Every October, a version of the same message lands in our inbox from a client: “Should we set up seasonality adjustments for Black Friday?”
For years the standard answer, including from Google itself, was yes. Tell Smart Bidding the spike is coming, let it bid harder, capture the peak.
Then Optmyzr published a study that made a lot of us rethink it, and honestly it matched what we’d already been seeing in a few accounts without being able to prove it.
This piece covers what the feature actually does, what the data says about the events everyone uses it for, and the handful of situations where we still reach for it.
Start with what it is, in plain terms
A seasonality adjustment is a note you leave for Google’s bidding system about a future window of time. You are telling it: conversion rate is going to be roughly X% different from normal between these two timestamps.
You are not raising bids by X%. You are changing the system’s estimate of how likely a click is to turn into a sale, and bids follow from that estimate.
Two details matter more than people realise:
It has a hard start and end time. You pick the hour, not just the day. When the window closes, the adjustment stops applying and bidding goes back to what it was doing before.
Google’s own advice is restrictive. The Help Centre says to use them only “if you expect major changes to conversion rates,” because Smart Bidding already handles seasonal events on its own. Google recommends them for events of one to seven days and warns they may not work well past fourteen days.
Availability, as of now: Search, Shopping and Display campaigns running Target ROAS or Target CPA, plus Performance Max and App campaigns on any bid strategy. Travel campaigns are not supported. If you’re running Maximise Clicks or manual bidding, the feature does nothing for you.
One more thing worth knowing if you manage several accounts: you can create an adjustment once at manager account level and push it down to linked client accounts. Accounts added to the MCC afterwards will not inherit it, so check your list before a big event.
The study that changed how we answer this question
In November 2025, Frederick Vallaeys and the Optmyzr team published an analysis of three consecutive Black Friday and Cyber Monday periods, 2022 through 2024. They looked at up to 6,000 advertisers per year (5,886 accounts in 2024, 5,114 in 2023, 4,292 in 2022) and split them into two groups: advertisers who set a seasonality adjustment during BFCM, and advertisers who did not.
The results were consistent across all three years.
Smart Bidding caught the spike on its own. For the group that set no adjustment at all, conversion rate still rose: 17.5% in 2022, 11.9% in 2023, 7.5% in 2024. Nobody had to tell the algorithm that Black Friday was happening.
Adjustments roughly doubled CPC inflation.
| Year | CPC change, no adjustment | CPC change, with adjustment |
|---|---|---|
| 2022 | +17% | +36.7% |
| 2023 | +16% | +32% |
| 2024 | +17% | +34% |
ROAS went the wrong way.
| Year | ROAS change, no adjustment | ROAS change, with adjustment |
|---|---|---|
| 2022 | -2% | -17% |
| 2023 | -1.5% | -10% |
| 2024 | +5.7% | -15.7% |
The group that left Smart Bidding alone held ROAS roughly flat, and actually improved it in 2024. The group that intervened lost ten to seventeen points of efficiency every single year.
Vallaeys explains why in one line that’s worth pinning above your desk: “Smart Bidding takes this literally. It does not hedge your bet.”
That’s the whole problem. If you say +50% and the real lift is +40%, Google does not split the difference or test into it. It bids as though you were exactly right, and you overpay on every auction for the duration (Optmyzr study, also covered by Search Engine Journal).
The honest counterpoint
The same study found the adjustment group grew revenue faster: +50.5% vs +25% in 2022, +52.8% vs +30.3% in 2023, +39.9% vs +33.8% in 2024.
So it is a real trade. If your instruction from the business is “clear this inventory” or “take share this weekend, we’ll worry about margin later,” seasonality adjustments do buy you volume. You are just buying it at a worse rate. Say that out loud to the client before the event rather than explaining it afterwards.
The question that decides it
After reading that study and testing it against our own accounts, we replaced our old checklist with a single question:
Does Google already know this event is coming?
Black Friday, Cyber Monday, Christmas, Valentine’s Day for a florist, back to school for a stationery brand. These are baked into Google’s models. It has years of data on how buyer behaviour shifts in those windows, across millions of accounts, in your country and your category. Adding your forecast on top is not new information. It is a second push in a direction the system is already moving.
