Google Ads no longer gives you more than you asked for: why a 600% ROAS becomes 400%

On 17 August 2026 Google switched on a change that affects almost every store running on a limited advertising budget. The wording in the help centre sounds harmless: campaigns will “perform more consistently toward your bid target”.

In plain language: if your advertising was doing better than you asked for, it will stop.

The change has been live for over a week now. That means you don’t have to guess at anything — your account already holds “after” data, and all you need to do is open it and compare.

Below I break down what exactly changed, who it affects, how much inaction costs and what to do about it — step by step, no jargon. Checking your own account takes about twenty minutes.

I’ll also say two things people usually keep quiet about. First: if this change hit you, the problem didn’t start on 17 August — you had it before. Second: for most online stores the right answer isn’t to fix the target, it’s to drop it altogether.

This article isn’t for you if:

• you only run video or mobile app campaigns — the change didn’t touch those at all;
• your budget never runs out before the end of the day — then it doesn’t affect you either;
• you’re on “Maximize conversions” or “Maximize conversion value” without a target — nothing changes here. If it’s the latter, you’re already where I recommend everyone else move by the end of this article. If it’s the former, part 7 is still worth reading: for a store it counts the wrong thing.

Part 1. What actually happened

A simple analogy

Imagine handing a buyer $1,000 and saying: “Buy stock, I’m willing to pay up to $20 per unit.”

The buyer goes to the market and finds some of the stock selling at $10. There isn’t enough money for everything, so they take the best deals — and bring back 100 units at $10 instead of 50 units at $20. You said “up to $20” and they came in at $10. Everyone’s happy.

From 17 August this buyer thinks differently: “The boss said up to $20. So I’ll buy at $20.” Same money — $1,000. They bring back 50 units instead of 100.

Technically they didn’t cheat you. You named the $20 figure yourself.

That’s exactly what Google did to Smart Bidding.

Google’s own wording

Here’s the verbatim example from the official help page:

If your campaign’s Target CPA is $10, but your recent actual CPA performance is $5, your campaign will deliver more closely to a $10 actual CPA.

And on the same page, a warning worth reading twice:

Campaigns currently performing more efficiently than their set targets may see campaign performance trending toward your set target.

No ambiguity there. Overperformance is being removed.

Exactly who this affects

The change only applies where two conditions overlap. You need both at once.

Condition 1. The campaign is limited by budget. In the Google Ads interface such a campaign carries a “Limited by budget” note in the Status column. It means demand exists, there could have been more impressions, but the daily budget ran out.

Condition 2. The campaign runs a target-based strategy:

• Target CPA;
• Target ROAS;
• Target CPC — Demand Gen only.

Campaign types in the firing line: Search, Shopping, Performance Max, Demand Gen, Travel. Display and Hotel already behaved this way — nothing changes for them.

Not affected: App campaigns, Video reach and Video view campaigns.

Note Performance Max on that list. For most online stores PMax is the main campaign taking the lion’s share of the budget. And it almost always runs into its budget ceiling.

Part 2. What it costs in money

Here’s the unpleasant part, and for some reason almost nobody spells it out.

When a campaign is limited by budget, your spend is fixed. You spend $1,000 a month and you’ll keep spending $1,000. What changes isn’t the spend — it’s what you get for it.

Running the numbers on ROAS

Revenue from ads = spend × ROAS. That’s simple arithmetic, not an opinion.

That’s $2,000 a month less on identical spend. Over a year, $24,000.

Your spend didn’t rise by a cent. The same money simply buys more expensive traffic now.

Running the numbers on leads

For those paying per lead rather than per purchase:

Half your leads disappear. The advertising invoice stays the same.

That’s the heart of it: you’re not paying for the change, you’re paying for having typed a number off the cuff once and forgotten about it.

See which of your campaigns hit their budget ceiling and how far actual performance has drifted from target. Free, on your own account data.

Calculate your gap →

Honestly, the other side of it

Google frames this not as a downgrade but as predictability — and there’s some truth to that.

Budget and efficiency used to be tangled together: you raised the budget and ROAS dropped unpredictably, because the system started entering pricier auctions. That’s why owners spent years afraid to increase budgets — the familiar “I’m scared to raise it, my ROAS will fall”.

After the change those two levers are separate. The target is what you ask for. The budget is how much volume you buy at that price. Raise the budget and efficiency holds at the stated target.

That genuinely is more convenient. But it only works on one condition: the number in the target field has to mean something. If it’s been sitting there since last year as a rough guess, it now becomes your real efficiency.

Part 3. If this change hit you, the problem was already there

Here’s the uncomfortable but important part.

A campaign that overperforms its target for months while running into its budget ceiling — that isn’t luck. It’s a job half done. And it existed long before 17 August; Google was just quietly making up the difference, and now it has stopped.

