How to Allocate Your Marketing Budget

Allocation is where most marketing plans quietly fail. Not because the total is wrong, but because the money gets divided by department, habit and last year's spreadsheet rather than by the results each pound is supposed to produce.

How should you allocate a marketing budget?

Start from the growth target and work backwards, not from last year's plan nudged up or down. Divide the money against three buckets — demand you can measure, brand with a stated mechanism, and a ring-fenced experiment fund — and make every line defend itself by the outcome it owns.

The discipline is subtraction. Anyone can add channels; the skill is refusing to fund the ones that survive only because they existed last year. A useful exercise: zero-base the plan once a year. Assume no line has a right to exist, then re-justify each against the target. Most teams discover two or three activities that have been running on inertia for cycles, producing nothing anyone can name.

Three buckets in practice:

• Demand: the channels that produce traceable pipeline. This is where the majority of the budget lives and where returns should be clearest.
• Brand: investment that makes future demand cheaper and buyers warmer. It needs a mechanism, not a mood.
• Experiments: a protected fund for finding the next channel before the current one plateaus.

This only works when your total is right in the first place. If you have not set that, start with how much a B2B company should spend on marketing, then come back to divide it.

What is the 70-20-10 rule and when does it apply?

The 70-20-10 rule puts 70% of budget into proven channels, 20% into promising-but-unproven ones, and 10% into genuinely experimental bets. It is a sound default because it forces you to keep discovering growth while most of your money still sits on things that reliably work.

The rule earns its keep when you already have a proven channel to put the 70% behind. Early on, when nothing is proven yet, it inverts — you may run closer to 40-40-20, spending far more of the budget on discovery because you have not yet found the engine that deserves the 70.

Where teams misapply it is by letting the 70 harden into permanence. A channel that was "proven" two years ago can decay quietly — rising costs, falling conversion — while still commanding its 70% because no one revisited the label. Treat "proven" as a status that must be re-earned each planning cycle, not a badge that is awarded once.

How much should go to brand versus demand generation?

A common starting split is 60% demand to 40% brand, then adjusted for sales cycle and stage. Long, considered B2B purchases justify more brand, because buyers spend most of the cycle out of market and choose the name they already trust when they finally enter it. Short-cycle or survival-mode businesses tilt harder towards measurable demand.

The reason brand gets underfunded is structural: demand shows up in this quarter's dashboard and brand shows up in next year's win rate. When budgets tighten, the line without an immediate number attached loses. That is precisely backwards for a considered purchase, where 90-odd percent of your future buyers are not looking today and the entire job of brand is to be the default when they start.

To protect the split, hold brand to the same standard as demand — a stated mechanism and a way to observe it working:

• Give brand a job in a sentence: "so in-market buyers already recognise us before sales makes contact."
• Track leading signals — branded search, direct traffic, unprompted recall, share of voice — rather than demanding last-click revenue it was never designed to produce.
• Resist the urge to reclassify brand as demand just to make it measurable. That is how you end up with expensive lead-gen wearing a brand label.

How do you decide what to cut?

Cut against marginal contribution, not average performance. Rank every line by what the last pound into it produced, and remove from the bottom until the remaining lines all clear your return threshold. Habit, sunk cost and internal politics are not reasons to keep a line alive.

The average trap catches everyone. A channel with a flattering blended return can be dead at the margin — the first slice works, the last slice does nothing — and the average hides it. Always ask what the next increment would do and what the last increment did. If neither has an answer, you have found something to cut.

A short sequence for a squeeze:

1. Kill activity nobody can attach a mechanism to. If no one can say what it is for, it goes first.
2. Trim channels past their efficient frontier — the point where more spend stops moving the number.
3. Pause, do not delete, anything strategic but slow. Brand is a candidate to reduce, rarely to abolish.
4. Protect the experiment fund. Raiding it feels free and quietly ends your pipeline of future channels.

The prerequisite for all of this is trustworthy measurement. You cannot cut on marginal return you cannot see, so get honest about how you measure marketing ROI before you start cutting on instinct.

How often should you revisit the allocation?

Revisit the mix quarterly and the total annually. Quarterly is frequent enough to reallocate towards what is working and away from what has decayed, without the thrash of chasing every weekly wobble. The annual pass is where you zero-base the total and reset the strategy.

Between formal reviews, let evidence trigger movement. If a channel clears its threshold with room to spare, feed it before the next quarter rather than making it wait. If one falls below, throttle it early. Rigidly frozen budgets are as damaging as constantly churning ones — the aim is disciplined responsiveness, not annual paralysis or monthly panic. This kind of ongoing reallocation is the core of our growth marketing work, and it is where most of the compounding gains actually come from.

The takeaway

Allocate from the target down, into demand you can measure, brand with a mechanism, and a protected experiment fund. Use 70-20-10 as a default once you have something proven to anchor it, cut on marginal return rather than average, and revisit the mix every quarter. The total sets the ceiling; the allocation decides whether you hit it.