What changed for budget-limited Target ROAS and Target CPA campaigns in August 2026?
From 17 August 2026, Google steers budget-limited campaigns on Target CPA or Target ROAS to deliver closer to the target you set. Its own example is a $10 Target CPA running at $5, which now moves toward $10 (Google Ads Help). The change hit Search, Shopping, Performance Max, Demand Gen (including Target CPC) and Travel. Display and Hotel already worked this way, and App, Video reach and Video view campaigns keep the old behaviour. The rollout was completed on 27 August (Google's FAQ).
Google won't adjust your target or your budget for you, unless someone has switched on auto-apply for target recommendations (see below). In a portfolio or shared budget, the fix has to be made at that level.
A campaign stuck on Limited by budget while beating target is often the cheapest growth in the account. To find the ones this change hits, look for budget-limited campaigns whose ROAS or CPA has moved toward target since late August.
Worked example. A shop selling in Germany at a 40% gross margin runs PMax-DE, so break-even ROAS is 250%. It runs on Target ROAS 400%, Limited by budget at €1,500 a day. Until mid-August it ran at 512% ROAS, or €7,680 of revenue and €1,572 of contribution after ad spend per day. Since the change it has been sliding toward 400%.
| What you decide | Daily spend | ROAS it lands at | Revenue a day | Contribution after ads |
|---|---|---|---|---|
| Nothing, so it drifts to target | €1,500 | 400% | €6,000 | €900 |
| Raise the target to 500% | €1,500 | about 500% | €7,500 | €1,500 |
| Keep 400%, raise the budget to €2,000 | up to €2,000 | about 400% | up to €8,000 | up to €1,200 |
Example figures for a hypothetical account. The 40% margin and the ROAS each option lands at are our assumptions; Google says only that budget-limited campaigns now optimise to the target "regardless of your budget limit" (Google Ads Help). The 500% row assumes the campaign still spends its full budget. Contribution is revenue × 40% minus ad spend.
Drifting costs €600 of contribution a day against the target raise, about €18,000 over 30 days. Raising the budget at the old target buys more sales at 400%, and at a 40% margin that still beats doing nothing by €300 a day. Google now steers the extra spend to the 400% you typed, though, so the budget raise trails the target raise by €300 a day.
Raise the target first, toward 500% this week. Google's FAQ pairs a budget buffer with the target: "providing a budget buffer while setting a target that reflects your desired efficiency" (Google's FAQ). How much headroom to add, and when, depends on what a new customer is worth after the first order. Google only knows that through the customer acquisition goal or the values you send. These figures are Google-attributed revenue; if a holdout shows part of it would have come anyway, break-even moves and so may the answer.
Running a campaign like PMax-DE right now? We can look at it with you on a 30-minute call.
Book a DemoWhat can you still control in Smart Bidding and Performance Max?
Google owns the bid and, in PMax, most of the channel mix. It has to follow most of the settings you feed those decisions; audience signals and search themes are hints it can go beyond. This guide covers bidding, budgets and exclusions. Product feeds and listing groups need a page of their own.
| Lever | Binding or hint | What it changes |
|---|---|---|
| Bid strategy and target | Binding, as an average | The return or cost per conversion Google bids toward |
| Conversion goals, values and the customer acquisition goal | Binding | What counts as a conversion and what each one is worth |
| Budget | Binding monthly (30.4×); a day can reach 2× | Average daily spend, or a total for a fixed flight |
| Seasonality adjustments and data exclusions | Binding input for the dates you set | The conversion rate Smart Bidding expects, or the days it learns from |
| Channel priority and budget allocation | Partial | Search partners and the Display Network set to High or Low, with an optional cap as a % of the daily budget |
| Brand, keyword, placement, URL, customer-list and demographic exclusions | Hard limit | Where PMax may not serve |
| Audience signals and search themes | Hint | Where Google starts looking, with room to go beyond |
PMax has no bid limits. Google offers them only on Search and Shopping portfolio strategies (Target ROAS), and PMax can't use a portfolio strategy or a shared budget (PMax compatibility). The October tiers group Search, YouTube, Maps, Gmail and Discover together as "Google channels", so Search still can't be split from YouTube. Low priority caps are best effort, and manager accounts can't set them. The tiers reach accounts progressively and don't apply to campaigns on a total budget, Demand Gen aside (channel prioritization).
