How to split your monthly marketing budget — the keep-grow-test rule

At the end of the month there's one number left: the amount set aside for marketing. Then comes the question — where does it go? Into ads, into content, into testing a new channel? Most business owners answer that question from scratch every single month, because there's no clear line connecting last month's result to this month's decision.

The pattern is familiar. One month the whole budget goes into a single channel because it performed well last time. The next month it's split five ways because every option looks promising. Neither is a strategy — both are guesses. The split below is built on one simple question: which money keeps what already works running, which money grows what's working well, and which money is spent purely to learn something.

First, split the budget into two separate lines

Before dividing anything, one separation has to happen: ad spend and the management fee should never sit on the same line. Ad spend goes straight to a platform — Google, Meta, TikTok — and if the account is opened in the business's own name, you can see exactly where that money went at any moment. The management fee pays for the work of the person setting the campaign up, watching it, and adjusting it — a separate job entirely.

In a model where the two are blended — say, "our fee is 20% of ad spend" — the fee grows automatically as spend grows, even when the amount of work stays the same. That's a straight path to a hidden markup on media. These two lines should be invoiced apart: no markup added to media spend, and the management fee standing on its own as a fixed arrangement. The split below is about ad and channel spend only — the management fee sits outside it and gets planned separately.

The three-bucket rule: keep, grow, test

Split the ad and channel budget into three buckets. The size of each one differs from business to business, but the logic behind them is the same everywhere.

The keep bucket — feeds what already works

This is the largest bucket, and it should be. It holds the channel that has delivered a proven result over the last two or three months — a search campaign bringing in most of the sales, for instance. The job of this money is stability: cut the budget here and your reach with an audience that already knows you drops, and a competitor fills the gap.

The grow bucket — pushes what's working, but hasn't hit its ceiling

This money goes to a channel that has already proven itself but hasn't maxed out yet — widening the audience, adding a new creative, opening a new region. The difference from the keep bucket is risk: the same result isn't guaranteed here, but the foundation is already there.

The test bucket — tries a new channel on a small amount

The smallest bucket, and the most expensive one to skip. This is where an untried channel, a new format, or a new audience segment goes. The rule is simple: if you lose it, the amount shouldn't sting. After a month of testing, the result either moves into the keep bucket or gets shut down entirely — there's no third option. "Let's give it a bit longer" is how a test budget quietly becomes a permanent one.

Where to start with a percentage

A precise percentage is hard to give, because it varies sharply by industry. The U.S. Small Business Administration's own blog notes that advertising spend averages somewhere from roughly one percent of revenue up to several, that retail tends to sit closer to 4%, and that B2C businesses can run as high as 10–12% (SBA's guide to marketing budgets). The number itself isn't a formula for the Azerbaijani market — the point is that saying "I set aside 10%" without knowing your own industry's average is a decision with nothing under it.

A more reliable starting point isn't a number at all, it's a ratio: a new, still-being-tested channel should get the smaller share of the budget, an established one the larger — never the other way round. If your test bucket keeps growing on its own, that usually means channels are graduating out of it into the keep bucket.

A worked example: a $1,000 month

Say your monthly ad budget is $1,000, and so far only one search campaign has been tested, with a steady result. In that case the keep bucket should be the largest share, since there's no other proven channel yet — say $700 stays in that campaign. Of the remaining $300, part of it (say $200) goes toward widening that same campaign's audience, and the last $100 goes toward testing a genuinely new channel for a month — Instagram ads, say.

By the end of the month there are three outcomes, not three separate decisions: the steady campaign keeps its place, the widened audience grows the grow bucket if it delivers, and the Instagram test either works (becoming a candidate for the keep bucket next month) or doesn't (it gets shut down, with the loss capped at that $100). The numbers will differ from business to business, but the logic doesn't: each bucket answers its own question, and none of them answers all three.

When to change the split

A split isn't a formula you set once and forget. The signals for revisiting it are clear:

• A channel in the keep bucket drops two months in a row — it has lost its "proven" status and needs re-evaluating, not automatic protection.
• A channel in the grow bucket hits its ceiling — extra money stops producing extra results and just reaches the same audience more often.
• The season changes — around a natural demand spike, every extra customer costs more even through a channel that already works, so it's worth shrinking the test bucket and growing the keep bucket instead.

The most expensive mistake: putting everything in one bucket

The same SBA guide makes a point of saying it plainly: don't put all your eggs in one basket. A channel performing well right now is no guarantee it performs well tomorrow — the platform's algorithm changes, a competitor moves into the same audience, the audience itself gets tired of seeing you. A budget tied to a single channel means the whole marketing effort stops the day that channel weakens.

The test bucket isn't there to win — it's there to learn. Keep the amount small enough that being wrong stays cheap.

You can build this split on your own, but telling which channel is genuinely "proven" and which has actually hit its ceiling usually needs an outside read — it's hard to see clearly from inside your own numbers. Our budget and channel-split audit exists for exactly that: we show, with data, which bucket should grow and which should shrink.