The most expensive mistake on Meta with a small budget is not made a week in, looking at the numbers. It is made in the first hour, setting up the structure.
And it is almost always the same one: three campaigns to "test the audiences", two ad sets each, and $15 a day spread across six containers getting $2.50 apiece. None of the six learns anything.
Why does splitting a small budget not work?
Meta optimises per ad set, not per account. Each ad set has to gather enough conversions to work out who to show the ads to, and the indicative threshold Meta itself states is around 50 conversions a week to exit the learning phase: precisely, about 50 results in the week after the ad set's last significant edit. The same page advises against high ad volumes, because combining similar ad sets also combines their learnings.
At $15 a day with an $8 cost per lead, that is about 13 conversions a week. On a single ad set it is thin, but the system moves. Split six ways it is two conversions each: the algorithm stays blind, and stays blind permanently.
On a small budget you do not test audiences. You test messages.
How should you structure a small-budget campaign?
- NOIOne campaign — Leads or Sales depending on what you actually want, not on what sounds better
- NOIOne ad set — broad audience, the country and a wide age range. No interests, no clever exclusions
- NOIThree to five creatives inside it — that is where the testing happens, not in the audiences
- AIOne optimisation event, the most frequent one that still means something commercially
- NOISeven days without touching anything — every edit restarts learning
Broad audiences unsettle anyone who learned this a few years ago, back when you built layered stacks of interests. Today that instinct works against you. Fine targeting on a small budget shrinks the pool, raises the cost, and takes away the room the algorithm needs to find the right people by itself.
The event you optimise for, which is the decisive choice
This is the part people get wrong most often, and it costs more than all the others put together.
Someone selling a $3,000 service sets optimisation to "Purchase" or some deep funnel event, because that is what they care about. Then at $20 a day they collect one conversion every ten days, and the algorithm has nothing to learn from.
| If the budget is | Optimise for | Why |
|---|---|---|
| under $20/day | View content or Lead | you need volume, not depth |
| $20-50/day | Lead | balance between volume and quality |
| over $50/day | the real conversion event | enough data arrives |
The rule under the table is single: optimise for the deepest event that can still gather roughly fifty conversions a week. If it cannot, move one step up the funnel. It is also the fix Meta gives for "Learning limited" ad sets: choose a more frequent optimisation event, for example moving from purchases to add to cart, alongside expanding the audience and combining ad sets and campaigns.
The four mistakes that burn the budget
Touching the campaign every day. Every substantive edit — budget, audience, event — restarts the learning phase. Anyone checking and adjusting each morning keeps the campaign permanently learning. Meta's list counts any change to targeting, creative or optimisation event as significant, along with adding a new ad and pausing for 7 days or longer; budget changes depend on their size.
Judging after two days. The first days are the most expensive and the least representative. Wait seven full days. Or wait until you have enough conversions for a number rather than an anecdote.
Launching on a new account at full budget. A freshly created account running aggressive campaigns is the fastest route to getting suspended. Start slow and climb.
Sending traffic to an untracked page. If the pixel is badly installed, or the event does not fire, Meta optimises blind and the money goes without you even knowing why. Tracking gets verified before you switch on, not after. Alongside the pixel there is Meta's Conversions API, which sends events from your server for measurement, reporting and optimisation. It is exactly the kind of check worth automating, which we cover in the AI agents guide.
How much do you need to spend to know if it works?
The minimum useful number is not in currency, it is in conversions: below 30-50 you decide nothing, whatever you have spent.
In practice: if the expected cost per lead is $8, you need $240-400 before you have a grounded opinion. At $15 a day that is two to three weeks. Anyone without that budget does not have a campaign problem: they have an expectations problem, and it is worth knowing that before starting rather than after.
The method we use to alternate machine and people on campaigns is in the vibe marketing guide.
Frequently asked questions
What is the minimum budget for Meta Ads?
There is no technical minimum, but below $10-15 a day the algorithm gathers too little to optimise. The real constraint is not what you spend per day, it is how many conversions you gather per week: below thirty or so nothing can be decided, whatever you have spent.
Is it better to run one campaign with several ad sets or several separate campaigns?
On a small budget, one campaign and one ad set. Each ad set needs its own conversions to leave the learning phase, so splitting a handful across several containers means none of them learns. Creatives get tested inside the same ad set.
Should I use interest audiences or a broad audience?
On a small budget, broad. Interest targeting shrinks the pool and raises the cost, and takes away the room the algorithm needs to find the right people. Interests become useful once volume is high and you need to segment by message, not before.
How long should I wait before editing a campaign?
At least seven full days without touching anything, because every substantive edit restarts the learning phase. Adjusting every morning keeps the campaign permanently learning and never produces a readable number.
Which event should I optimise for?
The deepest one that can still gather roughly fifty conversions a week. If purchases are too rare for the budget you have, move one step up the funnel: optimising well on an intermediate event beats optimising badly on the final one.
Why is my campaign spending less than the budget I set?
Usually because the audience is too narrow, the optimisation event too rare, or the account is new and Meta is being cautious. The first two are fixed by widening; the third is only fixed by time, and forcing it with sharp budget increases makes it worse.
Want us to look at your campaigns?
We will tell you where the budget is burning and what we would change first. Seven questions, founders answer within 48 hours.
ApplyKeep reading
Founder of As Growth, an AI-first growth partner. He runs growth and strategy across partner businesses: OroEtic, Giojello, Equity Plus and others. Apply if you want to work with us.