Should you use Target ROAS, Target CPA or Maximize conversion value?
Use a value strategy when your conversions carry revenue you trust, and a target when holding efficiency matters more than spending the whole budget. Since June 2026 Google has been relabelling the strategies: "Maximize conversion value with a Target ROAS" becomes Target ROAS and the CPA version becomes Target CPA, with no change in behaviour (Target ROAS, Target CPA). Editor, the API and the mobile app may show the old names for a while (Google Ads Help). Google wants 15 conversions in 30 days before Target ROAS on Search and Shopping. Below that, run Maximize conversion value without a target if your values are reliable, or Maximize conversions if they aren't, until volume allows a target.
A target change applies from the next auction, and Google says updates to ROI targets "never trigger 'Learning' status" (Google Ads Help). Learning follows a new strategy, a changed strategy setting, or campaigns and keywords moving in or out (bid strategy statuses).
Worked example. A lead-gen Search campaign spends €1,200 a day for 30 leads at a €40 Target CPA, and most leads convert within 9 days of the click. On 1 October the team cuts the target to €34. Google's rule is to let it converge for 1 to 2 conversion cycles, then evaluate over 2 cycles or 2 weeks, whichever is longer, and to account for conversion delay. With a 9-day cycle that means convergence by 10 to 19 October, then an 18-day window, then time for that window's last clicks to convert. The first clean read is in early to mid-November. A second cut on 8 October, inside the first cycle, would blur it. Suppose the campaign settles at about 24 leads a day. Spend falls to €816 a day, so €384 a day, up to €11,904 over October, goes unspent. That is only a saving if no other campaign could use it above break-even.
If a target moved and nobody on the team moved it, check the auto-apply History tab. "Adjust your CPA targets" and "Adjust your ROAS targets" can be auto-applied, account-wide (auto-apply).
How do conversion goals and values steer Smart Bidding?
Smart Bidding optimises toward the actions in the Conversions column, at the values you send. Deciding which actions count as primary is the strongest lever left to you, and it is easy to set once and forget. Changing them significantly sends the campaign back into Learning (learning period), so settle them well before a sale.
Account-default goals apply to every campaign that uses them; campaign-specific goals let one campaign bid toward its own selection. Secondary actions stay out of the column used for bidding (conversion goals). A newsletter sign-up left as primary next to purchases, in a goal the campaign bids on, will pull a Maximize conversions campaign toward sign-ups.
Conversion value rules adjust value by audience, location or device at auction time, for Target ROAS and Maximize conversion value (value rules). Google warns against repeating signals Smart Bidding already uses, such as geography, device and first-party lists. Use rules for what Smart Bidding can't see, a region with high return rates for instance, and put product margin into the conversion value itself.
The customer acquisition goal, set to New Customer Value, bids higher for new customers, defined from your customer list and website tag, on Search, PMax, Shopping and Demand Gen with a value strategy (customer acquisition). If Google Ads and GA4 disagree on the count, work through why Google Ads and GA4 report different conversions before touching a target.
How do you prepare Smart Bidding for a sale or a tracking outage?
For a short sale, schedule a seasonality adjustment with the conversion-rate lift you expect. When tracking breaks, apply a data exclusion for the affected click dates, starting one conversion delay before the break, as soon as you notice.
