Written by
BOLD
6 min read
Section
For business owners who are about to start paid ads, or are in their first month, and want to know when it's fair to say "this isn't working".
Day three. You open Ads Manager and see €60 spent and no purchases. The conclusion feels obvious: ads don't work for your business. Or the opposite happens - three sales on day two, and you're already planning how to spend the profit.
Both reactions are built on far too little information. Early numbers are noisy, and the decisions people make from them are some of the most expensive mistakes in paid advertising - either stopping something that would have worked, or scaling something that was just lucky.
Why early numbers lie
Flip a coin ten times and you might get seven heads. Nobody concludes the coin is broken. Ads behave the same way, just with money attached.
Say roughly 2 in every 100 people who click on your ad end up buying. Send 100 clicks and you'll on average get two purchases - but getting zero, or five, would not be unusual. So a campaign showing "0 purchases" after €60 might be terrible, or it might be perfectly fine and just unlucky. From the numbers alone, you can't tell.
There's also a second reason, which comes from the platform itself. Meta's own guidance says an ad set needs about 50 optimisation events - purchases, if that's what you're optimising for - within seven days to leave the "learning phase" and deliver predictably. Until then, costs swing around, and the numbers say more about the system exploring than about your offer.
Two thresholds, not one
"How long until I can judge?" is really two different questions.
When can I stop a single ad? A common rule of thumb: once an ad has spent two to three times what a purchase is worth to you, with no purchases at all, that's a fair signal to turn that ad off. Not the campaign. That one ad.
When can I judge the whole campaign? When you have enough time and enough purchases. As a rough guide, at least two weeks and at least 30 purchases. Fewer than that and you're still reading the coin flips.
Both thresholds come from one number: what you can afford to pay for a customer. That's your break-even acquisition cost - what you earn on a first order after product cost, shipping, payment fees and returns. Here's what the minimums look like at different price points:
(Insert your thresholds table here: cost per purchase from €10 to €300 against the judging thresholds and daily budget required.)
These are rules of thumb, not laws. But look at the right-hand column before you plan a budget.
The uncomfortable part
For many businesses, the honest minimum is bigger than the budget.
Take a product where a purchase costs you around €100, and a monthly ad budget of €1,500 - about €50 a day. A fair campaign verdict needs roughly €3,000, which is two months of budget. And at €50 a day, the ad set will probably never gather 50 purchases a week, so Meta will keep it in "learning limited", with costs bouncing around.
That doesn't mean ads are pointless for you. It means nobody can honestly tell you in ten days what the maths says takes eight weeks. If you're in this position, you have real options:
Use a single ad set. Every extra ad set splits the same money into pieces too small to learn from.
Optimise for an earlier step, such as add-to-cart or initiate-checkout. The signal is weaker, but it happens often enough for the system to learn. Just know you're judging a proxy, not sales.
Lengthen your window. Judge on 14 to 28 days, not 7.
Fix the site first. If a small budget is all you have, a page that converts better makes every euro of it go further.
In a small market like Lithuania there's a further squeeze: audiences are limited, so a tight budget also raises how often the same people see your ad. Be careful about setting up narrow targeting on a small budget.
Time matters as much as money
Even with enough budget, judging too early goes wrong for reasons that have nothing to do with the ads.
Behaviour changes through the week - weekends often look different from weekdays - so a window shorter than seven days gives you a lopsided picture. Purchases can also take a few days to be attributed, so day-one and day-two results are usually understated. And any large edit resets the learning process: swapping creative, changing the audience, or moving the budget by more than roughly 20% in one go can all restart it.
So the minimum is: don't judge before day seven, don't make big edits in the first three to five days, and when you do change things, change them together rather than one setting a day.
What you can judge early
Waiting doesn't mean flying blind. In the first few days you can't tell whether the campaign is profitable, but you can tell whether something is broken.
Is tracking working? On day one, make a purchase yourself and check that it appears in Ads Manager with the right value. If tracking is off, every number after that is meaningless.
Is the ad getting attention? Look at link click-through rate against your own other ads, not against internet averages. A very low number means the creative isn't landing.
Are clicks turning into page views? If many clicks never become a loaded page, your site is slow or the link is broken.
Are people moving down the funnel? Cost per add-to-cart is a useful early signal - not proof, but a direction.
These tell you whether to fix something. They don't tell you whether to stop.
What not to do
Don't judge on day-one ROAS. Attribution lags, sample sizes are tiny, and the platform is still exploring.
Don't split a small budget across many ad sets. €30 a day across five ad sets is €6 each, and none of them will ever gather enough data.
Don't edit constantly. Every significant change restarts learning. A campaign edited every other day may never have had a clean week.
Don't move the goalposts. If you decide what counts as success after you see the numbers, you'll always find a reason to keep going or to quit. Decide beforehand.
Before you launch: write down four things
The most you can pay for a customer without losing money (your break-even acquisition cost).
How much you're willing to spend to find out if the campaign works, and whether that's actually enough per the table above.
The earliest date you'll judge it.
What you'll do if it works, and what you'll do if it doesn't.
This takes ten minutes, and it's the difference between a test and an expensive guess.
The short version
Early ad numbers are mostly noise. Work out your break-even cost per customer, then use it to set a minimum: a few multiples of it to stop a single ad, and roughly 30 purchases over at least two weeks to judge a campaign. If your budget can't reach that, don't pretend it can - use one ad set, optimise for an earlier step, and judge over a longer window. And decide the rules before you press launch.
Not sure if your test budget is big enough to tell you anything? Send us your numbers and we'll tell you honestly what you can and can't learn from them - including if the answer is "wait". Book a free intro call →