Why can actual ROAS decline month over month despite unchanged campaign settings and biddings, especially during a predictable shift from peak-season demand to non-peak-season demand?

Last month a campaign was hitting 220% on a 200% target. Two weeks into this month — same target, same budget, same campaign — actual ROAS dropped to 170%. If I can’t explain it, I can’t act on it before the drop hits.

Answers from my research

The campaign settings stayed the same, but the demand environment didn’t. Three things shift between peak and non-peak:

1. The pool of high-intent buyers shrinks. Same targeting, weaker audience underneath it.
2. Conversion rate drops as a result. Same clicks, fewer turn into revenue.
3. Smart Bidding lags. Google’s algorithm is still bidding aggressively based on last month’s strong conversion data — it’ll recalibrate, but in those first weeks it’s overpaying for traffic that’s no longer worth what it was.

Three angles on the same underlying truth: same settings, different market.

My move when I see this is to raise target ROAS. A higher target ROAS forces Google to be more selective. It stops bidding on the marginal auctions where conversion rate is weakening, and narrows back to only the strongest signals. That protects actual ROAS even as the broader market softens.

The trade-off is: fewer customers, lower volume. But the customers you get are the high-intent ones, and the ROAS holds.

What I learned

When I can predict a demand drop — end of peak season, post-promo period, predictable seasonal lull — I raise target ROAS before the drop hits, usually by 10–25 percentage points (e.g., from 200% to 220–225%).

I accept fewer conversions for the period. Weak performance also feeds the algorithm’s learning, which can make recovery slower when demand returns — so protecting ROAS now also protects the campaign’s bounce-back later.

When demand comes back, I lower the target back to where it was. The campaign stays profitable across the full cycle.