Performance Max Budget Sizer
Performance Max needs enough conversion signal to learn, then a steady budget to hold its bids. Enter your target CPA and how many conversions a day you want, and the sizer turns it into a launch budget, a steady state, and a ramp.
Size a Performance Max budget from outcomes, not guesses. Daily budget ≈ target CPA × desired daily conversions. While the campaign learns, Google’s guidance is to keep the daily budget comfortably above the target CPA — a useful floor is roughly three times the target CPA per day — so the algorithm gathers enough conversions to stabilize. Launch above the floor, then settle toward the steady-state number once learning ends.
Performance Max Budget Sizer inputs and result
| Phase | Plan |
|---|
How to use this calculator
- Enter your target CPAUse the cost per conversion you set as the campaign goal. The whole budget scales from this number, so make it realistic for your account.
- Set desired daily conversionsEnter the conversion volume per day you want at steady state. Daily budget is simply target CPA times this figure.
- Check the learning-phase floorThe sizer shows roughly three times your target CPA as a daily floor for the learning window — the budget Google needs to gather enough signal to stabilize bids.
- Plan the rampLaunch at or above the floor for the learning window, then settle toward the steady-state daily budget. The ramp table lays this out.
- Export the planCopy a share link, download the CSV, or print the PDF for the account plan or client approval.
RGM Expert Says
The mistake we see most often with Performance Max is launching it on a thin budget to ‘test cheaply.’ PMax is a conversion-hungry, machine-learned campaign type; starve it during the learning phase and it never gathers enough signal to bid well, so the cheap test produces expensive, unstable results. We size the launch budget above the learning floor on purpose, because the cost of a slow learn is higher than the cost of a few extra days of full spend.
Steady state is the easy part — target CPA times the conversions you want a day. The discipline is leaving it alone once it is set. Every large mid-flight budget change can re-trigger learning, which throws away the signal you already paid to accumulate. We cap budget moves at roughly twenty percent at a time and give the campaign a few days to settle before reading results.
We also keep clients honest about what the number assumes. The sizer presumes your target CPA is achievable in the auction; if the real cost to convert is well above it, no budget rule fixes that. When the floor budget cannot buy the conversions the math promises, the problem is the CPA target or the offer, not the spend — and that is the more useful conversation.
How it works
Daily budget is the product of target CPA and desired conversions; the learning floor is a multiple of the target CPA.
- Target CPA — the cost per conversion you set as the goal.
- Desired daily conversions — the volume the campaign should deliver per day.
- Learning window — how long bids take to stabilize, typically one to two weeks.
The ~3×-CPA learning floor is a widely used rule of thumb based on Google’s guidance to budget well above target CPA during learning, not a fixed published formula. Math is rounded for display; the model uses full-precision values.
Feed the learning phase, then hold the budget steady
Performance Max bids with machine learning, and machine learning needs examples. During the learning phase the campaign is gathering conversions to calibrate its bids, and a budget set right at the target CPA rarely produces enough of them. Google’s own guidance is to keep the budget comfortably above the target CPA while it learns — the three-times rule of thumb is a practical way to do that — so the algorithm reaches stability faster and wastes less of your spend on guesses.
Once it has learned, the danger flips from too little to too jumpy. Large budget swings can reset the learning phase, discarding the signal you paid to build. The steady-state number exists so you can fund the conversions you want without yanking the budget around; change it gradually, in increments, and let the campaign settle between moves.
Above all, the budget rule assumes your target CPA is reachable. If the real auction cost to convert sits well above your target, no daily-budget formula will conjure the volume the math implies. When the floor budget cannot deliver, revisit the CPA target, the bidding strategy, or the offer rather than simply spending more.
Performance Max budget rules of thumb
These are practical guidelines drawn from Google’s guidance and common practice, not exact published formulas. Validate against your own account.
| Situation | Practical guideline |
|---|---|
| Learning phase | Budget at or above ~3× target CPA per day |
| Steady state | Target CPA × desired daily conversions |
| Mid-flight changes | Keep budget moves under ~20% at a time |
| Conversions per signal | Aim to accumulate enough conversions to stabilize bids |
What operators say about automated campaigns
Automated bidding rewards patience and signal; the fastest way to ruin it is starving the learning phase, then panicking and changing everything.
Set a target you actually believe, then let the system gather the data to hit it before you judge the result.