Google Smart Bidding, Explained for People Who Pay for It

When you run Google Ads with Smart Bidding, you're not setting bids anymore. Google's system sets a bid for each individual auction, based on the context of that exact moment: the search query, device, location, time of day, browser, audience membership, and dozens of other signals. A bid that makes sense for a ready-to-buy searcher on their phone near your store makes no sense for a casual desktop browser at midnight — Smart Bidding knows the difference, if you've taught it properly.

The strategies, in plain English

• Maximize Conversions — spend the budget to get as many conversions as possible, no cost target. The usual starting point while you gather data.
• Target CPA (tCPA) — tell Google the average cost per conversion you'll accept. It bids up when a click looks likely to convert cheaply, down when it doesn't, aiming to average out at your target.
• Maximize Conversion Value — like Maximize Conversions, but optimizing for total revenue value instead of count.
• Target ROAS (tROAS) — tell Google the return you want (conversion value ÷ ad cost). It prioritizes high-value conversions and may skip cheap ones that dilute the return.

The fuel: conversion tracking quality

Smart Bidding is an optimizer. It will very efficiently chase whatever you define as a conversion — including the wrong thing. If your tracking fires on a page view instead of a purchase, counts the same sale twice, or never records phone calls, the system will spend your budget finding more of those. Garbage in, garbage out — at machine speed.

Before any Smart Bidding strategy can be trusted: track the actions that actually make you money (completed purchases with real values, submitted forms, calls), deduplicate them, and use enhanced conversions where you can. For tROAS, the values matter as much as the count — a system optimizing toward values you made up will make decisions you'll regret.

Data volume: the part nobody wants to hear

Smart Bidding learns from conversion history, so volume matters. Google's own guidance is that tCPA works best with roughly 30 conversions in the past 30 days in the campaign, and tROAS with around 50 conversions with value over a similar window. Below that, predictions get shaky — the usual play is to run Maximize Conversions first (or pool campaigns that share a goal) until there's enough signal, then add a target.

Budget matters too. If your daily budget is barely above your target CPA, the system has almost no room to explore auctions and learn. A budget several times your target cost per conversion gives it the swings it needs to find the good ones.

Learning periods are not a malfunction

After you launch a Smart Bidding strategy — or make a big change to targets, budgets, or conversion definitions — the campaign enters a learning period, typically about one to two weeks depending on volume. Performance will wobble while the system re-tests the auction landscape. The classic mistake is panic-editing every two days: each significant change restarts the learning, so the campaign never stabilizes. Set it up right, then judge it on a full learning cycle, not on Tuesday's mood.

What you still control

• The conversion definition — what counts, and what it's worth. This is the strategy, everything else is tactics.
• Structure — which campaigns and goals to pool so each strategy gets enough data.
• Budgets and targets — realistic ones, changed gradually.
• Ads and landing pages — Smart Bidding picks the moment; your ad and page still have to close the deal.

Smart Bidding isn't magic and it isn't a scam — it's a power tool. Pointed at accurate conversion data with enough volume, it out-bids any human. Pointed at broken tracking, it burns money efficiently. We set up the tracking first, then let the machine work — see AI Google Ads management.