AI + Paid Media
Google Says Performance Is Too Low—So Why Is It Recommending More Budget?

Updated: October 2026
Google Ads API v26 introduces a recommendation type that can suggest raising a campaign budget because performance is “too low.” That label is not a business case. Before approving more spend, advertisers should prove that the next dollar can produce acceptable incremental value—not merely that the platform sees room to spend.
What the new recommendation actually tells you
Google added RAISE_BUDGET_PERFORMANCE_TOO_LOW and a corresponding raise-budget recommendation object in its October 14 API release. The release note identifies a new machine-readable recommendation, but it does not define your margin, sales capacity, cash constraints or acceptable marginal return.
That distinction matters because average campaign performance can hide a weak frontier. A campaign may report an acceptable blended CPA while the additional impressions available at a higher budget are materially less efficient.
Build a budget-approval evidence packet
| Question | Evidence | Stop condition |
|---|---|---|
| Is demand constrained? | Lost impression share, reach, query or audience coverage | No credible inventory gap |
| Is measurement healthy? | Tag, import, consent and CRM checks | Recent tracking break |
| Is quality stable? | Qualified leads, revenue, margin, retention | Volume rose while quality fell |
| Can operations absorb it? | Inventory, call coverage, sales capacity | Backlog or stock constraint |
| Is the marginal return acceptable? | Small-step budget test | Incremental value below hurdle |
Test the recommendation in a controlled step
Do not jump from the current budget to the largest suggested amount. Choose a bounded increase, preserve conversion definitions, and pre-register the review window. Compare marginal spend with marginal qualified value, allowing for conversion lag.
- Record the recommendation, timestamp and campaign state.
- Confirm no major tracking or creative incident is active.
- Estimate the additional spend and required conversions.
- Increase budget in a reversible step.
- Watch marginal CPA or ROAS, not only blended results.
- Review downstream quality after the normal sales delay.
- Keep, roll back or extend according to the written rule.
Do not let an AI recommendation approve itself
Recommendation ingestion can be automated; approval should still follow the business’s authority model. Route small changes inside a proven range to an operator. Require finance or channel leadership when the change crosses cash, margin or capacity thresholds. Preserve the raw recommendation and the decision rationale.
Connect this workflow with MTC’s AI paid media rollback plan and limited-by-budget profitability check.
Separate average performance from the next dollar
Report the current blended result beside the result of the budget increment. If an extra $1,000 produces two additional qualified sales, evaluate those two sales against the extra $1,000. Blending them with years of cheaper conversions can make an unprofitable expansion look healthy.
Also annotate promotions, seasonality, competitor pressure and inventory changes. A successful budget test during a temporary surge should not become a permanent baseline without another review.
Run a seven-question approval review
For every recommendation, ask: What problem is the added budget meant to solve? Which segment is expected to absorb it? What is the required incremental outcome? How long is the observation window? Which signals would invalidate the test? Who can stop the spend? When will the result be written back into policy? If any answer is missing, the recommendation is not ready for approval.
Compare the suggested increase with at least two alternatives: leaving the budget unchanged and reallocating the same amount from a weaker campaign. The platform’s recommendation is campaign-local; the business decision is portfolio-wide. A campaign can deserve more budget in isolation and still be the wrong use of the next company dollar.
Document rejection as useful data
Store rejection reasons such as poor lead quality, insufficient inventory, unreliable tracking or a lower-return alternative. Review those reasons quarterly. Repeated rejections caused by the same constraint reveal a process problem that automation alone cannot solve.
Frequently asked questions
Does “performance too low” mean the campaign is failing?
Not necessarily. It is a platform recommendation label. Judge performance against the business target and marginal economics.
Should recommendations be auto-applied?
Only inside a narrow, pre-approved policy with measurement checks, spending caps and rollback criteria.
What is the most important metric?
Use incremental qualified value or contribution margin from the added spend, supported by delivery diagnostics.
