On August 17, 2026, Google Ads will begin changing how campaigns marked “Limited by budget” bid when they use a target-based strategy. For campaigns using Target CPA (tCPA) or Target ROAS (tROAS), the system will optimize more consistently toward the target entered in the account, including when the budget changes.

The update will not automatically increase the budget or rewrite the target. It does, however, change the relationship between budget and efficiency. A campaign that has consistently exceeded its stated target may begin delivering closer to that target after August 17. In practical terms, a favorable gap that an advertiser considered “normal” should no longer be treated as guaranteed.

The essential point: if the target in Google Ads is more permissive than the company’s real business objective, the update may make that difference visible in conversion cost or value. Not every account needs a change, but budget-limited campaigns that clearly and consistently outperform their target should be reviewed before August 17.

What Google has confirmed and what should not be assumed

Google states that the new behavior applies to Target CPA and Target ROAS across Search, Shopping, Performance Max, Demand Gen and Travel. Target CPC is also included for Demand Gen. App, Video Reach and Video View campaigns will continue using the previous mechanism, while Display and Hotel already use the new behavior.

The Google Ads auction itself is not changing. The update concerns the way automated bidding optimizes a budget-constrained campaign. Google also makes three points clear: budgets and targets will not be adjusted automatically, target-based campaigns that are not limited by budget are not affected by this update, and daily and monthly account budget limits will continue to apply.

A tROAS target of 800% that consistently delivers 1,300% illustrates the risk. After the update, the campaign may accept opportunities with lower marginal value and move toward the stated 800% target. This does not mean that it must reach exactly 800% or that every account will lose the same amount of value. It means the gap between the target and current performance can no longer be ignored.

Who needs to review campaigns and who can leave them unchanged

The right filtering process starts with the campaign’s current status, not an old notification and not the assumption that every automated strategy is affected.

  • Campaign without a target: Maximize Conversions or Maximize Conversion Value without tCPA or tROAS is not part of this specific change.
  • Campaign with tCPA or tROAS but no budget constraint: Google says its behavior will remain unchanged.
  • Budget-limited campaign with no clear advantage over its target: there is no special reason to change the target solely because of the August 17 date.
  • Budget-limited campaign that clearly and repeatedly outperforms its target: this is where data quality, stability and the real business objective need to be reviewed.

“Outperforming the target” points in opposite directions for the two strategies. With tCPA, performance is better when actual CPA is lower than the target. With tROAS, performance is better when actual ROAS is higher than the target. An adjustment intended to preserve current efficiency would therefore usually mean lowering tCPA and raising tROAS. Generic advice to “increase the target” is wrong for tCPA.

Weekly tCPA and tROAS performance analysis highlighting an unusual result
A strong average is not enough. Weekly distribution, unusual conversions and reporting delay can change the decision.

Step zero: verify that the conversion signal is real

Before calculating anything, check the events Smart Bidding actually uses. In Google Ads, Primary actions appear in the Conversions column and are used for bidding when the associated goal is active. Secondary actions are normally for observation. A poor setup can turn a convincing report into a false picture of performance.

A minimum audit should answer these questions:

  • is each order or form submission recorded only once?
  • do calls, forms, purchases and offline imports use clear counting rules?
  • have actions with no direct commercial value, such as a page view or intermediate click, accidentally been left as Primary?
  • are order values real, or are identical placeholder values being sent?
  • are cancellations, returns and uncontactable leads reflected in the business analysis?
  • is the campaign optimizing the same definition of a conversion that the sales team uses?

If the answer is uncertain, do not adjust the target to fit an unverified number. Correct measurement first, document when the change was made and allow the data to mature. Google advises against applying data exclusions or new bid limits solely in response to this update. Those controls are intended for real tracking incidents, not for hiding an inconvenient period.

Choose an analysis window that can support the decision

The last seven days are rarely enough for a target change. Use the longest of these three windows: at least 14 days, two complete conversion cycles, or the period needed to collect a reasonable volume of conversions. For seasonal accounts, compare similar weeks instead of relying only on a blended total.