Now compare that to a Friends and Family sale you invented last month, a clearance you have never run before, or a big email blast landing Tuesday morning that will drive a wave of warm traffic to search. Google cannot see any of that. It has no history for it. That gap between what you know and what Google can possibly know is the actual job of this feature.
Vallaeys puts the rule about as simply as it can be put: if the spike is global and historical, Google has the data. If it’s unique to your brand, you need to tell it.
When we do use them
Short version of our internal rule at River Stone: the event has to be short, sharp, ours, and backed by something other than optimism.
We’ll set an adjustment for:
- A one-off flash sale with no precedent in the account
- A first-time promotion or bundle we’ve never run
- A large email or SMS push driving a concentrated surge of high-intent traffic
- A launch or restock with a date we control
- A niche industry deadline that doesn’t show up in global trend data (a licensing cut-off, a regional compliance date, a trade show week)
- An expected slowdown that’s specific to the business rather than the calendar
That last one gets ignored. Michelle Morgan of Paid Media Pros makes the point well in her WordStream guide: in B2B accounts, there are stretches like spring break, the back end of summer and the week between Christmas and New Year where conversion rate drops predictably, and a downward adjustment stops Google from bidding into a dead week (WordStream). If your sales team is on leave and nobody is answering the phone, that’s real information Google doesn’t have.
When we don’t
- Black Friday and Cyber Monday. The data is clear enough now that we default to no adjustment and spend the energy on budget pacing instead.
- Anything longer than a week. Google’s own limit is 1 to 7 days ideal, 14 days maximum useful. A whole busy summer is a budget and structure problem, not a bidding problem.
- Permanent changes. New landing page converting better? That’s your new baseline. Let the system learn it.
- Weekly patterns. If Saturdays are always strong, Smart Bidding knows.
- Thin conversion volume. If you cannot see a stable baseline conversion rate, you cannot forecast a change to it.
- Broken tracking. Adding a bidding instruction on top of bad data makes the bad data more expensive.
- “We think it’ll be big.” Hope isn’t a signal.
Working out the number
The maths is simple. The judgement is not.
Adjustment = (expected event CVR - normal CVR) / normal CVRNormal conversion rate 2.0%, expected event rate 3.0%, so 1.0 / 2.0 = a 50% lift.
Two habits that have saved us money:
Set it below your estimate. If last year showed +50%, we’ll often set +30% or +35%. Given how Smart Bidding treats your number as gospel, being slightly under is cheap and being over is not. Overshooting your forecast by ten points means you’re overbidding by roughly 7% relative to what the traffic is actually worth, on every auction, for the whole window.
Check the number against the right conversion. A lot of “lifts” from past events are lifts in form fills, not in qualified leads. For lead gen especially, pull the sales data before you trust the account data. If last year’s spike was mostly tyre kickers and price shoppers, bidding harder for more of them is not a plan.
For reference, the API accepts a conversion rate modifier between 0.1 and 10.0, so the interface will happily let you set something wild. That’s not permission.
The post-sale question, and a correction worth making
You’ll read a lot of advice, including in the older version of this article, telling you to set a negative adjustment after a sale ends so Google doesn’t keep bidding aggressively into weak traffic.
Google’s documentation says that isn’t necessary: campaigns return to their pre-adjustment behaviour once the event window closes, and no negative adjustment is required after a promotion.
Both things can be true in practice. The adjustment itself does expire cleanly. But your recent conversion data still feeds the bidding models, and after a heavy discount period, real conversion rate often sits below normal for a few days while everyone who was going to buy has already bought.
Our approach is to handle that with budget and targets rather than a second adjustment. Bring daily budgets back to normal on the first quiet day, and if the drop is severe and predictable from last year’s data, tighten the target instead of stacking another instruction onto the bidding system. Fewer moving parts, easier to diagnose when something goes wrong.
Three tools people mix up
Seasonality adjustments are for expected conversion rate changes in a future window.
