The split you're seeing between larger and smaller accounts is exactly right, and the breakeven point I've found is around 30-50 conversions per month in a single campaign before automated bidding starts reliably outperforming manual. below that threshold, tCPA has too little signal to work with. It makes predictions based on noise and you get erratic spend sometimes underspending because it's being overly cautious, sometimes blowing budget on a cluster of traffic that happened to convert twice in a row. Manual CPC with tight exact/phrase keywords keeps you in control during that data-building phase. above 50 conversions a month, automated starts paying for itself because the algo can actually spot patterns you'd never find manually device, time, audience overlap signals that don't show up cleanly in the reports. That's when handing over bid control makes sense. one thing worth checking on your smaller accounts: if conversions are spread across multiple campaigns, the algorithm sees them as separate pools. Consolidating into fewer campaigns can push individual campaigns over that threshold faster and unlock automated bidding earlier than you'd expect. my sequence for new accounts: manual CPC to build conversion history, tCPA once I hit 30 conversions in 30 days, then reassess at 60-90 days whether Maximize Conversions makes sense. Never skip the manual phase regardless of what Google recommends in the interface.