As a cautious operating benchmark, not an official Google requirement, a tCPA review becomes more useful around 30 conversions and a tROAS review around 60. Google does not produce a recommended target below seven conversions, but seven conversions may still be too little for a stable commercial decision. The useful volume depends on value variation, sales frequency and conversion-cycle length.

Important for tROAS: conversion count is necessary but not sufficient. If one order is worth as much as the other 59 combined, average ROAS primarily describes that order rather than the campaign’s usual behavior.

Two tests that expose tROAS inflated by outliers

1. Compare the aggregate average with the weekly distribution

Split the period into weeks and calculate ROAS for each one. Then compare the full-period average with the weekly median. If the 90-day average is 1,700% but the weekly median is 1,100%, one or two exceptional weeks may be driving the difference. The median does not replace ROAS, but it gives a clearer view of typical performance.

2. Recalculate without the strongest week

Temporarily remove the week with the highest value and repeat the actual-versus-target comparison. If the verdict changes from “adjust the target” to “no meaningful difference,” the signal is not yet robust. The analysis is sensitive to one fortunate result, and the bidding target should not be recalibrated around that spike.

It is also useful to inspect the share contributed by the highest-value conversions. If the top one or two orders represent a disproportionate share of total value, extend the period and review separately the products, markets or campaigns that produced that value.

Do not judge recent days before value has matured

Recent data is systematically incomplete in accounts with a long sales cycle. Google attributes conversions to the date of the ad interaction, while orders, qualified leads or offline imports may arrive later. During that period, CPA can look artificially high and ROAS artificially low.

Review the bid strategy report and conversion-lag information. Google recommends waiting one to two conversion cycles after a change before evaluating performance. If most conversions arrive after ten days, a decision based only on the last seven days compares mature spend with conversion value that has not yet been attributed.

Compare performance with the company’s real target

The campaign target is an instruction for the algorithm. The business target is the threshold at which a sale remains useful after margin, returns, operating costs and lead quality. These two numbers should not be confused.

For tCPA, start with the maximum acceptable cost for a customer and adjust it for the lead-to-customer close rate. If the company can pay 600 lei for a customer and only one in four leads closes, a 600 lei form CPA is not sustainable; the simplified starting point would be 150 lei before further quality adjustments.

For tROAS, the break-even threshold needs to be connected to margin, not revenue alone. With a 40% gross margin, a simplified calculation starts at 250% ROAS to cover media cost before returns and operating expenses. A healthy target may be considerably higher. This is why the Google tool cannot decide the correct target on its own: it does not know the company’s full economics.

A practical filter for the target gap

As a triage method rather than a universal threshold, Web Hat Solutions recommends against automatically changing a campaign for a gap below 20% versus target. A modest difference may come from normal variation, conversion mix or reporting delay. Above 20%, the review deserves more attention, but the final decision still depends on profitability and distribution stability.

  • Small or unstable gap: keep the target and monitor.
  • Large gap caused by an outlier: extend the period; do not adjust around one order.
  • Large, stable and profitable gap: choose between more budget, a target aligned with validated efficiency or a different bidding strategy.
  • The current target already reflects the commercial objective: keep it, while accepting that performance may move closer to it.

What to change for tCPA and tROAS

If you use tCPA

Example: the target is 100 lei and mature actual CPA is consistently 60 lei. If 60 lei is compatible with volume and the business objective, aligning the target with current efficiency means lowering tCPA, not increasing it. You may choose 60 lei or an intermediate value, such as 75 lei, if you want a controlled compromise between efficiency and volume.

If you use tROAS

Example: the target is 800% and stable, mature actual ROAS is 1,300%. If value distribution supports the result and the commercial objective requires that efficiency, alignment means raising tROAS. A more efficient target may reduce spend or volume, so it should be modeled and monitored rather than applied mechanically.