Data exclusions are for the opposite problem in the past: your tracking broke, a tag fired wrong, conversions went missing for two days. You exclude that window so it doesn’t poison the model. Use this when the data is wrong, not when the demand changed.
Target CPA / Target ROAS changes are for when your business goal changed, not the calendar. Pushing harder for all of Q4 is a target and budget decision. A 48-hour flash sale is a seasonality decision.
If you can’t say which of the three your problem is, you’re not ready to touch any of them.
Setting one up
Tools > Budgets and bidding > Adjustments > Seasonal, then the plus button. Choose “Conversion rate” as the adjustment type.
The interface moves around, so if it’s not where I’ve said, look under Budgets and bidding.
A few things to get right while you’re in there:
Scope it narrowly. You can pick campaign types, specific campaigns, and devices. If the sale is on two product categories, don’t apply the adjustment to your branded search campaign. Brand traffic behaves nothing like non-brand during a promotion, and pushing brand bids up during a sale is a fast way to pay more for people who were already coming.
Set the hours, not just the dates. If the sale goes live at 9am, start it at 9am.
Name it so future you understands it. Not “BF test.” Something like “Black Friday 2026 | Non-brand Shopping | +30% CVR | Nov 27-30.” Next year that name is your only record of what you did and why.
Before you press save
- Is the offer genuinely new, or does Google already have years of data on this event?
- Is the window seven days or less?
- What was the actual conversion rate lift last time, measured on qualified conversions and not just form fills or micro-conversions?
- Is the offer live at the exact hour the adjustment starts?
- Are landing pages, promo codes, feeds and Merchant Center promotions all live and tested?
- Which campaigns should be excluded? (Usually brand.)
- Is budget high enough to actually capture the demand, or will you hit the cap at 2pm on day one?
- Who is watching spend during the event, and at what number do they turn it off?
- What happens on the first day after?
If those questions feel tedious, that’s rather the point. The setup takes four minutes. The decision should take longer.
What to watch during and after
During: spend pacing, CPC, conversion rate against your forecast, budget-limited status, search terms, and whether the site is actually working. Most of the disasters we’ve cleaned up were not bidding problems. They were a broken promo code or a product feed that went out of stock at 11am.
After: write down what you set, what you expected, and what actually happened. Include profit and not just revenue, because a heavily discounted event can post a fine ROAS and a poor P&L. That one paragraph is worth more to next year’s planning than any blog post, including this one.
FAQ
What are seasonality adjustments in Google Ads? They’re a scheduled instruction telling Smart Bidding to expect a higher or lower conversion rate during a specific future window, so it can adjust bids for a short event instead of waiting to learn from the data after the fact.
Do they raise my bids directly? No. You’re changing Google’s conversion rate expectation. Bids move as a result of that, but you’re not setting a bid multiplier.
Should I use one for Black Friday? Probably not, if efficiency matters. Optmyzr’s three-year study across thousands of advertisers found that BFCM adjustments roughly doubled CPC inflation and cost ten to seventeen points of ROAS, while Smart Bidding detected the spike on its own. If your goal is raw volume rather than efficiency, they did drive higher revenue growth, so it becomes a deliberate trade rather than a best practice.
When are they genuinely useful? When you know something Google can’t: a first-time flash sale, a surprise clearance, a big email push, a niche event with no global signal.
Can I use a negative adjustment? Yes, for a short predictable slump. Google says you don’t need one just because a promotion ended, since the adjustment expires on its own. Use negatives for genuinely quiet windows you can see coming, like a shutdown week when nobody is answering leads.
How long can the event be? Google recommends one to seven days and warns performance may suffer beyond fourteen. Longer than that, use budgets, targets and campaign structure.
Do they work with Performance Max? Yes, PMax supports them on all bid strategies. Search, Shopping and Display need Target CPA or Target ROAS. Travel campaigns aren’t supported.
What if I’m wrong about the number? You overbid or underbid for the whole window, in proportion to how wrong you were. Which is exactly why we set conservative numbers and scope them tightly.
About The Author
