How to build a marketing budget you can defend to your CFO

Marketing budgets are usually set top-down: a percentage of revenue, or last year's number adjusted for mood. Then marketing spends the year trying to justify it. If you want to know how to build a marketing budget that holds up under scrutiny, flip the process: start from the customers the business needs, price them by channel and agree in advance when spending stops. The budget conversation gets a lot shorter.

This is the bottom-up method we use, step by step, plus the checklist to bring into the room with your CFO.

Why most marketing budgets don't survive the CFO

• They start from spend, not outcomes. A number is picked first and the plan is reverse-engineered to fit it.
• They rely on platform-reported results that don't reconcile with what finance sees in revenue and the bank account.
• They assume costs stay flat as spend grows, when acquisition usually gets more expensive at the margin.
• They have no stopping rule, so nobody knows what result would cause spend to pause.

A CFO doesn't need a perfect forecast. They need assumptions they can check, numbers that tie back to the books, and a clear point at which spend stops.

How to build a marketing budget from the revenue target

Work backward from the number the business needs, in this order:

• 1. Take the revenue target for the year from the business plan.
• 2. Subtract expected revenue from existing customers, based on your actual retention and repeat purchase history.
• 3. Divide the remaining new-customer revenue by first-year value per new customer. That's how many customers marketing must acquire.
• 4. Multiply by your blended new-customer CAC, adjusted for scale (more on that below). That's your acquisition budget.
• 5. Add non-working costs: tools, creative production, and agency or team costs.
• 6. Phase it by month to match seasonality, sales cycles and cash flow.

A simple illustration with made-up numbers: say the target is $5M, existing customers are expected to deliver $3M, and a new customer is worth $250 in first-year revenue. Marketing needs to acquire 8,000 new customers. At a blended CAC of $80, that's $640,000 in acquisition spend before scale adjustments and non-working costs. The figures are hypothetical; the logic is what you'll defend.

Price each customer by channel

• Use your real blended new-customer CAC from the last two quarters, not platform-reported CPA.
• Assume marginal CAC rises as spend scales. Build that into the model using your own history of spend versus CAC at different levels, and be explicit about the assumption.
• Count only new customers. Returning buyers converted by ads make CPA look better than the cost of real growth.
• Separate proven channels from experiments, and give experiments their own capped line.

Platform-reported CPA tends to flatter performance, for reasons we cover in ROAS vs MER. Your CFO will reconcile against revenue and cash, so the budget model should too.

Modeling rising costs at scale

The first dollars in any channel reach the most interested buyers. As spend grows, platforms have to reach people who are less ready to buy, so each additional customer tends to cost more than the average. A budget that assumes today's average CAC holds at double the spend will overpromise.

A hypothetical: a channel delivers customers at $80 on average at $40,000 a month. Pushing to $60,000 might bring the extra customers in at $110 each, which lifts the blended average even though the channel is still working. Model spend in tiers, each with its own assumed CAC, and show the CFO where the tiers stop making sense. Use your own history of spend against new customers to set the tiers; the numbers here are only there to show the shape.

A CFO doesn't need you to be certain. They need to see the assumptions and the point at which you'd stop spending.

Set payback guardrails

Agree up front on the maximum CAC payback period the business can fund. Subscription businesses with strong retention can often fund longer paybacks; many e-commerce brands with thinner margins aim to break even on the first order or within a few months. The right answer depends on cash position, margin and retention, which is exactly why finance should help set it.

To calculate payback, divide blended new-customer CAC by the contribution margin a customer generates per month. If acquiring a customer costs $120 and each customer produces $30 of contribution margin a month, payback takes four months. For businesses with lumpy purchases, track cumulative contribution margin by monthly cohort and note the month it crosses CAC.

Spend scales while new cohorts pay back inside the guardrail and pauses when they don't. That's a rule the CFO can audit. Typical guardrails include:

• A maximum blended new-customer CAC
• A maximum payback period, measured by cohort
• A minimum contribution margin after marketing
• A monthly spend ceiling tied to available cash

Split the budget: proven, scaling and experimental

• Proven: channels with a stable CAC history at current spend. This is the bulk of the budget.
• Scaling: channels that have hit targets at small scale and are being pushed further, with closer monitoring.
• Experimental: new channels, formats or markets, with a capped line and success criteria agreed before launch.

Keeping experiments in their own line protects them from being cut first, and protects the core budget from them. It also stops long-horizon investments like content and referral programs being judged on last-click results; our piece on growth loops vs funnels explains why those compound differently.

Review monthly, reallocate quarterly

A monthly scorecard keeps everyone honest: spend, new customers, blended CAC and payback by cohort. Reallocation between channels happens quarterly, once there's enough data to see a real trend rather than noise.

Agree in advance what triggers a change, so the monthly review is a check rather than a negotiation:

• Scale up a channel when it beats its CAC target for consecutive months at current spend and the next spend tier still fits the payback guardrail.
• Hold when results sit inside the target range or the data is too thin to call.
• Cut back when a channel breaches its guardrail for consecutive months and the cause isn't a known one-off, such as a tracking outage or a stock-out.
• Reforecast when a core assumption (retention, first-year value, pricing) moves materially.

This is where reliable marketing analytics earns its keep. If spend, customers and revenue come from different sources that don't reconcile, every review turns into an argument about the data instead of a decision about the budget.

What to bring to the CFO meeting

Use this checklist to prepare:

• The revenue target and the new-customer gap it implies
• First-year value per customer, with the data source named
• Blended new-customer CAC by channel for the last two quarters
• Your rising-cost assumption at higher spend levels
• Payback guardrails and the stopping rule
• The experiment budget, capped, with success criteria
• Monthly phasing and the review cadence

If you'd like a partner to build that model with you, our full-funnel growth marketing engagements start with exactly this exercise, and our pricing page shows how those engagements are structured.

Frequently asked questions

What percentage of revenue should go to marketing?

Percentage-of-revenue rules are a sanity check, not a plan. The right figure depends on margins, growth targets, retention and how efficiently you acquire customers. Build the budget from customers needed and CAC, then compare the result to revenue to check that it's plausible.

How often should a marketing budget be updated?

Review performance monthly and reallocate between channels quarterly. Reforecast the full-year budget whenever something material changes: a big shift in CAC, retention, pricing or the revenue target itself.

How do I budget for channels I haven't tried yet?

Give experiments a separate, capped line with a defined test period and success criteria agreed in advance. If a test hits its targets, it graduates into the scaling budget; if it doesn't, the loss is limited and the learning is documented.