Manual control versus algorithmic optimization tradeoffs
Manual CPC (cost-per-click) gives you the tightest control. You set your maximum bid per keyword, and Google won't exceed it. This works well when you understand your margins precisely and want predictable spending. But you have to adjust thousands of bids manually as markets shift, and you leave efficiency on the table because you can't react to every auction moment. Target CPA (cost-per-action) flips the equation. You tell Google your acceptable cost per conversion, and the algorithm bids to hit that target while maximizing volume. It requires historical conversion data to train on, but once trained, it adapts in real time to auction conditions you can't track yourself.
Target ROAS (return on ad spend) builds on Target CPA but optimizes for revenue, not just conversions. If your products have different margins, ROAS adjusts bids higher for high-margin conversions and lower for low-margin ones. Max Conversions is a full-automation play: Google bids to maximize the count of conversions within your daily budget, no target in mind. It's best for top-of-funnel work where you want volume and don't have a fixed cost tolerance.
Choosing based on data maturity and margin predictability
Manual CPC suits teams running new campaigns with limited history or those with extremely tight, non-negotiable margins. It's also the fallback when your conversion tracking is noisy. Target CPA requires at least 15-20 conversions per month in the learning phase but becomes more efficient than manual as spend scales. Target ROAS is premium pricing: you're handing the algorithm your margin structure, so it only works if your product costs and prices are stable.
Max Conversions is the starting point for brand-new accounts. Feed it a budget, let it learn for 2-4 weeks, then migrate to a CPA or ROAS strategy once you have enough conversion volume. The biggest mistake is mixing strategies across campaigns and comparing their efficiency too early. Each strategy needs 4-6 weeks of data to stabilize its bidding behavior. Patience beats constant optimization attempts.