Target ROAS Explained: How the Target Affects Bids, Revenue and Testing

Question 1

What is target ROAS?

Accepted Answer

According to the article, Target ROAS is the return Google aims for across the bidding strategy by predicting conversion value for an auction and then adjusting the bid in pursuit of the target. It is not an average Google promises to deliver and it does not impose a literal minimum return on every click. Calling the target a floor is useful shorthand, but the article states the analogy breaks because a higher target mainly makes bids less competitive.

Question 2

Why should I fix conversion values before changing Target ROAS?

Accepted Answer

The article states Google estimates predicted value as the chance of conversion times expected value, for example a 4% chance of a $50 order has an expected value of $2.00, while another auction might be $0.40 or $6.00. That estimate depends on values sent back, so passing every lead as $100 hides the difference between a $29 filter replacement and a $900 install, just as a fixed 50 euro value hides a 20 euro item versus a 300 euro order. The article cites an audit where duplicate tracking lifted reported ROAS from 6.0 to 7.0 with no extra sales.

Question 3

What does raising Target ROAS do to my bids?

Accepted Answer

According to the article, the working model is bid equals predicted value divided by Target ROAS, so with predicted value held steady a higher target means cheaper bidding. Under that model a $50 predicted value at a 500% target produces a $10 bid. For $12 in predicted value, a 300% target gives a $4.00 bid, 500% gives $2.40, and 800% gives $1.50, so if winning requires a $3.00 bid only the first is competitive.

Question 4

Why can my reported ROAS go up while revenue goes down?

Accepted Answer

The article states a higher target can make bids too low to win auctions, so spend may fall even with budget room and remaining traffic may lean toward brand searches and past buyers. In its example, $20k spend recording $64k revenue is 320% blended ROAS, then after raising the target to 500% spend falls to $11k with about 460% ROAS on roughly $50k in revenue. The ratio is higher while revenue is about $14k lower.

Question 5

Should I judge Target ROAS by average ROAS or marginal ROAS?

Accepted Answer

The article distinguishes blended ROAS, total recorded value divided by total spend, from marginal ROAS, what the next or removed stretch of spend earns. In its example Campaign A shows 1.5x average but only 0.8x on its next $20k of spend, or $16k of revenue, while Campaign B shows 1.1x average and still returns 1.1x on its next increment. For thresholds, the article states break-even ROAS equals 1 divided by contribution margin, so at 45% margin that is about 2.22x, and retaining 10 percentage points as profit makes the threshold about 2.86x.

Question 6

How do I test a Target ROAS change to find marginal return?

Accepted Answer

The article recommends a 50/50 experiment for steady-volume campaigns, keeping the control at its current target and lowering only the test side by 10% to 15% with other settings aligned. It suggests a four-week read on conversion value and impression share and checking the roughly 15-conversion, 30-day Search and Shopping reference point before testing. At readout, it states to calculate (test value minus control value) ÷ (test cost minus control cost) as the estimate of marginal ROAS, as in $4,000 extra spend returning $9,200 for 230% marginal ROAS versus $13,600 for 340%.