Ad budget for small business: how much you need to learn anything

Aug 18
3 min read

Updated: 5 days ago

Introduction

There are two questions hidden inside this one. How much can I afford, and how much do I need for the spending to teach me anything? They have different answers, and the second is the one that usually gets ignored.

A budget below the learning threshold does not produce a small result. It produces no usable information, which is worse than not spending at all.

1. Set an ad budget for small business from customer value

Start with what a customer is worth: the gross profit an average customer produces across their whole relationship with you.

Divide by three for a working maximum acquisition cost. Multiply by the number of additional customers you want in the period. That is a budget with a rationale, rather than a figure that felt comfortable.

If a customer produces $210 in gross profit, your ceiling is around $70, and twenty additional customers means roughly $1,400 for the period.

2. Check it clears the learning threshold

Now the second question. Platforms need a certain volume of conversions before their optimisation performs predictably — below it, results swing widely for reasons unrelated to your decisions.

Practically: your budget over the test period needs to buy enough conversions that a difference between two options would be visible rather than plausibly random. A handful is not enough.

If your target acquisition cost means the budget produces only a few conversions in the period, either extend the period or optimise for a reliable earlier step — an enquiry, a booking — where volume is higher.

3. If it does not clear the threshold, do not split it

This is the crucial consequence. A budget that is marginal for one campaign is useless spread across three.

Concentrate everything on the single channel that best matches your situation. Accept narrower coverage in exchange for actually learning something. Running two channels badly is the most expensive option available.

4. Budget for the whole period, not the month

Small budgets need patience more than large ones, because they take longer to accumulate signal.

Decide the total for a test period — commonly a couple of months — and commit to it. Judging a campaign after a fortnight of a small budget means judging noise, and changing it resets the platform's learning, which starts the accumulation over.

Set the period, spend it, then decide.

5. Include everything, not just media

The media spend is not the budget. Include production, tools, and any first-purchase discount you offer, since a discount given to acquire someone is an acquisition cost even though it appears as reduced revenue.

Businesses that count only media consistently understate their true acquisition cost, sometimes by enough to turn a profitable campaign into an unprofitable one on paper — or the reverse.

6. Scale on evidence, not on optimism

Once cost per customer is comfortably below your ceiling, increase the budget. You are buying profit, and the constraint is how much of it you can buy.

Increase gradually rather than doubling. Large jumps disturb the platform's optimisation and efficiency usually falls at least temporarily, which makes the result hard to read.

Watch cost per customer as you scale. It typically rises with volume, and the point where it reaches your ceiling is your practical limit on that channel.

7. Know when to reduce it

Cutting is sometimes the right answer, and it is worth deciding the trigger in advance.

Reduce or pause when cost per customer exceeds your ceiling for a sustained period and you have diagnosed why. Reduce during a genuine seasonal trough rather than paying premium rates for absent demand. And reduce when you cannot service more customers, since demand you cannot fulfil is worse than demand you did not create.

Conclusion

Derive the budget from customer gross profit, then check it clears the volume needed to learn something. If it does not, extend the period or optimise for an earlier step — and never split it.

Commit for a defined period, count production and discounts as part of it, scale gradually once cost per customer sits below your ceiling, and set the reduction trigger before you need it.