Google Ads Bid Strategy: The Target Decides

The account came with one Search campaign, a Target ROAS of 400 percent, and a handover note saying it had been set "at launch". That was fourteen months earlier. Nobody in the room could say where 400 came from. The campaign had been running at 620 percent ever since, which the client read as good news.

It was the opposite. Google had been holding volume back to protect a number nobody believed in.

Short answer: A Google Ads bid strategy is not an algorithm you pick, it is a constraint you set. Maximize Conversions spends your budget whatever a result costs. Target CPA and Target ROAS hand the same system a ceiling. The choice that matters is the number, and whether your conversion data can carry it.

The takeaways

• The ceiling is the strategy. Maximize Conversions and Target CPA drive the same machine; the difference is whether you typed a number into it. Google split those labels back apart in June 2026, and Search Engine Journal called the split a labeling change only.
• 30 conversions is a reading threshold. Google's 30 (and 50 for Target ROAS) is the window you need before you can judge a result with confidence, rather than a gate you must pass before switching strategies.
• Go read your targets before 17 August 2026. Budget-limited campaigns on a target strategy could previously beat their target. After that date they track it, so a €10 Target CPA running at €5 drifts up toward €10.

What does a bid strategy control?

It controls one thing: how much Google may pay for a result, and against which definition of a result. That gives you two axes. The first is what you are counting, conversions or conversion value. The second is whether you imposed a ceiling. Four boxes, not eight strategies.

Maximize Conversions is the count axis with no ceiling. Target CPA is the count axis with one. Maximize Conversion Value and Target ROAS are the same pair on the value axis. Everything else in the dropdown optimizes toward something that is not a result: Maximize Clicks buys traffic, Target Impression Share buys visibility, Manual CPC buys the illusion of control over an auction that resolves in milliseconds. Sorting the dropdown into those four boxes is most of the work.

Should you use Maximize Conversions or Target CPA?

Ask a different question. The two are the same system with and without a ceiling, so the decision is whether you know what a result is worth to you yet. If you do not, a ceiling is a guess dressed as a constraint, and it will starve delivery for no reason.

The evidence sits in Google's own labeling. Target CPA and target ROAS had been folded into the Maximize strategies, then in June 2026 Google split them back out as standalone options. Search Engine Journal, reporting on the change in July 2026, called it a labeling change only, with the behaviour underneath unchanged.

So the practical order is: run uncapped first, watch what a conversion costs across a few weeks, then decide whether that price needs a lid. Setting a target on day one is setting it from nothing.

How many conversions do you need before a target?

Fewer than you have been told. The 30-conversions-in-30-days rule gets repeated as an entry requirement for Target CPA, with 50 quoted for Target ROAS, and that framing is wrong in a way that costs small accounts real time.

Search Engine Journal frames those figures as the volume you need over an evaluation window to read performance with confidence, rather than a threshold you must clear before the strategy will work. An account producing eight conversions a month can run Target CPA. What it cannot do is look at last month and conclude anything, because eight events swing enough to tell any story you want.

So volume does not gate the switch. It gates your right to draw a conclusion afterwards, which most guides never separate.

What does switching a bid strategy cost you?

A week or two of numbers you cannot read. Every strategy change puts the campaign back into calibration, and the CPA usually rises before it settles. That period is not wasted spend, but it is unusable evidence.

The trap is what people do next. The numbers look bad, so they change something else, which restarts the clock. An account that switches strategy monthly never leaves the noisy stretch and never gets a clean read on anything it tried. I have watched a good Target CPA get abandoned on day nine, twice, in one quarter.

The rule is boring: change one thing, then sit still for a full window. If you cannot commit to leaving it alone that long, do not switch yet. The Google Ads learning phase is the mechanism underneath this.

Why can a met ROAS target still lose money?

Because the target is measured against the conversion value Google reports, and that number is not your revenue. It carries modeled conversions, view-through credit, and whatever value you passed in at the tag, which for most accounts is order value rather than margin.

Hit a 400 percent target on a product with a 30 percent contribution margin and you are underwater before overhead. The strategy did its job. The instrument you pointed it at measured the wrong thing. This failure mode survives every strategy change, because switching from Target ROAS to Target CPA moves the same flawed value into a different formula.

Set the target against your own break-even ROAS, then treat the platform's reported figure as a directional signal you reconcile against real revenue monthly.

What changes on 17 August 2026?

Budget-limited campaigns stop overdelivering against their target. Until now, a campaign constrained by budget rather than by its target could come in well under the target CPA you set. From 17 August 2026 those campaigns track the target you typed, including when you adjust budgets.

The German agency Die Online Experten, writing about the change in July 2026, uses the plain example: a €10 target CPA on a campaign that has been delivering at €5 will move toward €10. Nothing broke. The number you set months ago simply started binding.

Go open every budget-limited campaign this week and read the target as if you were typing it for the first time. Plenty were set at launch by someone no longer on the account.

Deciding that number is the part no bidding strategy can do for you. It is what Adscalr's budget intelligence is built around: three to five prioritized campaign plans drawn from twelve weeks of your real performance, each with a conversion goal and a CPI estimate labelled as an estimate, so the number you type has something behind it.