Google calls seasonality adjustments "ideal for short events of 1–7 days". It reserves them for major conversion-rate changes, since Smart Bidding already manages seasonal events. They work on Search, Shopping, Demand Gen and Display with a target strategy, on PMax with any strategy, and need no reverse adjustment afterwards (seasonality adjustments). Data exclusions cover Search, Display, Shopping and PMax and leave reporting untouched. Use them rarely and never remove one once applied (data exclusions); Google's holiday guide adds that the tool suits issues caught "within a week" (holiday guide).
Worked example. Search-DE and PMax-DE run Target ROAS 400% into Black Friday weekend, 27 to 29 November 2026. The usual conversion rate is 2.5% and the team expects 3.5%, so they set +40% for those three days. Suppose the sale only reaches 2.75%. Smart Bidding now values each click at 3.5 ÷ 2.75, about 1.27 times its worth, and as a first-order estimate ROAS lands near 400% ÷ 1.27, or 315%. Each €1,000 spent brings back about €3,150 in revenue where the plan said €4,000. Size the adjustment from last year's sale.
Example figures. The 2.75% outcome, a 10% lift where the plan assumed 40%, is our assumption.
For €3,000 of expected daily spend, Google's Shopping and PMax holiday guide says to set a €9,000 budget, 3 times as much, or 2 times without last year's data (holiday guide). A €3,000 budget tops out at €6,000 on any one day under the 2× overdelivery rule. If finance won't sign off on that, decide before the sale who may raise the budget over the weekend, and by how much.
What can you exclude or steer inside a Performance Max campaign?
Asset groups decide which creative goes with which audience signal and search theme. The hard limits are the exclusions in the table below.
| Control | Type | Scope in PMax |
|---|---|---|
| Brand exclusions | Hard | Search, Shopping and YouTube search inventory; Shopping ads can optionally still show on excluded brands |
| Campaign negative keywords | Hard | Search and Shopping inventory only |
| Account negative keywords | Hard | Up to 1,000, applied across the account |
| Placement exclusions | Hard | Account level, or manager-account lists |
| Final URL expansion and URL exclusions | Hard | Expansion is on by default; switch it off per campaign |
| Your data exclusions | Hard | Customer Match, web visitor and GA lists, on Search, Display, Shopping, Video and Discover inventory |
| Age and gender exclusions, device targeting | Hard | Demographic exclusions stop at age and gender |
| Audience signals | Suggestion | "Not a hard targeting criteria", in Google's words |
| Search themes | Suggestion | Up to 50 per asset group since 2025 (2025 update) |
Rows without their own link: PMax best practices and the compatibility page, Google Ads Help, October 2026.
Search campaigns with an exact match keyword are prioritised over PMax when the query matches it (About Performance Max). Keeping the few queries that carry most of the revenue in exact match Search makes their bids and terms easy to read.
What does Performance Max reporting show you?
Every PMax campaign has had asset-level metrics since 2025, and the three reports below are the ones to act on.
- The search terms report shows each term with its landing page and ad format, from March 2023 onwards. Act on it with negatives (Search and Shopping) or excluded content keywords (Display and Video).
- The channel performance report splits results across Search, Display, YouTube, Discover, Maps, Gmail and Search partners from 6 June 2025. Its help page still says "you can't directly control budget allocation per channel". That line doesn't reflect the October tiers.
- Auction insights for PMax cover Search and Shopping ads, segmented by ad type at account and campaign level, and show nothing below 10% impression share.
If the channel report shows Display taking a large share of spend with conversions you can't find in the CRM or confirm with a holdout, set Display to Low and cap it, once the control has reached your account. On a target strategy, the priority only changes spend while the campaign is budget-limited; the cap applies either way. Whether PMax conversions would have happened anyway needs a test, covered in incrementality testing vs MMM.
Which decisions does Google's automation leave to you?
