Because ROAS measures revenue, and you are not paid in revenue. An account can hold its ROAS target while shifting budget towards low-margin products, high-return lines, discounted baskets or existing customers who would have bought anyway. The target is being met; the mix underneath it has changed.
This is the most common cause and the hardest to see. Smart bidding is not choosing products, it is choosing whatever hits the target most reliably — and the easiest sales are usually the cheapest, most discounted, lowest margin ones.
The target is met every week while the composition of the revenue quietly deteriorates. Nothing in the Google Ads interface shows this, because the interface does not know what anything costs you.
The check is to segment revenue by margin band over time. If the low-margin share is climbing, the account is doing exactly what it was told and the instruction was wrong.
Returned orders stay in your conversion data unless you actively remove them. Categories with high return rates therefore look far better in Google Ads than they are, and bidding steadily favours them.
Clothing and footwear are the obvious cases, but any range with fit, colour or size variation behaves the same way.
Brand search converts cheaply from people who were coming anyway. Mixed in with everything else, it lifts blended ROAS and hides generic performance underneath it.
If brand volume has grown — through a PR hit, a Meta campaign, or seasonality — blended ROAS improves while the part of the account you are actually paying to acquire from gets worse. Separating brand from generic is the only way to see it.
A promotion lifts conversion rate, so ROAS holds or improves, while margin per order falls. Run enough promotions and the account looks steady on the dashboard and thinner in the bank.
The tell is average order value holding up while contribution per order falls.
If existing customers are converting through paid search, they inflate ROAS without adding a customer. New customer acquisition can be flat or falling while the headline number looks fine.
Google's new customer acquisition reporting exists for this. Without it you cannot tell growth from repeat.
In order: get margin into the measurement, split brand from generic, exclude or discount returns, and only then reset targets. There is very little point scaling an account that is measuring the wrong thing, so the order matters more than the strategy.
None of this requires rebuilding the account in week one. It requires knowing what a sale leaves behind before deciding what to bid for it.
Raising the target reduces volume, which usually reduces profit as well as revenue. It treats the symptom. The cause is that the target is set against the wrong number, so the account is optimising precisely towards the thing you do not want more of.
Not natively. Either send margin as the conversion value so the ROAS column becomes profit based, or write margin bands into custom labels in the feed so Shopping and Performance Max can be segmented and reported by profitability.
It can, because it decides the mix for you across Shopping, Search, Display and YouTube and reports it as one number. That is not an argument against Performance Max. It is an argument for giving it profit as its objective rather than revenue.
Not sure where your budget is going?A free Google Ads audit, written down and walked through on a screenshare. You keep the findings either way.
Request an auditThirty minutes, no obligation and no sales script. If I am not the right fit I will say so on the call and point you somewhere better.
I will reply within one working day. If you would rather not wait, pick a time that suits you now.
Book a time now30 minutes, no obligation.