Look at what that situation actually means. A campaign returning 600% ROAS against a 400% target pays back one and a half times better than you considered good enough. And at the same time it’s capped by budget — meaning there is demand you aren’t buying. People are searching for your products, the campaign could bring them in, but the money ran out.

That’s money sitting on the table. It has been sitting there for months.

There were exactly two right moves, and both were available long before 17 August:

1. Scale the budget up. The advertising pays back better than plan and you’re throttling it. So give it more money — until efficiency settles at a level you’re happy with.
2. Raise the target to match reality. If the budget genuinely can’t go up, then the target should reflect what’s actually happening rather than sit there as a just-in-case safety margin.

Neither happened — which means the account was running on autopilot. Set up once and left alone.

Whose shortfall this is depends on who runs the account. If you pay a freelancer or an agency, then spotting the gap between target and actual and coming to you with a proposal is precisely the work you’re paying for. If you run the account yourself, it’s simply a task you never got round to. There’s no shame in it — most accounts look like this.

What’s bad is something else: before 17 August nobody paid for that shortfall, and now you do. Every day.

Not sure whether your targets are being met, or where you’re underfunding campaigns that work? Analyse your account through our platform. Completely free.

Check your targets →

By the way, notice one detail: both correct responses lead back to the budget. Either you put in more money, or you honestly acknowledge what you’re earning. That’s not a coincidence — and in part 7 I’ll explain why, for an online store, the budget is a more reliable lever than the target in the first place.

Part 4. Why the usual reaction doesn’t work

Over the past week I’ve seen several typical responses. Here’s why each one is a bad idea.

“I’ll set the target higher, with some headroom — let Google work for it.”

Don’t confuse two different things here. Raising the target to the level of actual performance is right; that was part 3. Pushing the target above actual performance, with headroom, is harmful.

An excessively high target isn’t motivation for the algorithm, it’s a constraint: the system won’t find enough auctions that clear it, and the campaign will underspend. You’ll get a magnificent ROAS on a laughable volume. Stores don’t grow on that.

“I’ll drop the target completely and move to Maximize conversion value.”

The direction is right, and below I’ll explain why I consider it the primary option for a store. But not in this form. Dropping the target and changing nothing else means moving from “a badly set target” to “no target at all”: a strategy without a target will spend every last cent of the budget, weak days and weak products included. It only works together with the budget as your lever — see part 7.

“I’ll do nothing and give it a couple more days.”

The worst option, and the time has already been spent. The rollout isn’t instant, so the first few days may have passed quietly — but a week has gone by and you have the data. Silence in those early days never meant you’d dodged it.

“I’ll set a bid limit or exclude the data.”

Doesn’t treat the cause. This is a problem of wrong targets, not corrupted data. Excluding data additionally damages the measurement history the algorithm learns from.

Part 5. The three-column rule

Auditing your entire account comes down to one table with three columns. List your campaigns and fill in:

From there, three possible outcomes, each with its own action.

Column A. Actual beats target — act today

Actual ROAS above target, or actual cost per conversion below target.

This is exactly the money from the calculations above. Your job is to bring the target in line with reality: raise the target ROAS closer to actual, or lower the target cost per conversion closer to actual. And separately, answer the question from part 3: isn’t it time to simply give this campaign more money?

Brand campaigns suffer worst here — the ads on searches containing your store’s name. They almost always overperform any target, because the person was already looking for you specifically. Check those first.

Column B. Actual roughly equals target — do nothing

A gap within 10% is normal noise. Your target is already honest and the change won’t touch you.

Column C. Actual is worse than target — the change doesn’t apply

If a campaign isn’t reaching its stated target anyway, it has nothing to give back. You have a different problem here — the campaign isn’t hitting plan — but it has nothing to do with 17 August and gets fixed separately.

Three columns, three answers. Nothing else to remember.

You don’t have to fill those three columns in by hand: the platform sorts your campaigns for you and shows which ones sit in column A. Free.

Break down your account →

Part 6. How to do it by hand — step by step

Step 1. Find the campaigns limited by budget

Google Ads → Campaigns → the Status column. Look for the “Limited by budget” note.

One important detail: the status floats. A campaign can be budget-limited on peak days and free the rest of the time. So don’t only look at today’s snapshot — think back to whether that note appeared over recent months. If it did even once, the campaign is in the risk zone.

Step 2. Open Google’s own tool

Google built a dedicated screen for this — the Bid Target Adjustment Tool. It’s been live since 6 July.

Two ways in:

• the account notification “Review your campaign targets” → the “Review campaigns” button;
• Campaigns → the settings icon → the Bidding section → “Review campaigns”.

The tool finds your budget-limited campaigns itself and proposes a new target for each.

