how much should you spend.

every founder eventually asks what percentage of revenue should go to marketing, and every honest answer starts by explaining why the question is upside down. the percentage is an output of your economics, not an input to them. still, it is a useful sanity check — so here is the benchmark table, followed by the method that should actually set your number.

the reason the question persists is that percentages feel safe. a board approves 8% of revenue more readily than a number derived from assumptions about payback. but a percentage that ignores your margin can approve a budget your unit economics cannot support, or refuse one that would have been comfortably profitable — both of which are worse than an argument about assumptions.

— 01the benchmark, and its limits.

the ranges below are what we see in practice across the categories we work in. read them as a sanity check: if you are far below the band for your category, you are probably under-investing relative to competitors; far above, you may be buying growth at a loss.

the limits are important. a percentage rule ignores margin, which is the single most important variable — a business at 70% gross margin can spend twice what a 35% business can on the same revenue. it ignores growth stage, since launching costs disproportionately more than maintaining. and it ignores channel mix: a brand with strong organic and word of mouth needs less paid than one starting cold.

use the band to notice an anomaly, then throw it away and do the arithmetic below.

one regional note on the bands: acquisition costs in the uae and ksa have risen steadily as more brands moved budget to paid social, so a percentage that worked three years ago buys less attention now. if you set your budget as a fixed share of revenue and have not revisited it since, you have quietly been reducing your real spend every year.

— 02the method that actually sets the number.

start with contribution margin per order or per customer — revenue minus product, shipping, payment and fulfilment costs. that figure is the most you could pay to acquire a customer before losing money on the first transaction.

then decide your payback tolerance. if you can wait three months to recover acquisition cost, and a customer contributes margin monthly, you can spend up to roughly three months of that margin. if you sell once, first-order margin is your ceiling and you should target well inside it.

multiply your target acquisition cost by the number of customers you need for your growth plan, and you have a budget. it is arithmetic rather than a benchmark, and it produces a number you can defend to a board — and adjust the moment your close rate or margin changes.

one caveat for the region: if cash on delivery is material, use delivered margin rather than gross. refused orders quietly destroy this calculation.

build in a test allocation as well — somewhere between 10 and 20% of the budget for things you cannot yet justify. new channels, new formats, a creator you have not worked with. without it, the budget optimises into whatever worked last quarter and you find out too late that the channel has saturated. the test line is what keeps you from being surprised.

— 03agency fee versus media spend.

keep these separate in your planning, always. media spend buys attention and scales with ambition. agency fees buy the work that decides whether the media spend is any good, and they do not scale linearly — the same team can manage a substantially larger budget without a proportional fee increase.

a reasonable starting split for a small to mid-size account is roughly two-thirds media and one-third fee, moving further toward media as spend grows. what you should not do is convert the fee into a percentage of spend, because that reintroduces the incentive problem: your supplier then profits from a bigger budget rather than a better outcome.

and if the total is small — under a few thousand a month all in — spend it on one channel done properly rather than splitting it three ways. a small budget spread thin produces three sets of inconclusive data.

— 04when to raise it, and when not to.

raise the budget when payback is comfortably inside your tolerance and the constraint is genuinely reach. that is the only condition under which more money reliably produces more customers.

do not raise it when conversion is below your category norm, when creative has been unchanged for a quarter, when the enquiries you already get are not being answered quickly, or when you cannot say what a customer costs. in every one of those cases, additional spend amplifies an existing problem — which is the most expensive way to discover you had one.

review the number quarterly rather than annually. margin, close rate and platform costs all move, and a budget set twelve months ago is answering last year's question. the review does not have to be a rebuild — usually it is one calculation and a decision to hold, raise or move money between channels.

if this is the problem: ecommerce with no budget, a shoot on a small budget, setting a social budget in dubai.