How much should a company spend on marketing? Enough to acquire the customers your revenue target requires, at a cost per customer your margins can carry, and you find that number by working backward through your own funnel rather than picking a percentage of revenue. Below is the bottom-up model I use, with a template where every line is an assumption a board can question and you can answer.
Why percentage-of-revenue rules mislead
The usual answer is a percentage of revenue, borrowed from companies that look nothing like yours. The problem is the logic.
- It runs backward. The budget follows last year’s revenue, while marketing is supposed to create next year’s. A company trying to double and one trying to hold flat get the same answer.
- It ignores unit economics. Two companies with identical revenue can have very different margins, deal sizes and payback tolerance. The one that earns back acquisition cost in six months can afford far more.
- It says nothing about allocation. A total doesn’t tell you what to put into search, events or a second hire.
- It can’t be defended. When a board member asks “what do we get for this?”, “it’s what similar companies spend” is not an answer.
A percentage can be a sanity check at the end. It shouldn’t be the input.
Start from the revenue target and work backward
Begin with the number the board already cares about: next year’s revenue target. Then strip out everything the marketing budget doesn’t need to buy.
- Revenue from existing customers. Renewals, expansion and repeat purchases, forecast from your retention data. Lifecycle marketing supports it, but acquisition spend doesn’t pay for it.
- New revenue from other sources. Outbound sales, partners, referrals, the founder’s network. Be honest about what these deliver without marketing spend.
- The gap. What’s left is marketing-sourced new revenue. That’s what your budget has to buy.
Agree on this split with sales and finance before anyone opens a spreadsheet. Most budget fights I’ve sat through were really about who owns which slice of the target. It’s the first thing I pin down in a growth strategy engagement.
Funnel math: from revenue to pipeline to spend
Walk the gap down your funnel using your own conversion rates from the last two to four quarters, not industry averages.
A worked B2B example
Made-up round numbers, to show the mechanics.
| Step | Input | Result |
|---|---|---|
| Marketing-sourced new revenue | $1,200,000 | |
| Average first-year deal value | $20,000 | 60 new customers |
| Win rate (opportunity to closed) | 25% | 240 opportunities |
| Qualified lead to opportunity rate | 20% | 1,200 qualified leads |
| Blended cost per qualified lead | $150 | $180,000 media and program spend |
That $180,000 is working spend. It isn’t the whole budget.
Cap it with a target CAC
Now check the result against what a customer is worth. Suppose your finance model says marketing can spend up to $6,000 to acquire a customer at this deal size, given gross margin and the payback period you’ll accept. Sixty customers at $6,000 gives a $360,000 ceiling for everything: media, people, tools and agencies.
If the fully loaded plan lands above that ceiling, it doesn’t work at these conversion rates. The fix is in the funnel, the deal size or the target, not a quietly trimmed ad line. If you haven’t set a target CAC, invert the CAC payback formula: monthly gross profit per new customer times the payback months you’ll accept gives the most you can pay per customer across sales and marketing.
Adjust for timing and saturation
Two corrections most plans skip:
- Timing. If deals take a quarter to close, Q4 leads become next year’s revenue, so phase spend by quarter and start it one sales cycle ahead. If the target is recognized revenue rather than bookings, a customer signed in October contributes only a few months of first-year value, so you need more customers, or earlier ones.
- Diminishing returns. Cost per lead usually rises as you push a channel harder. If the plan needs twice last year’s leads from the same channels, price the extra volume higher or fund new channels to supply it.
The same math for ecommerce
For a DTC brand the chain is shorter. Forecast returning-customer revenue from your existing base first, divide what’s left by a new customer’s first-year revenue, then multiply by your target new-customer CAC. For example: $1.6 million needed from new customers at $200 of first-year revenue each is 8,000 customers, and at a $50 target CAC that’s $400,000 of acquisition spend.
Stress-test the assumptions
This is what makes the number defensible. Show what happens when each key input is worse than planned. In the example, a 20% win rate instead of 25% lifts working spend from $180,000 to $225,000; $180 per lead instead of $150 lifts it to $216,000. The conversation moves from “is this too much?” to “how confident are we in this rate?”
Working vs non-working budget
Put every line into one of two buckets:
- Working spend reaches the audience directly: ad media, sponsorships, paid placements, event booths, direct mail postage.
- Non-working spend makes working spend possible: salaries, contractors, agency fees, creative and content production, software, tracking, research.
