Check conversion tracking before anything else
Tracking is the most common cause of a ROAS drop and the most invisible. A consent banner update, a Shopify theme change, or a GA4 misconfiguration can cut reported conversions without anything in the market changing. The ads keep producing sales; the account just stops seeing them.
The fast check is a comparison: put conversions reported in Google Ads next to actual orders in your store backend for the same period. If the backend shows healthy sales while the ad account shows a collapse, you have a measurement problem, and cutting spend would punish campaigns that are working.
Separate market shifts from account problems
Once tracking is verified, compare cost per click and impression share year over year, not month over month. Rising CPCs with stable impression share means competitors are paying more for the same auctions, which is a market shift. Falling impression share at stable CPCs means your account is losing ground it used to hold.
Seasonality belongs in this check too. Comparing a quiet month against a peak month tells you about the calendar, not the account. Year-over-year comparison strips the seasonal pattern out and shows whether the account is genuinely weaker than it was.
Review what changed inside the account
Account changes show up in ROAS two to four weeks later, so review everything that happened in that window before the drop:
- Budget increases that pushed campaigns back into learning.
- New or changed bidding targets that forced the algorithm to chase different traffic.
- New campaigns competing with existing ones for the same queries.
- Feed changes, price changes, or out-of-stock bestsellers.
Most self-inflicted ROAS drops trace to a change someone made with good intentions and then forgot about. The change history in Google Ads keeps the honest record.
Why diagnosis must come before action
The wrong response to a ROAS drop usually costs more than the drop itself. Cutting budgets in response to a tracking gap starves campaigns that were profitable. Holding spend through a genuine market shift burns money a diagnosis would have saved.
Work the sequence in order: verify measurement, compare year over year to isolate market effects, then audit recent account changes. Each step rules out a family of causes, and the remaining explanation tells you whether the fix is a tag, a bid target, or patience.