How Much Budget It Takes to Know Whether a Campaign Works

The question always comes: how much should I invest? Whoever answers with a flat figure, sight unseen, is guessing. The correct answer is built backwards, starting from how many conversions you need before you can say anything sensible.

The reasoning in three steps

1. Estimate the cost per action

If you have historical data, use it. If not, start from a prudent assumption based on sector and geography — declared as an assumption, not a forecast.

2. Define how many conversions you need

With five conversions you learn nothing: they could be chance. You need enough to tell a trend from a coincidence, and that number grows when the differences you're measuring are small.

3. Multiply, and add time

Estimated cost per action times required conversions gives the test budget. Add a minimum duration on top, because cramming it all into three days produces distorted data.

The most expensive mistakes

• Splitting a small budget across too many campaigns: none collects enough data.
• Stopping after a few days because the first numbers look bad.
• Changing several variables at once during the test.
• Increasing spend before verifying the tracking.

What if the budget isn't enough

Narrow the field: fewer audiences, fewer variants, one objective. Testing one hypothesis properly beats testing five without being able to conclude anything about any of them.

An insufficient budget split across too many tests doesn't produce partial results: it produces no results.