Six weeks before Black Friday, your CFO will ask about paid media readiness. You'll mention seasonality adjustments, because Google recommends them and they sound like the responsible, data-driven thing to do. Here's the problem: across three years and nearly 6,000 accounts, advertisers who used those adjustments paid twice the CPC increase and lost more than 21 points of ROAS compared to those who left Smart Bidding alone.
That's not a rounding error. That's the difference between a profitable quarter and an uncomfortable board conversation.
Optmyzr's analysis of 5,886 accounts during BFCM 2024 found that advertisers using seasonality adjustments saw CPCs rise 34% versus a 17% increase for those who didn't intervene. The adjustment group's ROAS fell 15.7%. The hands-off group saw ROAS rise 5.7%. The pattern repeated in 2022 and 2023.
The Precision Problem Finance Will Understand
Seasonality adjustments work exactly as documented. Google's own help center explains that when you tell Smart Bidding to expect a 50% conversion rate lift, it raises bids by roughly 50%. The algorithm takes your input literally. It does not hedge.
This creates what Optmyzr calls the "precision-risk problem." If you estimate a 50% lift but the actual lift is 40%, you're overbidding by about 7% relative to the true value of that traffic. During BFCM, when volume is enormous and auctions are already competitive, that misalignment compounds fast.
The math is unforgiving: you're paying for certainty you don't have. And the data shows most marketers cannot predict conversion rate lifts to the decimal point required to make adjustments profitable.
The Shrinking Peak Makes Overrides Riskier
Here's the part that makes historical benchmarks dangerous. Among advertisers who didn't use adjustments, conversion rates rose 17.5% over baseline in 2022, 11.9% in 2023, and just 7.5% in 2024. The Black Friday spike is flattening because shoppers now spread purchases across a longer promotional season.
Global-e's 2025 BFCM report confirms this pattern: sales momentum now builds steadily through November, with significant uplift appearing two to three weeks before the actual weekend. The traditional "deal weekend" has become a deal month. Anyone setting adjustments based on what Black Friday "usually" does is calibrating to a peak that no longer exists at the same magnitude.
This isn't speculation. Ad Advance's 2025 analysis across hundreds of Amazon accounts found that traffic surged dramatically during BFCM, but conversion rates remained stable rather than climbing. The sharp distinction between Black Friday and surrounding days has eroded.
Smart Bidding Already Knows It's Black Friday
The core insight from three years of data is almost embarrassingly simple: Google's bidding models are trained on vast amounts of historical data. They know when Black Friday is. They know consumer behavior shifts globally during this week.
For advertisers without seasonality adjustments, Smart Bidding consistently detected the conversion rate spike, increased bids appropriately to capture volume, and kept ROAS stable or improved it. The algorithm behaved rationally without human intervention.
The adjustments did buy volume. In 2022, advertisers using them grew conversion value about 25 points more than everyone else. But by 2024, that advantage had shrunk to about 6 points, and it cost them a ROAS swing of more than 21 points. That's a lot of efficiency to sacrifice for marginal incremental revenue.
Guardrails, Not Overrides
The alternative isn't passivity. It's setting constraints that let Smart Bidding do its job while protecting against runaway costs.
Budget pacing matters more than bid overrides. Optmyzr's BFCM guide notes that without pacing, budgets can vanish before peak shopping windows. In 2024, orders accelerated midday on Black Friday and again in the evening on Cyber Monday. Running out of budget at 2 p.m. means missing the second wave entirely.
Max bid limits need attention before the week starts. If your limits are too low, Smart Bidding can't respond when CPCs climb. It will try to increase bids, but too little, too late.
ROAS targets should reflect what you actually achieved last year, not what you hope for this year. Setting an aggressive target during a period when the algorithm is already adapting to higher conversion rates creates conflicting signals.
The principle is straightforward: give Smart Bidding room to operate within boundaries you've defined, rather than trying to predict exactly what will happen and forcing the algorithm to execute your guess.
The Board-Ready Version
If you need to explain this to a CFO or board, here's the summary: seasonality adjustments are designed for unpredictable events where Smart Bidding lacks historical data. Black Friday is the opposite of unpredictable. It's the most anticipated, most analyzed retail moment of the year. Google's models have years of global data on exactly this week.
When you apply an adjustment, you're telling an algorithm with billions of data points that you know better than it does about a pattern it has observed thousands of times. The data says you probably don't.
The 21-point ROAS gap isn't because seasonality adjustments are broken. They work exactly as designed. The problem is that the design assumes you have information the algorithm lacks. For BFCM, you don't.
What to Do in the Next Six Weeks
Review last year's BFCM data by day, not just the aggregate. Look at search impression share lost to budget. If you saw double-digit losses, that's your actual problem, not bid strategy.
Set budget pacing rules that ensure you're live through peak hours. Check your max bid limits and open them up if they're constraining Smart Bidding's ability to compete.
Leave the seasonality adjustment toggle alone. Let the algorithm that's been watching Black Friday for years do what it's trained to do.
The CFO-safe answer isn't "we're using Google's recommended settings." It's "we're not overriding a system that has more data than we do on the most predictable retail event of the year." That's the math-backed position. And the data says it's worth 21 points of ROAS.
