questions in performance marketing, and it's rarely a tracking error. It's usually a difference in attribution window and attribution model, two settings that quietly shape every number on every platform's dashboard.

We spend real time on this conversation with new clients specifically because the confusion, left unresolved, tends to erode trust in the reporting itself. A client who doesn't understand why two numbers disagree starts to wonder which one, if either, is accurate, and that doubt is corrosive to a working relationship even when both numbers are technically correct in their own context and neither side has done anything wrong.

Meta's default attribution window credits a conversion up to seven days after a click or one day after a view, even if the person never returned through the ad again. Google Analytics, depending on configuration, often uses a different window and a different model entirely. Neither number is wrong, they're measuring genuinely different things, but comparing them directly as if they measure the same thing produces confusion every time.

We walk new clients through a concrete example rather than explaining the concept abstractly, since the abstract explanation rarely lands as clearly. Showing an actual conversion that happened five days after a Meta ad click, credited by Meta but invisible to a Google Analytics session-based view, makes the gap tangible in a way a general description of attribution windows never quite manages on its own.

Meta counts a conversion that happens within a day of someone simply seeing (not clicking) an ad as attributable to that ad. This is a defensible model for measuring brand influence, but it means platform-reported conversions will almost always run higher than a source-of-truth analytics tool that only counts actual click-throughs, and that gap isn't fraud or error, it's a different definition of "worked."

The tricky part is that view-through attribution isn't inherently wrong to include, brand exposure genuinely does influence some later conversions that a click-only model misses entirely. The mistake is treating a number that includes view-through credit as directly comparable to one that doesn't, rather than understanding both as legitimate but differently scoped measurements of the same underlying campaign.

We recommend clients use platform-reported numbers for day-to-day campaign optimization, since that's what the algorithm itself is optimizing against, but use a single independent source (server-side tracking into a data warehouse, or a consistently configured analytics tool) as the actual source of truth for budget and business decisions. Trying to reconcile every platform's number to match is a losing exercise; picking one honest source and understanding why the others differ is the more useful outcome.

Once a client accepts that platform numbers and source-of-truth numbers will never fully agree, the conversation shifts to something more productive: tracking the ratio between them over time rather than the individual numbers in isolation. A stable ratio suggests measurement is behaving consistently even if the absolute numbers differ; a shifting ratio is a genuine signal worth investigating, since it usually means something about tracking, not just attribution methodology, has actually changed.

Platform-level privacy changes over the past several years have widened the gap between platform-reported and independently tracked conversions further still, since a meaningful share of iOS conversions are now modeled or estimated by the ad platform rather than directly observed. That modeling is generally reasonable in aggregate, but it makes the platform number even less directly comparable to a strict, unmodeled analytics count, which is one more reason we push clients toward a stable independent source rather than chasing platform numbers as the primary decision input.

We explain the modeling piece specifically because it changes how a client should react to a sudden shift in reported numbers. A modeled conversion count can move for reasons entirely unrelated to actual campaign performance, a change in the platform's own modeling methodology, for instance, and reacting to that kind of shift as if it reflects a real change in results leads to decisions based on noise rather than signal, which is exactly the kind of overreaction a stable independent source of truth is meant to prevent.

Rather than debating attribution differences reactively whenever a client notices a discrepancy, we run a brief monthly reconciliation across every active ad account, comparing platform-reported numbers against the independent source of truth and noting the ratio between them. A stable, explainable ratio month over month means the discrepancy is business as usual. A ratio that moves meaningfully is the actual signal worth digging into, and having a routine check in place means that signal gets caught quickly rather than discovered months later during an unrelated budget review.

We build client dashboards that show platform-reported and source-of-truth numbers side by side with a short explanatory note, rather than picking one number and hiding the other, since a client who eventually stumbles across the platform's own reporting and finds a different number is far more likely to feel misled than one who's seen both figures laid out with context from the start. Transparency about the discrepancy, explained once clearly, does more for long-term trust than any attempt to simplify the story down to a single number ever would.

We've found that clients who receive this dual view early on actually stop asking about the discrepancy after a month or two, once the pattern becomes familiar and expected rather than a fresh surprise every time it's noticed. That's really the goal of the whole exercise, not eliminating the gap between platforms, which isn't possible, but making it thoroughly unremarkable, so budget conversations can focus on what actually changed rather than relitigating why two numbers never quite matched in the first place.