With a limited budget, every wrong euro weighs more. This is the guide I wish I had when I managed my first Meta campaigns for a small company: practical, without unnecessary theory.
Before opening Ads Manager: what you need to have ready
Before even creating the first campaign, you need three things: the Meta Pixel installed and working on the site, with the conversion events that really matter to you (purchase, lead, contact); at least three or four different creatives, between vertical videos and static images, because a single creative is not enough to test what works; and a clear idea of the starting audience, even just by age and geographic area.
How to structure the first campaigns, without overdoing it
With small budgets the temptation is to create many different campaigns to test everything at once: different audiences, different messages, different formats. It is the most common and most costly mistake. Meta needs a minimum volume of data to exit the learning phase and optimize a campaign well: if the budget is split across too many campaigns or ad sets, none of them ever reaches that volume, and results stay random.
It is better to start with a single campaign, a single ad set with a fairly broad audience (Meta works well when it has room to optimize on its own), and three or four different creatives inside the same ad set, letting the algorithm work out which one performs best.
Starting budget: how much is "enough" to get readable data
There is no universal figure, because it depends on the cost per result in your sector, but a useful practical threshold is: the daily budget should allow at least one result (lead, sale) every two or three days in the first weeks, otherwise the data volume is too low to optimize reliably. With very small budgets, it is better to test a single message angle at a time, instead of dividing the resources further.
The most common mistakes in the first 90 days
- Changing creatives or copy after only a few days, without giving the algorithm time to collect enough data.
- Narrowing the audience too much with very specific interests, thinking it is more "precise", when it often only reduces the volume available for optimization.
- Looking at cost per click instead of cost per final result, which is the only number that really matters for the business.
- Not having a landing page designed for social traffic, different from the one designed for people arriving from a Google search.
When it is time to scale
A campaign is ready to receive more budget when the cost per result has been stable for at least one or two weeks, not on a single good day. At that point it is better to increase the budget gradually, not double it all at once: too large a jump restarts the algorithm's learning phase, with a temporary worsening of results.