A deliberately small ad test you can actually measure in a week
Before spending real money, run a tiny test with one product, one audience and one week, set up so the result shows clearly in your own traffic sources.
The advice you have been given is probably some version of just try a few pounds a day and see. So you spend £200 across four ads, three audiences and two products, the ad platform reports a number that does not match your bank account, and at the end of it you cannot say whether ads work for you. The money was not the waste. The design of the test was.
Why most small tests tell you nothing
The first problem is confounding. Change three things at once and a poor result has three possible explanations, none of which you can rule out. Most creators run their first test during a week when they were also posting organically about the same product, which makes the ad's contribution genuinely impossible to isolate.
The second problem is that the two sets of numbers do not agree. The ad platform counts a conversion using its own attribution window and its own definition of a view. Your storefront counts an order. These will differ, sometimes substantially, and the ad platform's figure is the one with an incentive attached. When they disagree, believe your own revenue.
The third and least comfortable problem is sample size. At small budgets you may buy 300 or 400 visits, and from that you might get three sales or you might get none, and both outcomes are consistent with a perfectly ordinary conversion rate. A tiny test can tell you a channel is catastrophically bad. It cannot reliably tell you it is good. Design the test knowing that, and you will not overreact to either result.
The usual test and a measurable one
Everything in the right-hand column exists to make one question answerable: did the money produce buyers I would not otherwise have had?
| The usual first attempt | A test built to be read |
|---|---|
| Several products promoted at once | One product, chosen because you already know it sells organically |
| Multiple audiences and creatives running together | One audience, one creative, so a result has one explanation |
| Runs alongside a busy organic week | Runs in a deliberately quiet week, with a note of anything else you posted |
| Budget large enough to hurt if it fails | A fixed amount you would spend on a night out, agreed before you start |
| Success judged by the ad platform's dashboard | Success judged by buyers and revenue on your own product page |
| No stopping rule, so it drifts on for a month | Seven days, then it stops whatever the numbers say |
Setting it up so the answer is visible
Spend an hour on the setup before you spend a penny on the ad. The setup is what turns the spend into information rather than a story you tell yourself afterwards.
- 1Pick one product with a known price and at least a few existing organic sales, so you have a baseline to compare against.
- 2Write down your break-even before you begin: budget divided by your revenue per sale, minus card fees, gives the number of sales the test needs to wash its face.
- 3Set the ad's destination to the product page, and add UTM parameters so the visits are labelled even when the referrer is stripped by an in-app browser.
- 4Note your current baseline: views, buyers and revenue for that product over the previous seven days.
- 5Agree the quiet week. Do not launch anything else, do not run a discount, and keep organic posting about that product to your normal level.
- 6Run the ad for exactly seven days at a fixed daily budget, and do not touch the settings mid-flight, because every edit restarts the learning and muddies the data.
- 7On day eight, record views, buyers and revenue for the product again and subtract the baseline. That difference, not the ad dashboard, is your result.
- 8Write the outcome and the decision on the same page as the plan, so the next test starts from something rather than nothing.
Reading the result honestly
Suppose the product is £29, the budget is £70 for the week, and Stripe's processing fees take a small slice of each sale. Break-even sits at roughly three sales. In the baseline week the product sold two copies. In the test week it sold six. The difference is four, which is above break-even but by one sale, on a sample small enough that a single buyer changing their mind would have flipped the verdict.
The correct conclusion is not that ads work. It is that the test did not fail, which earns you the right to run it again at a slightly larger budget and see whether the pattern repeats. If instead the test week had produced two sales, the same as baseline, you would have learned something genuinely useful for £70: that this creative, to this audience, for this product, does not move anything. That is a cheap piece of knowledge. Since store.fan charges 0 per cent platform fee and payments land in your own Stripe account, the arithmetic is at least clean, and you can see how that works on the pricing page.
Set up the product page you would send ad traffic to, before you spend anything on ads.
Build the page firstThe mistake most people make
The biggest error is scaling a small win. Four extra sales in one week feels like proof, and the temptation to multiply the budget by ten is enormous. Nearly always the rate drops as the budget grows, because the cheapest and most receptive slice of the audience is reached first. Increase in steps, and re-check the numbers on your own analytics at each step rather than assuming the first result holds.
The second error is testing a product that does not sell organically. Ads amplify what already works; they do not fix an offer nobody wants. If your product has never sold to a warm audience, you are buying strangers to look at something your own people declined. Fix the offer first, then test the traffic. The product pages are quick enough to rework that this is a smaller job than it sounds.
If the spend would buy you fewer than about 200 visits, you will not get a readable result. Either raise the budget slightly or accept that you are testing whether the ad delivers traffic at all, not whether it sells.
The product page. Sending paid traffic to a page with ten options gives every visitor a chance to choose none of them, and it makes the result impossible to attribute to anything specific.
Your storefront, because it counts orders that produced money. The larger figure usually reflects a wider attribution window that credits the ad for visits it only partly influenced.
Better not to, on a first test. A discount changes the offer as well as the traffic, and you will not know which one moved the result. Discount codes are a Pro feature and are worth keeping for campaigns you are not trying to measure cleanly.
A good test is small, boring and finished on a date you set in advance. Run one, write down what it told you, and you will be further ahead than most people who spent five times as much learning nothing they could repeat.
Open a free storefront and get the baseline numbers your first test will need.
Start free todayTurn your knowledge into income
Launch your Store.Fan in minutes — sell digital products, courses, and calls straight from your bio. Free to start.
