How to allocate a marketing budget

Allocate by measured return per channel, and reserve a slice for finding out about the ones you have not measured.

Most budgets are allocated by last year's split adjusted slightly, which entrenches whatever was true before anyone measured.

Step-by-step

1. Enter your total budget.
2. Enter each channel's spend and results from the last period.
3. Read the return per channel and a suggested allocation.

Cost per acquisition, not cost per click

A channel with cheap clicks and poor conversion is expensive. What matters is cost per customer acquired, and beyond that, the value of the customers a channel produces — channels differ markedly in the quality of what they deliver, not only the quantity.

Keep a slice for testing

Allocating everything to what works today means never discovering what might work better, and channels decay — costs rise, audiences shift, platforms change their rules.

Around 10–20% for testing is the usual guidance. Test with enough budget to produce a real signal; a fortnight at a trivial spend teaches nothing except that small samples are noisy.

Attribution is harder than it looks

A customer who saw a social post, searched for you, read a review and then clicked an ad is usually attributed entirely to the ad. That flatters the last channel and starves the earlier ones — and cutting the "unmeasurable" channels often reduces the performance of the measurable ones, which is a confusing result if you believe the attribution.

Include time as cost

Organic channels look free and are not. Content, social and community work cost hours, and hours are the scarcest resource in a small business. Cost them at what your time is worth before comparing against paid channels.

Frequently asked questions

How much should I spend on marketing?

It varies by sector and stage; a share of revenue is the usual framing. What matters more is that spend is allocated by measured return rather than by habit.

How much should I reserve for testing?

Around 10–20%, with enough behind each test to produce a real signal. Channels decay, so a budget with no testing gradually stops working.

Why does attribution mislead?

Because it usually credits the last touch. Earlier channels that created awareness get nothing, so cutting them looks free and then quietly reduces the performance of everything downstream.