Daily Budget for Facebook Ads: Why Meta Predicts Zero

The last screen I saw before my campaign went live told me it would not work. "Your campaign may get 0 daily results with your current budget," it said, followed by an offer to double my spend. That warning is the single most useful thing Meta shows a small advertiser, and almost nobody explains what is actually behind it. If you are setting a daily budget for Facebook ads and wondering whether your number is too small, the answer is not a feeling. It is arithmetic, and you can do it in about thirty seconds.

I am not a Facebook ads expert. I run a marketing consultancy and a community, and paid social is the one channel I have deliberately avoided for years because it always felt like a hundred moving parts with Meta whispering instructions at every step. So I set up a real campaign, on real money, and recorded the whole thing. What follows is what I learned, including the part where the platform told me my budget was pointless.

https://youtu.be/9hRgGtmDa1w

What a daily budget for Facebook ads actually controls

A daily budget for Facebook ads is not a spending cap. It is the average Meta aims to spend per day across a week, and the platform is explicit that it can and will go over on any given day. My own budget screen spelled it out: an average of ten a day, a maximum daily spend of seventeen fifty, and a maximum weekly spend of seventy. The weekly number is the real ceiling. The daily one is a target.

That distinction matters more than it sounds. Advertisers set a small daily figure expecting a hard limit, then panic when day three comes in 40% higher. Meta is not overspending, it is front-loading toward whichever days the auction looks cheapest. As long as the seven-day total holds, the system is behaving correctly.

The second thing a daily budget controls is far more important, and it is the one nobody mentions. Your budget decides whether Meta's delivery system can learn anything at all.

Why Meta predicted zero: the 50-event problem

Meta's warning was not a sales tactic. Every new ad set enters what Meta calls the learning phase, during which the delivery system tests audiences and placements to work out who converts. According to Code3, Meta generally requires around 50 optimisation events per ad set per week to exit that phase. Below that threshold, the ad set is flagged "Learning Limited" and performance stays volatile.

Fifty events a week is the number that governs everything else. It is not fifty clicks and not fifty impressions. It is fifty of the specific event you chose to optimise for, which in my case was a purchase.

The maths that decides it

Here is the calculation, and it takes seconds. Multiply your target cost per conversion by 50, then divide by 7. That is the daily budget an ad set needs to exit the learning phase in a week.

Now put real numbers in it. Triple Whale analysed nearly 35,000 brands across the whole of 2025 and put the average Meta cost per acquisition at just under 40, with an average CTR of 2.19% (their figures are in dollars). Run that through the formula and you need roughly 275 a day per ad set to hit fifty purchases in a week.

I was spending ten. So Meta was not being pessimistic when it forecast zero daily results. At an average cost per acquisition, a ten-a-day budget buys about one conversion every four days, which is a fifth of what the learning phase needs. The warning was simply the platform showing its working.

Pro tip: run that formula before you write a single ad. If the number it gives you is wildly beyond what you can spend, you have three honest options, and picking one deliberately beats discovering it three weeks in.

Your three options when the maths does not work

The first option is to optimise for a cheaper event. A purchase is the most expensive thing you can ask for. Optimising for a lead, a sign-up, or a content view produces far more events for the same money, which gets the ad set out of learning and gives the algorithm something to work with. You are trading intent for volume, deliberately.

The second option is to accept you are buying data, not conversions. A small budget still tells you which image people respond to, what your real cost per click looks like, and whether your landing page converts anything at all. That is worth having, as long as you know that is what you bought.

The third option is to not run ads yet. If ten a day is genuinely your ceiling and your only conversion event is a purchase, that money almost certainly does more work as content marketing for a small business, because organic content compounds and paid traffic stops the moment you stop paying. I say that as someone who was mid-way through building a campaign at the time.

I picked the second option, knowingly. My goal was to understand the machine, not to squeeze a return out of a first attempt.

The setting that wasted my last campaign

Before this campaign, I ran one for traffic. It worked exactly as advertised, which was the problem. A few hundred people landed on my about page, and not one of them converted. Meta did its job perfectly: I asked it to buy visits, so it bought visits, and it optimised for the people most likely to click and least likely to do anything afterwards.

That is the trap in a single sentence. A traffic objective tells the algorithm to find clickers, and clickers are a genuinely different audience from buyers. You are not getting a cheaper version of the same people. You are getting different people.

So this time I set the campaign objective to sales and the ad set to maximise the number of conversions, rather than landing page views or link clicks. That change alone reframes the whole account, because now every metric you look at is measured against something that matters to the business instead of something that just looks busy.

None of that works without tracking. My Skool community's Meta pixel is connected to the ad account, with the conversion event set to purchase. Without a pixel firing a real event, "maximise conversions" is an instruction the platform cannot follow, so it quietly falls back to what it can measure.

