Three weeks into a quarter, a client's paid search conversions dropped 40% overnight in the dashboard. No bid changes, no budget changes, no new competitors we could find, no seasonal explanation that fit. The client was, understandably, alarmed. We spent two days investigating campaigns before we found the actual cause: the platform had quietly changed its default attribution window, and the drop was almost entirely a reporting artifact, not a performance one.
What actually happened
The ad platform shifted its default conversion attribution model as part of a routine update. Conversions that used to get credited under a longer look-back window were now being attributed differently — some to other channels, some not being counted within the reporting period at all. Actual sales, checked against the client's own order data, hadn't moved meaningfully. The dashboard number had, because the definition of "conversion" underneath it had quietly changed.
Why this is more common than people think
Ad platforms update attribution logic more often than most advertisers realize, and the changes rarely come with a loud announcement — usually a line in a changelog nobody on the client side reads, sometimes nothing at all beyond a shift in the numbers themselves. If you're not cross-checking platform-reported conversions against a source of truth you control, a change like this looks identical to an actual performance collapse.
What we check now before ever calling something a "real" drop
Compare against first-party data. Platform-reported conversions get checked against actual CRM or order data on a recurring basis, not just when something looks wrong. If the platform says conversions dropped 40% and your own sales data says revenue is flat, you've found a reporting issue, not a performance one.
Check for attribution window or model changes. Before assuming a genuine performance problem, we check whether the platform's default attribution settings have changed recently. This takes ten minutes and would have saved us two days on this account if we'd checked it first instead of last.
Look at raw click and cost data separately from conversion data. Clicks, impressions, and cost per click are less prone to this kind of definitional whiplash than conversion counts. If those numbers look stable while conversions crater, that's itself a signal pointing toward a measurement issue rather than a demand issue.
The part that's hard to communicate
Explaining "the number in the dashboard is wrong, not the business" to a client who's already anxious about a 40% drop is a genuinely uncomfortable conversation, and we've had it more than once. The fix isn't complicated once you've found the cause. Finding the cause quickly, instead of two days into a reactive scramble, is the actual skill — and it starts with never fully trusting a single dashboard's definition of "conversion" without a second source checking its work.