Testing and scaling are fundamentally different spending activities, even though both show up as the same line item in a monthly ad budget. Testing spends deliberately to learn, expecting a portion of it not to convert efficiently. Scaling spends to capture proven performance. Budgeting for both with the same assumptions produces bad decisions in both directions.
Why the two need separate budget logic
Testing budget should be evaluated on learning value — did it identify a winning angle, audience, or creative — not on immediate ROAS, since most individual tests are expected to underperform relative to already-proven campaigns. Treating test spend with the same performance bar as scaling spend leads teams to under-invest in testing, which starves the pipeline of new winners needed to sustain growth.
Scaling budget, conversely, should be evaluated on efficiency and marginal return as spend increases — the question isn't 'did this work' (it already has) but 'how much further can this scale before efficiency degrades,' which is a different analytical question requiring different data (diminishing returns curves, audience saturation signals).
Setting a testing budget as a percentage of total spend
A common practical approach allocates a fixed percentage of total ad budget to testing on an ongoing basis — often somewhere in a 10-30% range depending on how mature and stable the scaling campaigns already are — rather than testing opportunistically with leftover budget. Treating testing as a required, protected allocation rather than a discretionary afterthought keeps the pipeline of new winning campaigns flowing even during periods when scaling campaigns are performing well and tempting the team to redirect everything toward them.
The right percentage shifts over a business's lifecycle: newer accounts or newer verticals typically need a higher testing allocation since less is proven yet, while mature accounts with several stable scaling campaigns can run a smaller, more targeted testing allocation.
Cash-flow implications of each
Testing spend tends to be more volatile and less predictable in its return timing, which matters for cash-flow planning — it should be budgeted as a cost with uncertain payback, not forecast into expected revenue the way scaling spend reasonably can be. Scaling spend, because it's tied to proven performance, can be forecast with tighter confidence and tied more directly to expected revenue and cash inflow timing.
Mixing the two in one undifferentiated forecast makes both less accurate — testing spend's uncertainty pollutes the confidence of the scaling forecast, and scaling's predictability can mask how much of the budget is genuinely still speculative.
Structuring accounts and campaigns to keep the split clean
Separating testing and scaling campaigns clearly — by naming convention, by campaign structure, or even by separate ad accounts for higher-volume testing operations — makes it much easier to track spend, performance, and budget allocation against each category independently, rather than needing to reconstruct the split from a mixed campaign list after the fact.
This separation also makes it easier to protect scaling campaigns from being disrupted by testing-related account issues (a flagged test creative, for instance) when testing runs in a distinct account rather than alongside proven, revenue-critical scaling campaigns.
How account capacity affects this strategy
Running testing and scaling in separate ad accounts requires enough account capacity to make that separation practical, which is one of the operational advantages of an unlimited-account agency model. Power Ads' account structure gives clients the room to isolate testing from scaling without needing to ration account access between the two.
Key takeaways
- Testing budget should be evaluated on learning value, not immediate ROAS, since most tests underperform relative to proven campaigns.
- Scaling budget should be evaluated on marginal efficiency as spend increases, a different question than 'did it work.'
- A protected testing budget percentage keeps the pipeline of new winners flowing even during strong scaling periods.
- Forecast testing and scaling spend separately, since their return predictability and timing differ significantly.
- Separating testing and scaling into distinct campaigns or accounts protects proven revenue from test-related disruptions.
FAQ
What percentage of budget should go to testing?
It varies by account maturity, but a common range is 10-30% of total spend, with newer accounts or verticals needing a higher share.
Should testing spend be forecast into expected revenue?
Generally no — treat it as a cost with uncertain payback rather than forecasting it with the same revenue confidence as proven scaling spend.
Is it worth running testing in a separate ad account?
For higher-volume testing operations, yes — it isolates any account-level risk from test creative away from stable, revenue-critical scaling campaigns.
