There is no single right budget for Meta Ads. There is, however, a wrong way to choose one: picking a round number that feels comfortable, splitting it across six campaigns, and declaring after ten days that "Facebook doesn't work".
A budget has one job at the start: to buy you enough data to make a decision. If it is too small or spread too thin, you pay for ads and learn almost nothing.
This guide gives you a simple way to calculate a starting budget from your own numbers, decide how many ad sets it can support, and know when to spend more.
Start with what you can lose
Your first month on Meta is a test. Some of that money will go on ads that do not work, and that is part of what you are paying for.
Before you do any maths, answer one question: how much can you spend over the next 30 days and still run the business if you get nothing back? If the answer is close to zero, fix your finances first, or start with organic content and direct sales until you have room. Don't fund a test with money you need for stock or rent.
Work out your budget from your numbers
You need three figures. If you don't have them, read our guide to a good ROAS for your store first.
Average order value: what a typical order brings in.
Target cost per purchase: the most you can pay for a purchase and still reach your goal.
Number of purchases you need to judge a test.
Meta's own guidance offers a useful reference. In an October 2026 post reported by PPC Land, Meta said ad sets should aim for roughly 50 conversions a week, and that in its analysis those below that level ran about 28% more expensive per purchase. Meta did not publish its method, and smaller budgets may explain part of that gap, so treat it as direction rather than a law.
The same post gives a formula for the number of ad sets your budget can support:
Meta's formula Maximum ad sets = Weekly budget ÷ (Cost per result × 50) |
Here is what that means for a store, using an illustrative EGP 150 target cost per purchase:
Weekly budget | Cost for 50 purchases | Ad sets it can support |
|---|---|---|
EGP 7,500 | EGP 7,500 | 1 |
EGP 15,000 | EGP 7,500 | 2 |
EGP 30,000 | EGP 7,500 | 4 |
Many new stores can't reach 50 purchases a week at first. That doesn't mean you can't advertise. It means you should run fewer ad sets, so your budget and your data sit in one place rather than scattered across many.
What if your budget is smaller than that?
Most small stores will start below this level. You can still run a useful test if you follow three rules.
Use one campaign and one or two ad sets. Don't divide the budget across audiences you can't afford to test separately.
Give it enough time. Judge results over a couple of weeks rather than day by day, because a few days of data can swing widely.
Judge with an earlier signal if necessary. If purchases are too few to compare ads, look at add-to-cart rate, cost per checkout and click-through rate. These tell you which ads and pages are worth more money, though only purchases tell you whether they make money.
A low budget also makes your order value more important. If an order leaves you very little after costs, consider a bundle or a higher-value product before you advertise.
How to split your first budget
A practical starting split for a new store looks like this. Treat it as a starting point rather than a rule.
Part of the budget | Share | Purpose |
|---|---|---|
Testing new audiences and creative | About 80% | Finding people and ads that sell |
Retargeting store visitors | About 20% | Bringing back people who showed interest |
If your store has very little traffic, retargeting audiences will be small, so spending may be limited at first. Let that part grow naturally, rather than forcing it. Once you have winners, keep some budget for testing, because ads wear out.
If you are unsure how to structure your campaigns, our guide to CBO vs ABO explains how to decide whether Meta or you should control spending for each group of ads.
Spend on creative as well
Your budget isn't only what you pay Meta. If you can only afford one video, you can only test one idea. Set aside time or money for a few different concepts before launch:
A short video showing the product in use
A customer talking about it, if you have a real customer willing to do so
A clear static image with the offer
A video that answers the most common objection
For our Max Muscle campaign, we ran 8 creative concepts in 48 hours and rebuilt the offer around perceived value. Across seven days, US$2,108 brought 597 purchases at US$3.5 each. That was a Black Friday flash sale for an established retailer in Egypt, so a new store shouldn't expect those numbers. The lesson is the volume of ideas tested and the strength of the offer, not the ROAS.
Check that the money can buy a sale
Before launch, make sure a bigger budget won't simply waste more:
Your Meta pixel and Conversions API are firing correctly, and purchases are recorded with the right value.
Your product page works well on a phone, with the full price and delivery fee clearly shown.
Your confirmation process is ready, because orders will start arriving immediately.
Your stock can handle what the test might sell.
If you sell through cash on delivery, the budget should be judged on delivered orders. See our guide on how to reduce cash on delivery returns.
How to read the first results
Avoid making changes every day. Each significant edit can reset what the system is learning, and short-term swings are normal. A simple approach:
What you see after enough spend | What it probably means | What to do |
|---|---|---|
Few clicks, low click-through rate | The ad isn't getting attention | Test new creative and new hooks |
Many clicks, few add-to-carts | The page or product isn't convincing | Fix the page, price or offer |
Many add-to-carts, few purchases | Checkout, shipping or price surprise | Simplify checkout, show the full cost |
Good purchases, poor delivery | Weak orders, slow confirmation or courier | Fix confirmation, adjust the campaign |
When to increase the budget
Scale when you have evidence, not when you have a good day. A reasonable checklist:
The campaign has beaten your target cost per delivered order for at least one to two weeks.
It has done so on enough orders that a few sales would not change the picture.
You can fulfill more orders without damaging delivery.
You have new creative ready, since more spend wears out ads more quickly.
Raise the budget gradually, then watch for a change in cost. Results usually weaken as spend grows, because you are reaching people who are less ready to buy. As long as you stay above your profit target, that's acceptable.
When to stop or change
Stop or rethink if any of these is true:
You have spent several times your target cost per purchase on an ad and it has brought no sales.
Your numbers show a profit on the dashboard but not after delivery.
You have tested several genuinely different ideas and none comes close to your target. At that point, the problem may be the product, the price or the market rather than the ad.
A sample first month
Week 1: Get ready Calculate your target. Produce at least three different ideas. Check tracking, the mobile store and order confirmation. |
Weeks 2–3: Test Run one campaign with a limited number of ad sets. Avoid daily edits. Stop only the ads that have clearly failed. |
Week 4: Decide Compare ads by delivered orders. Keep the winners, make new versions, and decide whether to raise the budget, fix the page or test a new offer. |
Getting the most from your budget
The right budget is the one that gets you to a clear decision without risking the business. Start with a figure you can afford to lose, focus it on few ad sets, and judge it on delivered orders. If you would like a team to manage that process, we offer Meta Ads for e-commerce. And if you are building your whole plan, start with how to market an online store in Egypt.
Launching a project this quarter?
30 minutes, no cost. We’ll show you where your cost per lead is leaking.
Sources
PPC Land, reporting Meta's Peak Season 2026 post: https://ppc.land/meta-says-ad-sets-under-50-weekly-conversions-cost-28-more-per-purchase/
Last reviewed November 2026.