If you can raise the budget

Google presents a budget increase as an option for capturing more volume at the stated target. It makes sense only when the target is profitable, demand exists, tracking is accurate and operations can process the additional volume. A larger budget does not repair a poor commercial target.

If the budget is strictly capped

There is no single mandatory solution. You can align tCPA or tROAS with the validated objective, use Maximize Conversions or Maximize Conversion Value without a target if you accept efficiency fluctuations, or consolidate campaigns through portfolio strategies and shared budgets where the account structure allows. The right choice depends on the real priority: stable efficiency, maximum volume or strict spend control.

Why the Apply button should be checked rather than rejected automatically

The Bid Target Adjustment Tool is useful for quickly identifying campaigns and comparing the target with historical performance. Google says it does not calculate a recommended target for campaigns with fewer than seven conversions. The tool still cannot replace a review of value distribution, margin, Primary conversion actions or reporting delay.

In one anonymized real-account example, the interface displayed “Recent performance as of 6 July.” For one campaign, the recommendation proposed changing tROAS from 1,120% to 4,529%, although later mature performance was approximately 835%, below the existing target. One unusually large order in a single week had influenced the recommendation.

In the same account, another campaign appeared in the tool at roughly 830%, while refreshed data for the same period exceeded 1,100% after value matured. This example does not prove that the tool is wrong in every account. It shows why an average calculated from a snapshot needs to be compared with current data, weekly distribution and actual conversion value.

Check the campaign’s current Limited by budget status separately. Google explains that the general notification may be triggered by campaigns that were limited at some point in the previous 12 months, while recommendations use more recent data. A campaign appearing in the notice is not automatically constrained today.

A controlled plan before and after August 17

  1. Export a baseline: target, budget, actual CPA or ROAS, conversions, value, status, date range and conversion lag.
  2. Validate tracking: Primary and Secondary actions, duplicates, values, offline imports and counting rules.
  3. Check budget status now: do not treat a historical notification as the single source of truth.
  4. Analyze at least two conversion cycles: add the weekly distribution and the without-the-best-week test for tROAS.
  5. Compare with business economics: margin, close rate, returns and operational capacity.
  6. Change one important variable: target, budget or strategy, not all three at once.
  7. Record the date and reason: this helps separate your decision from Google’s system change.
  8. Wait one to two conversion cycles: Google also warns that planning forecasts may be less accurate between August 17 and August 31.

The analysis is best completed before August 17 rather than on the same day. If the target and the system’s behavior change at once, the cause of a performance shift becomes difficult to identify. For long-cycle accounts where a complete stabilization period is no longer possible, preserve a documented baseline and avoid a large adjustment based on incomplete data.

Frequently asked questions

Will Google automatically change my tCPA, tROAS or budget?

No. Google explicitly states that targets and budgets will not be changed automatically. What changes is the way the bidding system optimizes toward the target you entered.

Should every budget-limited campaign be changed?

No. The priority is target-based campaigns delivering significantly and consistently better efficiency than their target. Campaigns without a target, those that are not budget-limited and those without a meaningful gap do not require a change solely because of this update.

Is applying the tool recommendation enough?

Not without verification. The tool is a starting point. Compare the recommendation with mature data, weekly distribution, Primary actions, margin and the campaign’s current status.

Can the 20% threshold be used as an automatic rule?

No. It is an operating filter for prioritizing analysis. A 25% gap created by one order may be less meaningful than a 15% gap repeated consistently in a high-volume account.

What should I monitor after August 17?

Track actual versus target, spend, volume, value, channel allocation for Performance Max and Demand Gen, conversion lag and the commercial outcome in the CRM or order system. Evaluate after one to two conversion cycles, not after the first day.

Official technical references

The August 17 update makes the declared target more important than it was in a budget-constrained campaign. The sound decision is not to accept or reject Google’s recommendation automatically. It is to turn the target into an accurate expression of business economics, supported by clean conversion data and performance that remains stable even without the fortunate week.