Both are necessary, and there’s no universal right ratio. An SEO and content program is almost entirely non-working, because the cost is people and production. A paid-social-heavy DTC brand sits at the other end.
What matters is tracking the split. Non-working costs creep: another tool, another retainer, another contractor. When I audit budgets, the leak is more often in this bucket than in the ad accounts.
Test budgets for new channels
The funnel model funds channels you already understand. New channels need their own line, or they never get tested.
Size each test bottom-up: decide how many conversions you need to judge the channel, then multiply by an honest guess at cost per conversion. For example, needing 20 qualified leads at an expected $300 each means a $6,000 test, plus creative and setup. I usually cap the reserve at a tenth to a fifth of working spend, so experimentation neither starves nor quietly takes over.
Rules for the reserve:
- Every test has a written hypothesis, budget, end date and pass threshold before launch.
- Fund fewer tests fully rather than many partially. An underfunded test produces an inconclusive answer.
- A test that passes graduates into the core budget with its own line, and the reserve refills for the next one.
Budgeting for people, tools and agencies
These costs are often scattered across finance categories, which makes marketing look cheaper than it is. Put them in one planning view.
- People: use fully loaded cost (salary, employer taxes, benefits, equipment) and budget from the month someone is actually in seat. A March hire who starts in June frees up three months of budget that needs a decision, not silent carryover.
- Tools: list every subscription with its renewal date and pricing model. Many marketing tools price by contacts, profiles or seats, so costs rise as your list or team grows.
- Agencies and freelancers: separate fixed retainers from fees tied to ad spend. A percentage-of-spend fee grows every time you raise media, which raises your real marginal CAC.
- Fractional leadership: if you’re buying senior strategy part-time, see what a fractional engagement includes to budget it realistically.
Keep a small contingency line for what you can’t forecast, such as a platform change that forces a tracking rebuild.
Rules for increasing or cutting spend
Agree on the rules with finance before the year starts, so results trigger decisions instead of a budget renegotiation in every meeting.
| Signal, sustained over the agreed review window | Action |
|---|---|
| Channel CAC below target, lead quality holding | Raise that channel in steps, then hold and re-measure |
| Channel CAC above ceiling for two reviews in a row | Cut to maintenance level and diagnose before going to zero |
| Leads on plan, opportunities or deals behind | Fix qualification, follow-up or conversion before touching spend |
| Revenue ahead of plan, efficient channels not saturated | Pull spend forward from later quarters |
| Company-wide cash pressure | Cut the least efficient marginal spend first; protect the test reserve and compounding channels where possible |
Three principles sit behind the table. Judge channels on marginal CAC, what the last dollar bought, not the average. Reallocate between channels before asking for new money. And review monthly but reforecast quarterly, so short-term noise doesn’t trigger big moves. Without reliable tracking and attribution, every rule becomes a debate about whose numbers are right.
Budget template
Build the spreadsheet in three tabs. Every output is a formula, so changing one assumption updates the whole plan.
Tab 1: Inputs. The only place anyone types numbers. Example values continue the B2B case above.
| Input | Example value |
|---|---|
| Revenue target for the year | $5,000,000 |
| Forecast revenue from existing customers | $3,200,000 |
| New revenue from sales, partners and referrals | $600,000 |
| Average first-year deal value | $20,000 |
| Win rate / lead-to-opportunity rate | 25% / 20% |
| Cost per qualified lead, by channel | $150 blended |
| Target marketing CAC | $6,000 |
| Sales cycle | 3 months |
Tab 2: Funnel and spend. Formulas only: customers, opportunities, leads and working spend by channel, phased by quarter with the sales-cycle offset, plus the sensitivity rows.
Tab 3: Full budget. One row per line, by quarter: working spend by channel, test reserve, people, tools, agencies and freelancers, content and creative, contingency. The last row compares the total with the CAC ceiling and flags it if it’s over.
Before it goes to the board:
- Revenue split agreed with sales and finance
- Conversion rates taken from your own CRM, with the date range stated
- Spend phased by quarter and offset by the sales cycle
- Total sits under the CAC ceiling
- Sensitivity rows show the two assumptions the plan depends on most
- Test reserve and graduation rules written down
- Increase and cut rules signed off by finance
Get it built
If you want this model built from your own CRM and ad data, then run through the year rather than filed away, that’s the work I do. Most engagements start with a fixed-price Growth Audit, credited if we continue. See pricing or get in touch to pressure-test next year’s number.