Google Ads Bidding Strategies: When To Use tCPA And tROAS

Nearly every account I inherit has at least one campaign on the wrong bid strategy. The classic is target ROAS switched on for a campaign producing four conversions a month, and an advertiser who cannot work out why impressions fell off a cliff in week two.

Here is the short answer. Run Maximise conversions while you are building data and every lead is worth roughly the same. Add a target CPA once volume is steady and you have a cost you need to hold. Move to target ROAS only when your conversions carry genuinely different values and you are clearing at least 15 of them in 30 days, which is Google’s stated minimum for Search and Shopping. This post covers what each strategy does, the data each one needs, the order to graduate through them, and the settings that decide whether Smart Bidding works for you or quietly drains the budget.

What counts as Smart Bidding, and what doesn’t

Google reserves the term Smart Bidding for four strategies: Maximise conversions, Maximise conversion value, target CPA and target ROAS. These are the ones using auction-time bidding, which means a bid is calculated for every individual auction using signals like device, location, time of day and audience.

Maximise clicks and target impression share are automated, but they are not Smart Bidding. They optimise for traffic and visibility. Neither has any idea whether a click turned into money. Manual CPC still exists. Enhanced CPC does not: Google migrated existing Search and Display campaigns off it to Manual CPC in March 2025.

One naming note, because it confuses people mid-conversation. From June 2026 Google relabelled the strategies in the interface. “Maximise conversions with a target CPA” is now just Target CPA, and “Maximise conversion value with a target ROAS” is now Target ROAS. The bidding behaviour underneath is identical. Nothing changed except the label.

The strategies in plain terms

Strategy	Optimises for	Use it when
Maximise clicks	Volume of clicks	You are opening a new market and need traffic to learn from, or you have no conversion tracking worth trusting
Target impression share	Where your ad sits	Brand defence, or a competitor is squatting on your name
Maximise conversions	Conversion count, spending the budget	New conversion-focused campaigns, thin data, every conversion worth about the same
Target CPA	Conversion count at an average cost	Lead gen with steady volume and a cost you have to hold
Maximise conversion value	Total revenue, spending the budget	Ecommerce with real value tracking and headroom to spend
Target ROAS	Revenue at a set efficiency	Ecommerce or value-based lead gen where margin matters more than volume

The split that matters is not automated against manual. It is count against value. If a £40 sale and a £400 sale are the same event to your bidding, you are on a count strategy and you will get more of whatever is cheapest. That is fine for a plumber. It is expensive for a retailer.

How much conversion data each one needs

This is where most bad switches happen. People pick the strategy that matches their ambition rather than their data.

• Target ROAS, Search and Shopping: at least 15 conversions in the past 30 days at the conversion tracking level. Values must be greater than zero to count. Google recommends 50 conversions in 30 days before you judge performance.
• Target CPA: no published hard floor. Google’s guidance is to evaluate over at least 30 days and around 30 conversions before drawing conclusions.
• Maximise conversions: works from a standing start, which is exactly why it is the right first move on a new campaign.

My own floors sit above Google’s. Fifteen conversions is the point where target ROAS is permitted, not the point where it is a good idea. On Search I want 30 or more a month before I hand the algorithm a value target, because at 15 a single refund or one freak order swings the whole average and the bidding chases the noise.

The other half of the requirement is tracking quality. Smart Bidding optimises to whatever you feed it. Send it every newsletter signup as a conversion and it will buy you newsletter signups beautifully. Before changing a strategy, open your conversion actions and check what is actually counted in the Conversions column.

The decision path I use

1. Is conversion tracking correct and counting only actions worth money? If not, stop. Fix that first. Nothing below works without it.
2. Under 15 conversions a month? Maximise conversions. Let it run. Resist the urge to add a target while volume is thin, because a tight target on thin data just throttles impressions.
3. Steady volume, all conversions worth roughly the same? Target CPA. Set the target at your current achieved cost, not your target cost.
4. Conversion values genuinely vary? Move to value. Maximise conversion value first if you want to spend the full budget, target ROAS once you need efficiency more than volume.
5. Efficiency target hit and you want more scale? Loosen the target before raising the budget. A tROAS of 400% that spends half the budget is not beating a tROAS of 320% that spends all of it, unless your margin says otherwise.

