When to Use Seasonality Adjustments and Data Exclusions
Two settings that quietly rescue or wreck Smart Bidding. Here is exactly when each one applies.
Smart Bidding learns from the recent past and assumes tomorrow looks like last week. Most of the time that is a reasonable assumption.
Then you run a two day flash sale, or your tracking breaks for six hours, and the assumption becomes expensive.
Google gives you two tools for these moments. They do different jobs and using the wrong one makes things worse.
Seasonality adjustments: for planned short term changes
You use these when you know your conversion rate is about to change temporarily, and you want bidding to react immediately rather than learning after the fact.
The classic case is a sale. Your conversion rate normally sits at two percent. During a thirty percent off weekend it jumps to four. Without an adjustment, Smart Bidding spends the first day of the sale bidding as if conversion rate were still two percent, and you underspend on your best day of the quarter.
You tell Google to expect a conversion rate of four percent for that period, and it bids accordingly from hour one.
Rules for using them
Only for periods of one to seven days. Longer than that and Smart Bidding will learn on its own.
Only when you have a real basis for the number. Look at what happened during your last comparable sale.
Only for conversion rate changes, not traffic volume changes. Smart Bidding handles volume fine.
Set them in advance, and set an end date.
Do not use them for major annual events like Black Friday. Google has enough historical data for those and applies its own modelling. Adding your adjustment on top can double count the effect.
Data exclusions: for when the data was wrong
These do the opposite job. You use them when a period of your conversion data is unreliable and you want Smart Bidding to ignore it entirely.
The real use case is tracking failure. Your tag broke on Tuesday afternoon and was fixed Thursday morning. During that window Google saw clicks with no conversions, and it has now concluded that Tuesday afternoon traffic is worthless.
Without a data exclusion, that conclusion sits in the bidding model for weeks and quietly suppresses your performance.
Rules for using them
Apply as soon as you discover the problem, covering the exact broken window.
Keep the window as tight as you can. Excluding two weeks because six hours were broken throws away good data.
Do not exclude a period just because performance was bad. Bad performance is information. Broken tracking is not.
Do not use them for more than fourteen days. If your tracking was broken for a month, you have a bigger problem than a bidding setting.
The difference in one sentence
A seasonality adjustment says the world is temporarily different. A data exclusion says the data was temporarily wrong.
Using a data exclusion for a sale means Google learns nothing from your best conversion data. Using a seasonality adjustment for a tracking outage means Google bids hard into a period where it cannot see conversions. Both are bad in different directions.
The workflow that actually matters
Most accounts never use either of these, and the reason is that nobody notices tracking outages while they are happening.
So build the noticing first.
Set up an alert for a sharp drop in conversions relative to clicks. Google Ads has built in alerts, or use a scheduled check against your own order data.
When it fires, confirm whether it is a tracking failure or a real performance drop. Compare against your source system, not against another ad platform.
If it is tracking, fix it, then apply a data exclusion for the exact window.
Write down what happened and the dates. Future you will be looking at a strange dip in a chart and wondering.
The alert is the valuable part. The exclusion is just cleanup.
Common situations and what to use
Flash sale this weekend. Seasonality adjustment, conversion rate up.
Site down for four hours. Data exclusion for that window.
Checkout broken but tracking fine. Neither. The data is accurate and reflects real lost sales.
Conversion tag accidentally removed for two days. Data exclusion.
Price increase next month. Neither. That is a permanent change and Smart Bidding should learn it.
New product launch with unknown conversion rate. Neither. You do not have a basis for a prediction, so let it learn.
Duplicate conversions fixed, historical data inflated. Data exclusion for the inflated period.
That last one comes up more than you would expect. If you have just fixed a double counting problem, the historical data Smart Bidding learned from is wrong and worth excluding: duplicate conversions in Google Ads.
Where they live
Tools, then Bid strategies, then Advanced controls. Both adjustments are there.
You can apply them at account level or to specific campaigns. Campaign level is usually right, because a sale rarely affects every campaign equally.
One honest caveat
These are corrective tools, not performance levers. If you find yourself reaching for them weekly, the underlying problem is usually tracking reliability rather than bidding configuration.
Fixing the tracking removes the need for most of this: testing conversion tracking before launch.
Need this built properly?
These two settings exist for the weeks when reality and your data disagree. I set up and repair tracking for ecommerce stores and lead generation businesses, so the numbers in your ad accounts match the money in your bank account.
Have a look at what I do, or book a call and we can go through your setup together.
