What percentage of revenue should go to marketing?

5–10% to hold position, 10–20% to grow, and more for early-stage companies buying their way into a category. These are starting points rather than rules — the right number depends on margin, sales cycle length and how much of your growth is referral-driven.

What is the difference between demand capture and demand creation?

Demand capture reaches people already looking for what you sell — search, comparison pages, retargeting. Demand creation reaches people who are not looking yet and makes them aware a solution exists. Capture is measurable and finite; creation is hard to attribute and is what makes capture cheaper over time.

How much should I spend on experiments?

Around 10%, ring-fenced, with permission to fail. Without a protected experimental budget every euro ends up in the channel that reports best, which is usually the channel that takes credit rather than the one that creates demand.

Should a small business do brand marketing at all?

Yes, but not the way large companies do it. For a small business brand work looks like publishing genuinely useful things, being consistently recognisable, and being known for one specific thing — not awareness campaigns.

How to Split a Marketing Budget

Updated 16 July 2026·3 min read·чети на български →

the_short_answer

Most companies should spend 5–10% of revenue on marketing when holding position and 10–20% when growing. Split it roughly 60/40 between capturing existing demand — search, retargeting, sales enablement — and creating future demand, then keep about 10% for experiments. The exact ratio matters far less than not spreading the budget thinly across every available channel.

Two questions, usually asked in the wrong order. Most companies decide the split before deciding the size, and both before deciding what the budget is supposed to do.

First: how much

The common benchmarks are 5–10% of revenue to hold position and 10–20% to grow. They are useful as a sanity check and dangerous as a target, because they ignore the two things that actually determine the number.

Margin. A business with 70% gross margin can spend far more to acquire a customer than one at 20%, and the percentage-of-revenue heuristic hides that entirely.

Sales cycle. If your cycle is nine months, spend committed today shows up in results next year. Companies with long cycles that budget quarterly consistently cut spend just before it would have worked.

The more useful calculation is bottom-up: what does a customer cost to acquire today, what is one worth over their lifetime, and how many more do you want? If you do not know the first two numbers, finding them is a better use of this quarter than reallocating anything.

Second: the split

Divide the budget by job, not by channel. There are three jobs.

Demand capture — roughly 50–60%. Reaching people already looking. Search, comparison and pricing pages, retargeting, the sales materials that close deals already in motion. It is measurable, it converts well, and it is bounded — you cannot capture more demand than exists.

Demand creation — roughly 30–40%. Reaching people who are not looking yet: publishing, events, partnerships, PR, anything that makes someone aware a solution exists before they need it. It is hard to attribute, which is why it gets cut first, and it is what makes capture cheaper every year. Companies that only capture find their acquisition costs rising annually and blame the platform.

Experiments — about 10%. Ring-fenced, with explicit permission to fail. New channels, new formats, new answer-engine surfaces. Without protection this money is always reallocated to whatever reported well last month.

Why even distribution fails

The instinct under uncertainty is to spread the budget across every channel so nothing is missed. This reliably produces the worst outcome available, for two reasons.

Every channel has a threshold below which it does not work at all — enough frequency to be remembered, enough data to optimise, enough content to matter. A budget divided eight ways clears the threshold in none of them.

And you learn nothing. Eight underfunded channels produce eight ambiguous results, so next year's decision is made on the same absence of evidence as this year's.

Two or three channels, funded past the threshold, beat eight funded below it — even if one of the eight would have been the best channel.

Where the money actually leaks

Paying for traffic to a page that does not convert. Fix the page before buying more visitors. Everything spent upstream of a broken conversion step is multiplied waste.
Buying demand while the positioning is unclear. Advertising an unclear proposition buys you a faster rate of being misunderstood.
Agency and tool fees counted as marketing spend. Track them separately, or you will conclude that a channel is expensive when the fee was.
Production costs eating distribution. A film that consumes the whole budget and leaves nothing to put it in front of anyone is a common and entirely avoidable failure.

The review that keeps it honest

Once a quarter, three questions:

Which channel produced customers, not leads? Lead volume is the number most likely to be optimised at the expense of the business.
What did we stop doing? A budget that only ever gains line items is not being managed.
What did the 10% teach us? If the answer is nothing, either the experiments were too timid or nobody looked at the results.

Questions people actually ask

What percentage of revenue should go to marketing?: 5–10% to hold position, 10–20% to grow, and more for early-stage companies buying their way into a category. These are starting points rather than rules — the right number depends on margin, sales cycle length and how much of your growth is referral-driven.
What is the difference between demand capture and demand creation?: Demand capture reaches people already looking for what you sell — search, comparison pages, retargeting. Demand creation reaches people who are not looking yet and makes them aware a solution exists. Capture is measurable and finite; creation is hard to attribute and is what makes capture cheaper over time.
How much should I spend on experiments?: Around 10%, ring-fenced, with permission to fail. Without a protected experimental budget every euro ends up in the channel that reports best, which is usually the channel that takes credit rather than the one that creates demand.
Should a small business do brand marketing at all?: Yes, but not the way large companies do it. For a small business brand work looks like publishing genuinely useful things, being consistently recognisable, and being known for one specific thing — not awareness campaigns.