Step 5 is the one people skip. They raise budget against a target the account cannot reach and then wonder why the extra money never spends.

Setting the target, and fixing it when it’s wrong

Set the opening target at what the campaign already achieves over the last 30 days. If your CPA has been £62, set £62. Not £40 because £40 is what finance asked for. A target well below current performance tells Google to bid only on the auctions it is most certain about, and certainty is rare, so the campaign stops showing.

Then move in steps of roughly 10% to 15% and wait a couple of weeks between moves. Bigger jumps push the strategy back into learning and you lose the read on whether the last change worked.

Two things to hold in mind while you do it. Budget-capped campaigns can spend up to twice their average daily budget on any single day, so a strategy that looks like it has overspent may simply be balancing across the month. And if a campaign is limited by budget, your bid strategy is not the constraint. Fix the budget before blaming the target.

The controls most advertisers never touch

Three tools sit in Tools > Shared library > Bid strategies, and almost nobody uses them.

Seasonality adjustments. Tell Smart Bidding to expect a conversion rate change for a defined window, like a three day sale. They are built for short events of one to seven days and lose their usefulness beyond about a fortnight. Do not use one for Q4. Use one for Black Friday weekend.

Data exclusions. If your tag broke for 48 hours, Smart Bidding reads those two days as genuine failure and recalibrates on a lie. A data exclusion tells it to ignore that window. Use it for tracking outages only, never to hide a bad week.

Portfolio bid strategies. One strategy applied across several campaigns so they pool their conversion data. This is the honest fix for the account with six small campaigns that each sit below the volume threshold. Group them rather than pretending each one has enough data on its own.

And read the bid strategy report before you change anything. It shows what the strategy is actually targeting against what it is delivering, which is usually a faster diagnosis than staring at the campaign table.

What to check, and how often

Weekly, look at spend against budget and actual CPA or ROAS against target. That is a five minute check. Do not act on it unless something has moved hard.

Monthly is when you make changes: one target adjustment per campaign, then leave it alone. If you moved the target, the budget and the match types in the same week, you have learned nothing about any of them.

Worth knowing right now: Google has flagged that bidding system updates from mid-August 2026 may cause temporary performance swings on budget-limited campaigns. If your numbers wobble in the next few weeks and nothing in the account changed, wait a full cycle before reacting.

Not sure your bidding matches your data

Most bidding problems I find are not the strategy itself. They are a target set from a spreadsheet rather than from the account, or a conversion action counting things nobody would pay for. If you want a straight read on which of those is costing you, take the free Google Ads audit and I will tell you what I would change and in what order. If you would rather hand the whole thing over, that is what our Google Ads management does. Running paid social too? The logic is different over there, and I covered it in Meta ads bidding strategies.

FAQ

How long does the learning period last after changing bid strategy? Plan for one to two weeks of unstable performance, and longer on low-volume campaigns. Judge the change over at least 30 days rather than the first few days. Changing the target again mid-learning restarts the process and wastes the spend already used.

Should I use target CPA or target ROAS for lead generation? Target CPA for most lead gen, because leads usually enter Google Ads with no value attached. Move to target ROAS only if you pass real values back, for example by importing closed deal values from your CRM. Without values, tROAS has nothing to optimise against.

What is the minimum conversion volume for target ROAS? Google requires at least 15 conversions in the past 30 days for Search and Shopping campaigns, with conversion values greater than zero, and recommends 50 before you evaluate performance. Below that, use Maximise conversions or target CPA instead.

Does Manual CPC still work in Google Ads? It still exists and it still runs, but you give up auction-time bidding, which is where most of the gain sits. Manual CPC is defensible on tiny budgets or heavily restricted accounts. On anything with conversion tracking and steady volume, an automated strategy will beat it.

Why did my spend drop after switching to target CPA? Almost always because the target is below what the campaign was actually achieving. Google restricts bidding to the auctions it is confident can hit that cost, so impressions fall. Reset the target to your recent achieved CPA, let it stabilise, then tighten in steps of 10% to 15%.