How a data exclusion works
Automated bidding learns from the conversions your account reports. If the tag stopped firing for a stretch — a site migration, a broken thank-you page, a consent banner change, a developer removing a script — the account looks as though demand collapsed. Smart Bidding believes the report, bids down, and the damage outlasts the outage.
A data exclusion is the correction. You give Google a start and end date, and the bidding models leave that window out of their calculations entirely. It does not delete anything from your reports; the conversions, or the absence of them, still appear in the interface. It only removes that period from what the bidding system learns from.
Why data exclusions matter
Because a tracking outage is a measurement failure, not a change in demand, and letting bidding treat it as real means paying twice: once for the leads you did not record, and again for the weeks afterwards when the strategy is cautiously bidding its way back up. An exclusion breaks that chain.
It is also the honest way to protect a strategy through a planned technical change. A website replatform, a move to a new domain, or a switch in tag setup all carry a real chance of losing conversions for a day or two. Knowing the exclusion exists lets you plan the change rather than fear it — and it is a good reason to check conversion tracking before, during and after any site work rather than discovering the gap a fortnight later.
Common mistakes with data exclusions
The biggest is using one to hide a bad patch. If conversions genuinely fell because demand fell, a competitor moved, or your offer stopped working, that is information the bidding system needs. Excluding it teaches the strategy a version of your business that is not true, and the correction arrives later and harder.
The second is confusing it with a seasonality adjustment. They solve opposite problems: an exclusion removes a period of broken data from learning, while a seasonality adjustment warns bidding that conversion rates are about to change for a short, known event such as a sale. Using one where the other belongs makes bidding worse, not better.
The third is excluding long stretches. The tool is built for short outages, and a wide window removes so much learning that the strategy is left thin. If the tracking was broken for a long time, the honest answer is to fix it and let the account rebuild.
How to act on it
Find the exact window first. Use the conversion reporting to see when recording stopped and restarted, and check your tag setup rather than guessing, because an exclusion set to the wrong dates removes good data and leaves the bad in place.
Apply it as soon as you have confirmed the outage, keep the window as tight as the evidence allows, and note it somewhere the next person will find — a bidding change with no explanation is a puzzle in six months. Then expect a short unsettled spell while the strategy recalibrates, and resist the urge to change targets during it. Most importantly, treat the exclusion as the second step. The first is repairing the tracking so the same window never has to be excluded again.