What to cut when your budget is small

A small budget spread across every surface Meta owns is a small budget wasted. My ad set defaulted to 17 included placements, so I cut it back to Facebook and Instagram and excluded four, including Audience Network, Threads and Messenger. Concentrating spend gives each surface enough volume to produce a signal instead of a rounding error.

Then I found the checkbox that most people scroll straight past.

Underneath the placement list sits an option called "Allow limited spending to excluded placements", and the small print says Meta will aim to spend around 5% of your budget on each excluded placement. Read that again. You exclude four placements, leave the box ticked, and a fifth of your budget can go to the exact surfaces you just turned off. On a large account that is noise. On a small one it is the difference between a clean test and an unreadable one.

Untick it. It takes two seconds and it is the highest-value click in the whole setup.

The setup, step by step

Here is the order I built the campaign in, with the decision that matters at each stage.

1. Name the campaign plainly. Mine is "Test Ad for Clicks and Sales". Naming conventions matter at scale, and at one campaign they do not, so do not stall here.
2. Set the objective to sales, not traffic. This is the single highest-impact choice in the whole build. Everything downstream inherits it.
3. Connect the pixel and pick the conversion event. Confirm the event actually fires before you spend anything. Test it yourself and watch it appear.
4. Set the budget at campaign level. Then run the 50-event formula against your target cost per conversion so you know, before launch, what the budget can realistically produce.
5. Choose conversions, not landing page views. In the ad set, maximise number of conversions. The recommended option is correct here, which is not always true elsewhere.
6. Trim placements and untick the 5% box. Facebook and Instagram only, and no limited spending to excluded placements.
7. Build variations that change one thing. I published six image ads with identical primary text, headline and call to action. Only the image differs.

That last point is worth dwelling on. I deliberately kept the copy identical across all six ads, because if I change the image and the text at the same time, a winner tells me nothing about why it won. One variable at a time is slower and it is the only way the result means anything.

I also declined every Advantage+ enhancement Meta offered on the creative: no automatic text tweaks, no added animation, no image adjustments. If the platform rewrites my ad mid-test, I am no longer measuring what I built.

What to actually expect

Expect the first few days to look bad, and expect that to be normal rather than diagnostic. During the learning phase, cost per result swings wildly while the system tests, and reacting to day-two numbers is how advertisers reset learning over and over without ever reaching a stable state. Every edit to targeting, budget or creative restarts it.

Expect Meta to keep asking for more money. Mine recommended seventeen a day on the grounds that similar businesses spend that and get nine conversions. That is a real benchmark rather than a random upsell, and it is still not a reason to spend money you have not decided to spend. Meta's own best practices for minimum budgets exist precisely because the platform knows undersized budgets underperform.

Expect, above all, to learn something specific. In our experience with small business campaigns, the first run is worth it for one thing: finding out which single image makes a stranger stop. That answer transfers to organic content, thumbnails, and every ad you run afterwards, which is a return that outlives the campaign.

What I will not do is tell you it worked, because I do not know yet. The campaign published with one campaign, one ad set and six ads, at ten a day, with a warning attached. Honest reporting on that comes after the data does.

FAQ

What is a good daily budget for Facebook ads?

A good daily budget is your target cost per conversion multiplied by 50, divided by 7. That produces the roughly fifty weekly optimisation events an ad set needs to exit Meta's learning phase. If the resulting figure is out of reach, either optimise for a cheaper event such as a lead or a sign-up, or accept that the campaign is buying data rather than conversions.

Can you run Facebook ads on a small budget?

Yes, but change what you ask for. A small budget cannot produce fifty purchases a week, so optimising for purchases leaves the ad set permanently learning limited. Optimising for a cheaper event generates enough volume for the algorithm to work with. Concentrating placements on one or two surfaces also stops a small budget being spread too thin to register.

Why does Meta say my campaign may get zero results?

Meta compares your daily budget against the expected cost of the conversion event you selected and estimates how many results that buys per day. When the budget divided by the expected cost per result rounds down to zero, it shows the warning and offers a higher figure. It is an arithmetic estimate, not a judgement on your ads.

Does Facebook really spend on placements I excluded?

It does, if you leave "Allow limited spending to excluded placements" ticked. Meta's own wording says it aims to spend around 5% of your budget on each excluded placement. Exclude four placements with that box ticked and roughly a fifth of your spend can land on surfaces you deliberately turned off. Untick it before you publish.

How long before Facebook ads start working?

Plan for at least a week of stable delivery before drawing conclusions, and longer if your event volume is low. The learning phase needs around fifty optimisation events to complete, and every change to budget, targeting or creative restarts it. The most common mistake is editing the campaign daily, which keeps it in learning indefinitely.

If you want to follow what this campaign actually produces, along with the rest of the marketing systems I run in public, join our community and you will get the numbers as they land rather than a tidy case study written after the fact.

---

Disclosure: Some links in this article are referral links. If you use one, I may earn a commission at no extra cost to you. I only link to tools I actually use.