Set your Meta ads budget from purchases, not gut feeling.

€10 a day buys purchases, not decisions

Yes, €10 a day can bring in sales. If one purchase costs €35, a €70 weekly budget produces about two purchases a week. The problem is not the number of purchases. It is what you can conclude from two.

You cannot conclude anything from two purchases. You do not know whether the ad was good or you got lucky, and neither does Meta's system. It shows your ad by guessing, because it does not have enough examples of who buys.

That is why ”is €10 enough” is the wrong question. The right question is: how many purchases a week do you need to make a decision, and what do those purchases cost?

Why the threshold is 50 purchases a week

Meta's own guidance is that an ad set exits the learning phase after about 50 optimization events within seven days. When you optimize for purchases, the optimization event is a purchase. Before that threshold, Meta is still searching for who to show the ad to, and results swing from day to day.

The threshold is not a penalty. It is statistics. The system needs enough examples of buyers before it can tell who resembles them. The same applies to you: two purchases cannot judge an ad, fifty can.

You can also fall back into learning. A significant change to the ad set, such as new targeting, a new ad or a large budget increase, restarts the count. With a small budget, make few changes and let them work.

The formula you can do yourself

You need two numbers: how much one purchase may cost (your target CPA) and how many purchases a week you want. The formula is:

With 50 purchases a week as the goal, the numbers look like this:

These are big numbers for many online stores. That is why it pays to separate two goals: do you want to find out whether Meta can work for your store at all, or do you want to scale something you already know works?

Testing profitability: 20 purchases show the direction

When you test for the first time, 15 to 25 purchases already show the direction. It is not enough for Meta to learn, but it is enough for you to decide whether to continue. Same formula, fewer purchases:

If you cannot reach even this level, you are still testing. You are just testing more slowly: the same number of purchases accumulates over months instead of weeks, and your conclusions are delayed by the same ratio.

Where the target CPA comes from

The target CPA does not come from Meta. It comes from your margin. Work out what one order leaves you after the cost of goods, shipping, payment fees and returns. That is the money that pays for both advertising and profit.

Break-even matters more than the target. It tells you when advertising starts eating money, and you need to know it before you raise the budget. If customers buy again, you can calculate break-even from a customer's first-year purchases, but then you also need to know how many actually come back.

What a small budget really costs

A small budget does not show up on the invoice. It shows up as time and as money lost to wrong decisions. In our experience, three costs repeat on almost every small-budget account:

1. Time. €10 a day and a €35 purchase mean 50 purchases in about 25 weeks. €250 a day does the same in a week. Both pay the same for 50 purchases, but one gives you the answer six months earlier.
2. False negatives. On a small budget an ad looks bad by pure chance, and it gets paused before it could show anything. Working ads die because nobody waited for enough data.
3. Impossible tests. You cannot compare two ads when each gets one purchase a week. You are comparing noise with noise and drawing conclusions from it.

When the budget is not enough for 50 purchases a week

For most stores starting out it is not, and that is not a dead end. It just decides how the account should be built. These five things help the most:

1. One campaign, one ad set. If a purchase costs €20, ten ad sets at €20 a day never learn anything, because each gets one purchase a day. One ad set at €200 gets ten and exits learning in five days.
2. Do not split audiences. Splitting interests into separate ad sets divides the same data into smaller piles. In our experience, broad targeting and a good ad win on a small budget almost every time.
3. Still optimize for purchases. Add-to-cart produces events faster, but then Meta learns to find people who add to cart and do not buy. Use it as a temporary step at most, and return to purchases as soon as you can.
4. Give it seven days of peace. Every significant change resets learning. Decide changes weekly, not daily.
5. Test in sequence, not in parallel. If the budget covers one proper test, run one proper test. Two half tests answer neither question.

Three goals, three budgets

Who this is not for yet

If your budget is €10 a day, you should not hire us or anyone else. Money meant for advertising would go to an agency fee instead of learning. Do the calculation above, run one clear ad to a broad audience and collect the first 20 purchases yourself.

A partner starts to make sense when the budget is enough for learning and the bottleneck is what to test and in which order. That is what we cover in Test first what can double the result.