Two things you need to know about it.

First: Google won’t apply anything on its own. The help page states outright that neither targets nor budgets will be changed automatically. Until you press “Apply”, nothing has happened. Opening it, looking and closing it again leaves you exactly where you were — still losing money.

Second: for campaigns with fewer than 7 conversions there’ll be no recommendation at all. Google says so explicitly: at that volume performance is unpredictable and there’s nothing to calculate from. For a smaller store that means half your campaigns come with no suggestion — you’ll be deciding with your own head.

Step 3. Choose the new target

Four options.

Option 1 — target at actual. Set exactly what the campaign delivers now: target was 400%, actual is 600% → set 600%. You keep current efficiency and volume stays roughly the same. The safest choice.

Option 2 — target in the middle. Actual 600%, old target 400% → set 500%. You knowingly trade some efficiency for more volume. Makes sense if you genuinely want to grow and your margin allows it.

Option 3 — keep the target but give it more money. The case from part 3: if the advertising pays back better than plan, the right move isn’t to throttle it with a target but to scale the budget.

Option 4 — drop the target entirely and manage the budget. Google itself lists this as one of the available actions. For an online store I consider it the primary one — that conversation is part 7.

An honesty check on your target. Before typing a number in, answer one question:

• for cost per conversion — how much can I realistically pay for a single purchase and stay in profit?
• for ROAS — what is my gross margin? At a 25% margin, a 400% ROAS is break-even, not a target. Anything below it is trading at a loss.

A target that survives this check was set deliberately. A target that doesn’t is an old number — and the 17 August change simply handed you a reason to finally fix it.

Step 4. Change it properly

Three rules that will save you a month:

1. Not in one jump. Moving a target by more than 10–15% at once disrupts the algorithm’s learning period. If the gap is large, go in two or three steps with a 1–2 week pause between them.
2. Not everything on the same day. Don’t change target, budget and campaign structure simultaneously. You won’t be able to tell afterwards what actually worked.
3. Start where the money is. A 20% gap on a campaign spending $2,500 a month matters more than a twofold gap on one spending $150. Count dollars, not percentages.

Step 5. What to watch afterwards

• First 3 days: look, don’t touch. There’s no data yet.
• Days 3–14: judge by conversion cycles, not by the calendar. At 10 conversions a week, three days will show you nothing but noise.
• Weeks 2–4: this is where real “after” data appears. If you missed with the number, correct it now.
• Beyond that: set yourself a monthly reminder. A one-off fix doesn’t hold forever: performance moves, the gap between target and actual reopens, and in six months you’re back in column A.

Part 7. What I recommend to stores: drop the target and manage the budget

Everything above is about fixing the target properly, assuming you’ve decided to work with one. Now here’s what I actually recommend to most online stores.

Move to Maximize conversion value with no ROAS target. And manage the budget.

The problem with target strategies isn’t Google

A target strategy assumes you know the right number. That you calculated your ROAS from margin, accounted for returns, repeat purchases and differences between categories — and typed in a justified figure.

Honestly: for most stores that figure is a rounded guess, set once and never revisited. Until 17 August it got away with it, because Google quietly compensated for the imprecision and delivered better than requested.

From now on your guess is executed literally. The cost of being wrong in that number is now direct.

Why “conversion value” and not “conversions”

For an online store this matters enormously.

Maximize conversions counts orders as units. A $10 order and a $270 order are the same thing to that strategy. The algorithm will faithfully bring you plenty of cheap orders, because you asked for quantity.

Maximize conversion value counts money. It extracts the maximum revenue from the same budget. A store lives on revenue and margin, not on the number of rows in its CRM — which is why for a store the right strategy is almost always this one.

The budget as your only lever

From there, management reduces to one comprehensible thing: how much money you give the campaign.

Not “what ROAS do I want”, but “how much am I prepared to put into this segment”. The difference is enormous:

• a target is a promise to the system, which it will now keep literally, even if you got it wrong;
• a budget is a real safety catch. It caps losses hard and takes effect immediately.

The working routine is simple:

1. Split campaigns so the budget can be moved meaningfully — by category, by product group, by whatever you want to grow separately. One campaign for the whole store robs you of this lever.
2. Give each one a starting budget — the amount whose loss you could live with in the worst case.
3. Once every week or two, look at how each performed: revenue, margin, payback.
4. Performing well — add budget. Not performing — take it away. Don’t guess in advance, react to what happened.

Notice this is exactly where both correct responses from part 3 led. The only difference is that here the budget isn’t the fallback, it’s the primary instrument.

This is the same thing we do at GetProfit daily and automatically at the level of every product, category and brand: what sells today gets more money, what has only been spending for months gets the minimum or nothing.

