How much should you spend on ads?
Percent-of-revenue rules and competitor benchmarks are astrology. Budgets fall out of unit economics, learning velocity, and marginal returns.
"How much should we spend on ads?" is usually answered with folklore: five to ten percent of revenue, whatever the competitor supposedly spends, last year plus a stretch. All of these are answers to the wrong question. Spend is not an input you choose by convention — it is an output of three numbers your business already contains: what a customer is worth, what the ad platforms need in order to learn, and where your marginal returns cross your economics. Find those and the budget writes itself.
The floor: learning velocity
Modern paid media runs on algorithmic bidding, and algorithms learn from conversions. Below roughly thirty to fifty conversions per month per campaign, Smart Bidding and its Meta equivalents limp — decisions are made on thin data, performance whipsaws, and every judgment you make about "whether the channel works" is statistical noise. This defines a real budget floor: enough spend to generate learning-grade conversion volume on at least one focused campaign. If your realistic budget cannot clear the floor at your CPA, the answer is not to spread it thinner — it is to concentrate: one channel, one campaign, one conversion event, until volume earns the right to expand. A $3k budget across four channels is four failed experiments; the same budget on one campaign is a real one. To see where your own floor sits, the free ad budget calculator works the math backward from a revenue goal — learning floor included.
The ceiling: marginal returns meet unit economics
The maximum defensible budget is wherever the marginal acquisition cost — what the next tranche of spend actually pays per customer, not the account average — crosses your allowable CAC. The allowable number comes from your unit economics: contribution margin and payback tolerance for ecommerce, payback-ceiling math for SaaS. The marginal number comes from watching performance across spend steps as you scale. Averages hide the crossing point — an account can report a healthy blended CAC while its last $20k converts at double the allowable rate. If you only track one sophisticated thing about your budget, track marginal CAC by spend tier.
The two-question budget audit: (1) Is every campaign above learning-grade conversion volume? If not, consolidate. (2) Do you know the CAC of your last spend increment, as distinct from your average? If not, you do not know whether your budget is too big or too small — and neither does anyone quoting a percent-of-revenue rule.
Between floor and ceiling: allocation and cadence
- Fund demand capture before demand creation: search on high-intent terms and remarketing fill first, because they harvest existing intent at the best marginal economics. Prospecting and upper-funnel earn budget after capture saturates — visible as impression-share headroom running out.
- Move in steps, not leaps: budget changes of roughly 20% per learning cycle, one variable at a time, marginal performance reviewed at each step. The account tells you where the ceiling is if you climb slowly enough to hear it.
- Reserve a testing tranche — commonly 10-15% — with explicit permission to fail. Accounts where 100% of budget must perform never test anything, and accounts that never test decay on a delay.
- Rebase quarterly against economics, not calendar habit: margins, CPCs, and close rates all drift, and a budget frozen while its inputs move is wrong within two quarters — in whichever direction costs you more.
What the folklore gets you instead
Percent-of-revenue rules produce budgets that are simultaneously too small for winners and too large for strugglers, because they encode no information about returns. Competitor-matching imports a stranger's economics — different margins, different LTV, different funnel — into your P&L. And 'last year plus ten percent' assumes the marginal-return curve froze in your honor. Each heuristic survives because it ends the conversation quickly. The framework above takes a spreadsheet and an afternoon, and it replaces the conversation with two numbers you can defend in a board meeting: the floor your learning systems need, and the ceiling your economics permit. If you want an outside read on where your account sits between the two, that is a question our audit answers directly.
What percentage of revenue should I spend on advertising?
Percent-of-revenue rules are a starting heuristic at best. Growth-stage ecommerce brands often run 10 to 25 percent of revenue; mature brands 5 to 10; SaaS companies frame it as CAC payback instead. The better method is bottom-up: your unit economics set the ceiling (what a customer is worth) and your bidding systems set the floor (enough conversions to learn from).
What is the minimum budget for Google Ads to work?
Enough to generate roughly 30 conversions per month in your core campaign — that is where Smart Bidding starts functioning properly. If your CPA is $50, that implies about $1,500 a month minimum for one campaign. Below that, run manual or enhanced bidding on a tightly scoped keyword set rather than spreading thin across automation that cannot learn.
How do I know if my ad budget is too high?
Watch marginal, not average, performance: if your last budget increases bought conversions at a CPA meaningfully above your target while averages still looked fine, you are past the efficient frontier. Falling impression share losses to budget, rising CPCs on stable auction pressure, and MER drifting down while platform ROAS holds are the other tells.
Written by The ADSRUNNER team. If this resonated and you want to apply it to your own account, you can book a strategy call or run a free audit.