Smart Bidding sets the bid for each auction inside one campaign, toward the goal you gave it.
| Decision | What Google does | What stays with you |
|---|---|---|
| Budget across campaigns and platforms | PMax spreads spend inside one campaign. Recommended Investment Strategy suggests moves between campaigns in one Google Ads account, applied after you confirm | How much PMax gets against Search, Meta or TikTok |
| Whether the target still fits | Bids to the target you set, more strictly since August when the campaign is budget-limited | A price change, a margin squeeze or a promotion reaches Google only through the target or the values |
| When to fund a winner | Shows a recommended budget; auto-apply "won't increase your budget" | The call itself |
| Which conversion count is true | Bids on the Conversions column | Reconciling it with GA4, the CRM and the shop backend |
| A tag that breaks overnight | Keeps bidding on whatever data arrives | Noticing, then applying a data exclusion |
| Approval | Change history records what changed, when, and who changed it | Who may move a target, and why each change was made |
Change history has no field for why. The reason usually lives in someone's head or a Slack thread, and three months later nobody can tell a tested call from a hunch. Since August the gap costs more. Budget-limited winners need a weekly check, and each decision on them is worth logging with its reason.
Budget-limited winners flagged against your break-even ROAS.
Nupact checks every campaign on Google Ads, Meta, TikTok and Amazon Ads around the clock against its own per-market baseline. A campaign limited by budget while beating your CPA target or break-even ROAS for five days or more gets flagged, and Nupact drafts the budget move with a forecast. For ROAS campaigns the flag reads your break-even, so it still catches a campaign Google is steering down to a 400% target when break-even sits at 250%. Whether to raise the target first, as this guide suggests, or approve the drafted budget move stays with your team, and whatever you approve is logged with its reason.
Moves between platforms arrive the same way. In our product demo, Google Demand Gen-DE sat at 190% ROAS against a 240% break-even for seven days, so Nupact drafted a move of €1,360 a day to Meta Advantage+ Sales-DE, with its revenue forecast attached for whoever approves it. Nothing applies until someone approves it in Slack or the app, unless you have let that action type run on its own inside guardrails you set. The log records each change with its reason, rejected proposals included, and every row has a Revert control, so a change can be rolled back from the log. When a conversion feed goes stale, Nupact pauses its own autonomous actions and pages the team. See anomaly detection and what an AI agent should be allowed to do.
What else do people ask about Smart Bidding and Performance Max?
Does changing Target ROAS restart Smart Bidding's learning period?
No. Google says updates to budgets and ROI targets never trigger the Learning status, and Target ROAS uses the new target from the next auction. Once it has converged, which Google puts at one to two conversion cycles, judge the result over two conversion cycles or two weeks, whichever is longer, and avoid a second change inside one cycle.
Can you set bid limits on Performance Max?
No. Google offers bid limits on Search and Shopping portfolio strategies, and Performance Max can't use a portfolio strategy or a shared budget. You steer what PMax pays through the target and the budget.
How do you stop Performance Max showing on your brand searches?
Add a brand list as a brand exclusion on the campaign. In PMax it covers Search, Shopping and YouTube search inventory, and a checkbox lets Shopping ads still show on those searches if you want them. Google calls brand exclusions the better option than negative keywords for brand traffic, since they also block clear misspellings and foreign scripts.
Can you see search terms in Performance Max?
Yes. Since 2025 every PMax campaign has a search terms report, with the landing page and ad format for each term. Block the terms you don't want with negative keywords at campaign or account level. In PMax, campaign negatives apply to Search and Shopping inventory only.
Will Google raise my budget when a campaign is limited by budget?
No. Auto-applied recommendations never increase a budget, and Google says the August 2026 bidding change leaves targets and budgets alone too. Hover over the Limited by budget status to see the recommended budget.
Should you use a seasonality adjustment for a whole Black Friday week?
Only if the whole week brings a step change in conversion rate. Google calls 1 to 7 days ideal, warns that adjustments may not work as well beyond 14 days, and says to use them only for major changes because Smart Bidding already manages seasonal events. If last year's jump sat on Friday and Cyber Monday, adjust those days and leave the rest to Smart Bidding.