Look at your own data to see where the budget goes today and what it brings back on each campaign. Free.

Open your account →

Honestly, the downsides of this approach

I don’t want to sell it as a solution without costs.

• You now observe efficiency rather than set it. There’s no guaranteed ROAS — there’s an actual one you review after the fact and respond to with budget. Some people find that uncomfortable.
• The strategy will spend the whole budget. Always, bad days included. That’s precisely why the budget must be an amount whose loss you can absorb — it is the safety catch.
• Switching strategy resets learning. The algorithm needs 1–2 weeks to relearn. Don’t do it a week before your seasonal peak.
• It requires discipline. Every week or two you’ll have to look at a report. If you won’t look at all, a target strategy with an honest number is safer for you.

When to keep the target after all

One case: when your number is calculated, not guessed. You know your gross margin, you know your allowable acquisition cost, and the target follows from them. Then a target strategy does precisely what you need — and the 17 August change is arguably a plus for you: budget and efficiency are now separate levers, and the budget can be raised without fear that efficiency will drift.

If the number was there on a “roughly this” basis — drop it and move to managing the budget.

Part 8. Special cases

You use a portfolio strategy (one strategy across several campaigns). All adjustments are made at portfolio or shared budget level only. You can’t fix an individual campaign inside a portfolio.

You run Performance Max. Same rules, but checking is mandatory: PMax almost always hits its budget ceiling and almost always carries the store’s core revenue. First candidate for review after your brand campaign.

You have few conversions (fewer than 7 in the campaign). No recommendation will come from Google. Work it out by hand: take actual performance over 30–60 days and set the target next to it. If there isn’t even enough data for that, the issue isn’t your targets — it’s that the campaign is sliced too thin.

You have one campaign for the whole store. Then moving to budget management will do little: there’s nowhere to move budget to, and inside a single campaign you aren’t steering anything anyway. Split campaigns by major categories first — otherwise the budget isn’t a lever, it’s a single tap.

Peak season is coming. There are about two months to the autumn-winter peak. Sorting out targets is better done now, on calm data, than in November when every day is expensive and any edit disrupts learning at the worst possible moment.

Part 9. The questions people ask most

“Will Google fix my targets for me?” No. The help page states it plainly: neither targets nor daily budgets will be changed automatically. The tool only proposes — you apply.

“I haven’t touched the account in six months. Does this affect me?” Very probably yes — and more than those who watch their accounts. The longer a target sits unrevised, the wider the gap between it and reality.

“My budget is small, a couple of thousand a month. Does this matter to me?” The change affects you just the same. But set priority in money: if the gap costs you $10 a month, it isn’t worth dropping everything for. Checking is still worth it — it’s twenty minutes.

“What if I just raise the budget?” Then the campaign stops being budget-limited and the change no longer applies to it. Technically, yes, that’s a solution — and often the right one (see part 3). But raising the budget only makes sense if the target is honest: otherwise you’re just spending more at the same weak bar.

“Isn’t dropping the target risky? It’s the only brake I’ve got.” The brake isn’t the target, it’s the budget. The target is your guess at the right number, and the system now executes it literally, mistake included. A budget can’t be wrong: whatever you set is the most you can lose. One condition — the budget must be an amount whose loss you can absorb, and the report has to be reviewed every week or two.

“How do I tell whether the change has already hit me?” By the divergence from your previous level at unchanged spend: same spend, but cost per conversion has crept up or ROAS has slipped toward your target. The change has been live for over a week, so you already have the data for that comparison. If spend changed too, the cause is something else.

What to do right now

One evening is enough:

1. Open Google Ads and list the campaigns marked “Limited by budget” that run a target ROAS or a target cost per conversion.
2. Fill in three columns: target / actual over 30 days / who sits in column A. Start with your brand campaign and Performance Max.
3. For every campaign in column A, answer two questions. First: is the number in the target field calculated from margin, or set by eye? Second: if the advertising pays back better than plan and hits its budget ceiling — why haven’t I given it more money by now?
   • Target calculated → bring it in line with reality in 10–15% steps, not one jump, and not on the same day as other edits. And settle the budget question separately.
   • Target set by eye → drop it, move to Maximize conversion value and manage the budget (part 7).

If this review shows your targets were guesses, that’s nothing to be ashamed of — most accounts are like that. What’s bad is something else: until 17 August Google quietly covered that imprecision for you, and now it has stopped.

And the main conclusion, the one this whole article was for: managing money is more reliable than managing promises. A target is your assumption, which the system now executes literally. A budget is a fact you control entirely.

---

Sources: Google Ads Help, “Changes to target based bid strategies” (17061251) and “Bid Target Adjustment Tool” (17125145). The money calculations are ours, based on the arithmetic of “revenue = spend × ROAS” under a fixed budget.