Target ROAS

Target ROAS is the equivalent target field inside Maximize Conversion Value, entered as a percentage - 300% means the algorithm is trying to hold an average of $3 in tracked conversion value for every $1 spent. It only works when conversion values are both accurate and varied; feed it the same flat number on every conversion and it has nothing to differentiate one auction from another, at which point it behaves like Target CPA with extra steps.

Google predicts a value likelihood for each auction and bids up where it expects strong payback relative to the target. Raising the target does not make the algorithm try harder for the same volume - it makes the algorithm hold a stricter ratio by bidding on fewer, higher-value opportunities and skipping the rest, so lower spend at a better ratio is the expected result of raising the target, not a malfunction. Portfolio strategies apply the same target across several campaigns and pool their conversion value data, which can help thinner campaigns learn faster by borrowing signal from higher-volume ones in the same group, though it also means one unusually large or unusually small campaign in the portfolio can pull the shared bidding behavior in its direction.

This strategy fits ecommerce accounts with real variation in basket size and lead-gen accounts with scored lead values, where knowing which conversions are worth more is genuinely useful to the bidding decision. The target itself should be set from the business's actual margin, not just from revenue - a 300% ROAS can still be unprofitable once cost of goods, fulfillment, and other expenses are counted, so the number that goes into this field should reflect what the business can sustain, not what looks impressive in a report. It is a weaker fit for accounts where every conversion is worth about the same, since Target CPA achieves a similar outcome with a simpler setup and one fewer moving part to get wrong.

The most damaging mistake is setting a target the account has never come close to achieving - jumping straight to an aspirational ratio rather than starting near the account's demonstrated average and working up in increments. Because a high target is enforced by bidding on almost nothing, an unrealistic target does not produce gradually better efficiency, it produces a campaign that barely spends at all. A second mistake is treating volume loss after a target increase as a problem to fix rather than the mechanism working as intended - the trade-off between ratio and volume is the point of the setting, not a side effect. A third is changing the target frequently in response to daily noise, which keeps resetting the learning period instead of giving any one setting enough time to show a real trend.

Read Conv. value / cost as the ROAS figure, but judge it on a weekly or monthly trend rather than single days - one unusually large order can swing a day's reported ROAS in a way that has nothing to do with how the bidding is actually performing. It is also worth keeping a known breakeven ROAS on hand from the margin math above, so a reported ratio can be judged against what the business needs rather than against an arbitrary round number.