Target CPA vs. Maximize Conversions: Which Should You Choose?

I've watched this argument play out in maybe fifty Google Ads accounts. Target CPA or Maximize Conversions. People treat it like a fork in the road where one path is right and one is wrong, and they'll spend a week reading guides to pick correctly.

Then they pick correctly, set it up correctly, and the campaign still underperforms. Every time the conclusion is the same: wrong strategy, switch to the other one. So they switch. It still underperforms.

Here's the blunt read. Target CPA versus Maximize Conversions is a real question, and I'll answer it properly below. But it's the second question. The first one - the one that decides whether either strategy works - is whether the conversions you're feeding Smart Bidding are real. Both strategies optimize toward conversion data. If 24 to 31 percent of that data is bots or noise, you are tuning a model against a corrupted baseline, and no bid strategy fixes that.

This is not a bidding-strategy post pretending the signal is clean. It's a post about the signal first. DataCops gets one mention, as the architecture that fixes the signal. Then the actual comparison.

Quick stuff people keep asking

Should I use Target CPA or Maximize Conversions in Google Ads? Maximize Conversions when you're new, have little conversion history, and want Google to gather data fast. Target CPA once you have enough conversions to know a profitable cost per acquisition and need to hold the line on it. That's the textbook answer. It assumes your conversions are real.

When should I switch from Maximize Conversions to target CPA? Rough rule: once the campaign has cleared the learning period and is logging a steady volume of conversions - many practitioners use 30-plus in 30 days as a floor - and you can see a stable, profitable CPA in the data. Switch before that and Target CPA throttles you on too little signal.

Is Target CPA the same as Maximize Conversions with a target? Functionally, close. Google folded a target-CPA field into Maximize Conversions, so "Maximize Conversions with a target CPA" behaves much like classic Target CPA. The distinction is now more of a UI label than two separate algorithms.

How many conversions do I need before setting a Target CPA? Google's old guidance hovered around 30 conversions in 30 days. More matters than the floor - and more importantly, the conversions need to be genuine. 30 conversions where 10 are bots is not 30 conversions. It's 20, plus 10 lies.

Does Maximize Conversions spend your full daily budget? Yes. That's the defining trait. Maximize Conversions will spend every dollar of the budget chasing volume. If your budget is set loosely, it'll happily spend it on low-quality conversions to hit the count.

What happens to my bids if I increase budget on Maximize Conversions? Bids tend to spike. The strategy has more money to deploy and pushes into more expensive auctions to use it, so your CPCs climb and CPA often climbs with them. Scaling Maximize Conversions is where a lot of accounts get hurt.

Which bidding strategy is better for new Google Ads campaigns? Maximize Conversions, generally. New campaigns lack history, and Maximize Conversions gathers data aggressively. The risk: aggressive data gathering on a contaminated funnel just gathers contaminated data faster.

Why is my Target CPA campaign not spending? Usually the target is set too low for the auction, so Google can't find inventory that hits it. Sometimes thin conversion history. And sometimes the real CPA is fine but bot conversions made historical CPA look artificially cheap, so your target is anchored to a number that was never real.

Both strategies optimize toward conversions. What if the conversions are fake?

Strip away the marketing language and Smart Bidding is one loop. It looks at which clicks converted, builds a model of what a converter looks like, and bids more on traffic that resembles them. Target CPA does it with a cost ceiling. Maximize Conversions does it with a volume goal. Same loop, same fuel: your conversion data.

Now the part the comparison guides skip entirely. That fuel is dirty.

Of the conversion-adjacent traffic that gets collected, 24 to 31 percent is bots. Datacenter IPs, automated agents, click farms, scripted junk. On the other side, 25 to 35 percent of analytics and conversion events are blocked before they ever arrive - uBlock, Brave, Safari, extensions. So Smart Bidding is learning from a dataset that's simultaneously inflated with fake conversions and missing a quarter of the real ones.

Feed that into the loop. The model studies your "converters," and a chunk of them are bots. So it learns that bot-like traffic converts. Then it does its job - it bids up to find more traffic like that. Target CPA does it within a cost ceiling. Maximize Conversions does it to maximize the count. Either way, the algorithm is now actively, efficiently buying you more bots, because you told it bots were customers.

This is why "correct" setups underperform. The bidding strategy isn't broken. It's executing perfectly against a corrupted definition of success.

Let me make it concrete. PillarlabAI ran a honeypot signup flow and watched what came through. 3,000 signups. 77 percent fraudulent. 650 of them traced to a single device fingerprint - one machine wearing 650 faces.

Drop that into a Google Ads account. Those signups fire as conversions. Smart Bidding ingests them, with no idea 650 came from one device. It builds a converter profile heavily shaped by that fraud. Then it goes hunting for more of the same. Your conversion count looks great. Your Target CPA looks like it's holding. And your actual customer acquisition is a rounding error, because the algorithm has spent two weeks optimizing toward a ghost.

That's the foundational failure. Picking Target CPA over Maximize Conversions when the signal is contaminated is choosing how you'd like to lose money, not whether.

The fix is upstream of the bid strategy

You can't clean this inside Google Ads. By the time a bot conversion shows in the interface, it's already trained the model. You can exclude placements and add negatives all day - that's reacting after the fact, and the learning already happened.

The fix is to stop the bad conversion from being counted as a conversion in the first place. That means filtering at the point of collection: scoring each conversion event against IP reputation, device fingerprint, and behavior before it's recorded and before it's sent onward through the conversion API. A bot signup gets flagged at ingestion and never enters the conversion stream Smart Bidding learns from.

That's the architecture DataCops is built for - first-party collection with bot filtering at ingestion, an IP database over 361.8 billion addresses sorting residential from datacenter from VPN from proxy from Tor, and a clean CAPI feed to Google so the conversions the algorithm sees are the conversions that were real. Get that right and the Target-CPA-versus-Maximize-Conversions question finally becomes a real strategy decision, because both strategies are now optimizing toward humans.

Decision guide

Brand-new campaign, little to no conversion history. Maximize Conversions to gather data - but verify your conversion source is filtered first, or you're just gathering contaminated data quickly.

Mature campaign, 30-plus genuine conversions a month, known profitable CPA. Target CPA. You have the signal and a number worth defending.

Target CPA campaign won't spend. Check if the target is too tight. Then check whether historical CPA was made artificially cheap by bot conversions - you may have anchored to a fake number.

Scaling a Maximize Conversions campaign. Expect CPC spikes when you raise budget. Raise gradually, and watch CPA, because the strategy will buy volume at any quality to use the money.

"Correct" setup, still underperforming. Stop reswitching strategies. Audit conversion data quality. The problem is almost certainly the signal, not the bid model.

Heavy paid acquisition as your main channel. Conversion signal integrity is your single highest-leverage fix. Both bid strategies amplify whatever you feed them - so feed them filtered data.

You're tuning the engine while the fuel is contaminated

The mistake is treating Target CPA versus Maximize Conversions as the lever that decides performance. It isn't. It's a lever that decides how Smart Bidding pursues conversions - not whether the conversions are worth pursuing.

Both strategies are obedient. They optimize toward exactly what you label a conversion. Label bots as conversions and both will, with total competence, go buy you more bots. The strategy debate is real, but it lives one floor up from the foundation, and the foundation is signal integrity.

So before you switch strategies again, do this. Pull your last 200 conversions. Check the IPs - how many are datacenter? Check device fingerprints - how many conversions share one? If you find a cluster, you found why your "correct" setup underperforms. It was never the bid model. It was the data underneath it.

What share of your conversions are real - and have you ever actually counted?