What the first budget is actually buying
It is not buying sales. It is buying an answer to one question: can this product be sold profitably through paid ads, at this price, to this audience?
That means the budget has to be big enough to produce a readable result, and small enough that being wrong does not hurt the business. Those two conditions, not a market average, define the number.
Work it out from your own numbers
Start from margin, not revenue. Take the price the customer pays, subtract the product cost, the delivery, the packaging, the payment or courier fees, and the share of orders that come back. What remains is what you have to spend on getting one order.
Decide how many results make an answer. One or two orders tell you nothing. You need enough orders that the result is not luck, and you need them within a window short enough to act on.
Multiply, then check against the business. Roughly, your test budget is the cost you can afford per order times the number of orders you need to believe the answer. If that number is more than you can lose, the honest move is to change the offer or the price, not to run a smaller test that cannot tell you anything.
What wastes a small budget
- Too many products at once
A small budget split across a catalogue produces noise. Pick the few you would be happy to sell a hundred of.
- Turning ads on and off daily
Every restart costs you the learning you already paid for. Set the budget, set the end date, leave it.
- Judging on day one
Early numbers swing hard on small spend. Decide the review date before you launch and do not look for permission to panic before it.
- Spending on ads before the page is ready
Traffic to a page that is slow, unclear on price, or hard to order from is the most expensive way to find out the page is the problem.
- Counting the wrong thing
With cash on delivery, count delivered and paid orders. A cheap cost per lead means nothing if those leads never take the parcel.
- No plan for the answer being "no"
A test that says "this does not sell profitably at this price" is a successful test. Decide in advance what you will change if that is the answer.
Ad spend and agency fees are two different things
Your ad spend goes to the platform and stays yours: your account, your data, your audiences. What you pay anyone to manage it is separate, and it should be clear which is which before you start. If you cannot see the split, you cannot judge either one.
Ask for your own account rather than running inside someone else's. It costs nothing at the start and it is very expensive to fix later.
What a first month should leave you with
Not just sales. You should end the month knowing which product moves, which hook stopped people, what an order actually costs you once it is delivered, and where the drop-off is between the click and the door.
If a month of spending did not produce those four answers, the money bought traffic instead of information, and the second month will be a guess as well.
The honest answer
There is no fixed starting budget. Work out what one delivered order can cost you and still leave margin, multiply it by the number of orders you need to trust the result, and only run the test if you can afford the answer to be no.