The question comes back every autumn, when next year gets planned: how much goes where. Most companies answer it in one of three ways. They roll over last year's split. They look at what a competitor is doing. Or they follow their agency's proposal. All three share the same flaw: none of them starts from the company. There is a simpler method, and it begins with three numbers you already have.
The three numbers that come before everything else
Your gross margin per sale. Not your revenue. What is left once the product or the service has been delivered. That is the only money that can pay for acquisition.
The value of a customer, not of an order. If someone buys three times a year for two years, your acquisition budget is not calculated against the first order. Many companies underinvest because they compare their acquisition cost to a basket instead of comparing it to a relationship.
Your sales cycle. The time between first contact and money in the bank. A three-day cycle and a six-month cycle cannot be run on the same budget or with the same patience.
Out of those three numbers comes the only ceiling that matters: what you can pay for a customer without losing money. Until that number is written down somewhere, no budget split can be defended.
The rule that comes before the split
No dollar into a channel you cannot measure end to end.
Not "measure" in the sense of seeing clicks. In the sense of knowing how many real customers that channel produced, and what they paid. If the chain is broken somewhere between the ad and the invoice, you are not buying growth, you are buying an opinion.
It is frustrating, because fixing measurement produces nothing to show in a meeting. But a budget spent into a blind channel never gets evaluated. It gets renewed, year after year, because nobody can prove it should stop.
Capture demand before you create it
Every marketing dollar does one of two things. It captures demand that already exists, or it creates new demand.
Search captures. Someone is looking for your product, you place yourself on their path. It is immediate, it is measurable, and it is capped by the number of people searching.
Social, content and PR create. Nobody was looking, someone discovers you. It is slower, harder to attribute, and it is not capped.
A company that has not yet captured all the existing demand has no reason to create more. Start by picking up what is already on the ground. When search saturates and the cost per customer climbs without volume following, that is the signal to move to demand creation.
The split, at last
Once you know what works, the split fits into three blocks.
Most of the budget on what is proven. Proven means: this channel produced measured customers, at a cost that stays under your ceiling, several months in a row.
A share on scaling. The proven channel, pushed harder, to find where the cost degrades. It always degrades. The point is to find the exact spot, not to avoid it.
A share on tests, with a stop rule written in advance. A test without a stopping criterion is not a test, it is a subscription. Write the maximum amount and the decision date before the first dollar goes out.
When nothing is proven yet, the whole budget goes to the channel closest to the purchase, and nothing else gets touched until a first result is measured.
Two things not to do
Spreading. An SMB can run two channels properly. Not five. Five channels on a divided budget means five campaigns that never reach the volume needed to learn anything, plus a month spent producing reports about noise.
Concluding too fast on a run of zeros. A channel that produced no sale in two weeks may not have failed, it may simply not have had enough volume to produce a readable result. Before shutting it down, take the account out of the day-to-day: how many times was the offer actually seen by someone in a position to buy. Often the right call is to reduce, not to close.
The question to ask before approving a budget
For every line: what would make me stop this one.
If nobody around the table can answer, that line is not an investment, it is a habit. Habits never remove themselves, they get renewed.
We hold that role
Part-time marketing leadership: the budget arbitrated line by line, with a stop rule on each one and someone who owns the decision. See our marketing leadership offer
Frequently asked questions
What percentage of revenue should go to marketing? No universal number survives contact with a real company. A brand at 70% margin and a company at 15% cannot invest the same proportion. Start from your margin and your customer value, not from an industry average.
What if nothing is measurable right now? Then the first budget does not go to advertising, it goes to measurement. It is a few days of work, once, and it changes the value of every dollar spent afterwards.
Should everything stop while measurement is repaired? Not everything. But what is expensive and indefensible has to stop. That is usually one or two lines, and they are easy to spot.