The standard answer is a percentage of revenue — five percent, seven to eight percent for businesses in growth. It sounds authoritative and it is nearly useless in practice, because it takes no account of what a customer is worth to you or what one costs to acquire.
Start from the customer, not the revenue
Two numbers determine everything:
What is a customer worth? Not the first transaction — the total, over the length of a typical relationship, minus what it costs you to serve them. A pharmacy customer filling monthly prescriptions for four years is worth a very different amount from a single walk-in purchase.
What can you afford to pay to get one? A common working figure is a third of that lifetime value, which leaves room for delivery costs and profit. If a customer is worth twelve hundred dollars over their lifetime with you, four hundred to acquire one is defensible.
Once you have those, the budget follows from a target. Twenty new customers a month at four hundred each is eight thousand a month. That is the number — arrived at from the economics rather than from a percentage.
Where the reasoning breaks
Nobody knows their true lifetime value. Most small businesses have never calculated it. It is worth an afternoon with your records, because every subsequent decision depends on it.
Acquisition cost is not fixed. The first customers from a channel are usually the cheapest. Scale up and cost per acquisition rises as you exhaust the most responsive part of the audience.
Delivery is a constraint. If you can only serve fifteen new customers a month, buying twenty-five is not growth. It is a backlog and, eventually, bad reviews.
Splitting the budget
Search first, if people search for what you sell. Someone typing your service into Google has already decided they want it. That intent makes search the most efficient starting point in most businesses.
Social when the audience is not searching. Nobody searches for a job they have not decided to look for — which is exactly why driver recruiting works on Meta and why a plumber's budget usually should not start there.
Keep something back for testing. Ten to twenty percent on channels you have not proven. Most tests fail. The ones that work fund the next year.
The floor
Below a certain point, a budget cannot work. Advertising platforms need conversion volume to optimise, and a budget too small to generate that volume produces erratic results and no learning.
For most paid channels that floor is somewhere around a thousand to fifteen hundred dollars a month in media spend. Below it, the honest advice is usually to spend nothing on ads and put the effort into your Google Business Profile, your existing customers and referrals — all of which work at any budget.
We would rather tell someone that than take a retainer for a campaign that cannot succeed.