POAS automation needs commercial judgement first
Google Ads can optimise only for the conversion values it receives. If those values reflect revenue alone, a healthy reported ROAS does not tell you whether sales cover product costs, fulfilment, returns, payment fees and advertising.
Start with a trustworthy contribution measure
Contribution is revenue less cost of goods, fulfilment and shipping, returns and payment fees, before advertising. POAS relates that contribution to ad spend. Neither POAS nor platform attribution proves that advertising caused the sale. The underlying cost and return data must be current enough for the decision being made.
Know what the bidding signal does not know
Margin-aware values can help avoid favouring a high-revenue product with weak contribution. They do not settle whether demand is branded, whether a customer is new, or whether the business has the cash and stock to serve more orders. Compare attributed performance with the wider commercial picture before moving budget.
Keep product decisions with people
BOI® describes five product jobs: Scale, Profit, Protect, Recovery and Gateway. Recovery concerns ageing or overstocked inventory becoming cash within a contribution floor and time box; low stock is a constraint on spend. A person decides when a job changes. Automated bidding can execute a decision, but it should not define the job or invent the commercial limits.
Ask for accountable reporting
Ask which costs enter conversion values, when returns are reconciled, who owns changes, and how results are compared with finance records. The implementation rules, scripts, thresholds and account-specific settings are working methods, not a public template.
For the underlying measure, read what POAS includes. For buyer-facing product decisions, see BOI® for ecommerce teams. These explain different questions from this article: the measure, the jobs and the limits of automation.