In short: they are two opposite tools. The seasonality adjustment warns Smart Bidding about a short, predictable change (1-7 days) and its percentage comes from the change in conversion rate during the previous event, always erring on the conservative side. The data exclusion is retroactive and wipes a broken period — but first you fix the fault, with exact dates.
Module 2 of the Intermediate level explained when to use seasonality adjustments and data exclusions. This tutorial sets them up step by step, with the percentage calculation and three worked cases.
Where are the seasonality adjustments?
Tools → Bid strategies → Advanced controls (or "Seasonality adjustments" and "Data exclusions" in the Tools menu, depending on your version of the interface). Two tabs: Seasonality adjustments and Data exclusions. Both are created at account level and applied to the campaigns you choose. From a manager account you can create them for several accounts at once.
Part A · How do you set up a seasonality adjustment?
Step 1 · Work out the percentage
The adjustment is the expected change in the conversion rate, not in conversions and not in traffic. Using your data from the previous event:
% adjustment = (conversion rate during the event ÷ normal conversion rate − 1) × 100
Example: normal rate 2.0%, rate during last year's sales 3.1% → (3.1 / 2.0 − 1) × 100 = +55%. Be conservative: if in doubt, +40%. If the event is new, estimate cautiously (+20-30%) or set no adjustment at all. For drops (a public holiday when nobody buys but people still search), use a negative percentage.
Step 2 · Create the adjustment
- Seasonality adjustments → +.
- Name ("January 2027 sales"), description (the calculation).
- Scope: Search / Shopping / Display, depending on the campaigns affected (one adjustment per type if it affects several).
- Campaigns: all of them within the scope, or a selection (only the ones taking part in the event).
- Devices: all of them, or only the ones that change (e.g. mobile in an app promotion).
- Dates and times: the event's exact start and end. Google recommends short periods (1-7 days); for longer events, better with no adjustment (the algorithm learns anyway).
- Conversion rate adjustment: the percentage from step 1.
- Save. You can create it in advance (it stays scheduled).
Step 3 · Check and withdraw
During the event, the bid strategy report shows the adjustment as active. If you see that the real rate isn't going up as expected (day one: compare with the day before), edit the percentage downwards or delete the adjustment: keeping a wrong one makes every click in the period more expensive. When it ends, it expires by itself.
Part B · How do you set up a data exclusion?
Step 1 · Pin down the exact dates
From the conversion diagnostics, the change history and the upload sheet: the first day with broken data (conversions at zero, duplicated, a faulty bulk upload) and the last day (when it was fixed and verified). Include both. If the failure was partial (one device, one campaign), narrow it down.
Step 2 · Fix it first
The exclusion doesn't fix anything: it stops Smart Bidding learning from the period. First the tag, then verification with a real conversion, and only then the exclusion.
Step 3 · Create the exclusion
- Data exclusions → +.
- Name ("Measurement broken 10-16 August"), description (what happened).
- Scope (Search / Shopping / Display) and the campaigns affected.
- Devices if relevant.
- Dates: the start and end of the failure. It can be retroactive.
- Save. The affected strategies may go into "learning" briefly.
Step 4 · Afterwards
Don't touch targets or budgets for a few days. Note it in the change log. If the failure repeats, the cause (a redesign, a plugin) is what needs resolving, not creating exclusions on a production line.
Three worked seasonality cases
Black Friday (Friday to Monday)
Normal rate 1.8%; last year during BF, 3.4% → +89%; conservatively +70%. Scope Search and Shopping, product campaigns (not brand or remarketing, which go up by themselves), all devices, from Friday 00:00 to Monday 23:59. Created a week in advance. On Tuesday it expires; budgets and targets come back in steps (module 9 of the Intermediate level).
A week of broken measurement (thank you page deleted in a redesign)
The website's change history: redesign on the 10th; conversions at zero from the 10th; tag fixed and verified on the 16th. Data exclusion from the 10th to the 16th, scope Search, all campaigns with that conversion. Created on the 17th. No touching targets until the 30th.
A 48-hour flash promotion (−30% code by email)
No history for the event. A cautious estimate of +30%, only generic Search and Shopping campaigns, exactly 48 hours. Day one: real rate +45% → leave it. Day two: +20% → the adjustment is lowered to +20% at midday. It expires by itself.
Which mistakes are made with these adjustments?
- A percentage based on conversions or traffic instead of on the conversion rate.
- Adjustments lasting weeks (the whole of Christmas): they distort more than they help.
- A data exclusion for a bad week of trading.
- Approximate dates in the exclusion: it either leaves broken data in or takes good data out.
- Creating the exclusion before fixing the measurement.
💡 Ninja trick: the SBNS judge reads the active seasonality adjustments and recalibrates its window (it doesn't judge with the spike inside), and the Agent warns you on the first day of conversions at zero with normal clicks — the moment to fix things and note down the exclusion's start date. Both tools in this tutorial are manual by design: they are the ones that depend on what you know and Google doesn't.
What you should remember
- Seasonality: percentage = the change in conversion rate from the previous event, conservative; 1-7 days; the campaigns taking part; edit or delete it if it goes wrong.
- Data exclusion: fix it first, the exact dates of the failure, retroactive; no touching bids afterwards.
- Three cases: BF +70% / exclusion 10-16 / flash +30% adjusted on day two.