A performance drop is not an emergency (yet)

The worst marketing decisions are made in the hour after someone notices a drop. Budgets slashed, campaigns paused, agencies blamed, before anyone has established what actually happened. Here's the triage to run first. It takes less time than the panic.

Step one: is it real, or is it measurement? Tracking breaks quietly and often: a site update drops a tag, a consent banner changes what's counted, a platform changes a definition, a form silently stops firing. The tell: conversions fall off a cliff while clicks and traffic hold steady. Real demand rarely vanishes overnight; measurement does it all the time. Check whether the thing being counted can still be counted before concluding people stopped wanting it.

Step two: is it everywhere, or somewhere? One campaign, one channel, one geography, one device, or all of them at once? A drop isolated to one place has a local cause: that campaign, that audience, that page. A drop everywhere at once points to something shared: tracking (again), the site itself, or the market. This single question usually cuts the suspect list by ninety percent.

Step three: did anything change just before? Pull the timeline: your changes (budgets, creative, targeting, the website), the platform's changes (recommendations auto-applied, learning phases reset by edits), and the calendar's (seasonality, holidays, a competitor's launch). Most real drops sit downstream of a change somebody made, and the change is often three to seven days before the drop became visible, not the same day.

Step four: size it honestly. Against what baseline? A drop versus last week might be a normal weekly rhythm. A drop versus an unusually good month is regression to normal, not crisis. Small absolute numbers swing violently in percentage terms: down 40% can mean three fewer leads. Decide how big the problem is in money before deciding how big the response should be.

Only after these four steps does "what do we do" become answerable, and by then it usually answers itself: fix the tag, adjust the one campaign, wait out the seasonal dip, or, when it really is demand, make a considered call with the actual size of the problem in view.

The discipline is having the evidence ready before the panic. OmniCommand runs this kind of triage continuously, flagging drops when they emerge, separating tracking gaps from real declines, and lining the change up against what happened in your accounts that week, so the question "what happened?" arrives with its answer attached.

Request early access at omnicommand.ai.