A client comes in and says: "We'll put 50,000 CZK (≈ €2,000) a month into Facebook." Fine. But how do you split those 50,000? How much on testing new creatives, how much on scaling what works, and how much on retargeting? Without a clear answer to that question, it's just money thrown into a game of chance.
90% of the e-shops that come to us have the same problem — the entire budget goes into a single campaign with no system whatsoever. Either they test too much, or they don't test at all and ride one creative over and over until it starts falling apart. Both are wrong.
At LK Media we manage Facebook Ads campaigns for e-shops and agencies from 10,000 CZK (≈ €400) to 300,000 CZK (≈ €12,000) a month. And over the years we've developed an exact formula — not a gut-feel one, but a mathematically grounded one. Here it is in full.
Summary for those who don't have time to read the whole article: The ideal Facebook Ads budget split is 70% scaling, 20% testing, 10% retargeting. Testing campaigns need to complete the learning phase (50 conversions a week). Scale up by max 20% every 3–4 days. Creative fatigue comes sooner than you think — watch your frequency. And always calculate your break-even ROAS first; without it you don't even know whether you're running a business.
Why Most Companies Budget Wrong
According to Triple Whale data, on average 68% of the entire marketing budget goes to Meta (Facebook + Instagram). That in itself isn't a problem — Meta is still one of the most effective platforms for e-commerce. The problem is how that money is handled internally.
The most common mistakes we see:
Testing without data. You launch a new creative, give it 200 CZK (≈ €8) a day for 3 days, the result looks bad, you turn it off. But 200 CZK over 3 days isn't nearly enough for the algorithm — it has no data to learn from. The result is statistically worthless.
Scaling too fast. A working campaign gets four times the budget overnight. Meta treats it as a new campaign, resets the learning phase and the results collapse. The advertiser thinks scaling doesn't work — when in fact they just broke the 20% rule.
Ignoring break-even ROAS. A campaign generates a ROAS of 3.2. Looks good. But if the product margin is 25%, break-even ROAS is 4.0 — so the campaign is losing money. We see this in roughly every third new client.
Retargeting as the main strategy. Retargeting has great ROAS numbers — but only because it reaches people who would have bought anyway. Retargeting doesn't drive growth. Overloading a retargeting audience saturates it and performance collapses.
First: Calculate Your Break-Even ROAS
Before you start thinking about any budget split, you need to know one number: break-even ROAS. Without it, you don't know whether your campaigns are making or losing money.
The formula is simple:
Break-even ROAS = 1 ÷ gross margin
Examples by margin level:
| Gross margin | Break-even ROAS | Example product |
|---|---|---|
| 20% | 5.0 | electronics, pet food |
| 30% | 3.33 | sports equipment, drugstore goods |
| 40% | 2.5 | fashion, dietary supplements |
| 50% | 2.0 | cosmetics, handmade products |
| 60% | 1.67 | software, digital products |
TIP: Calculate with your margin excluding VAT. If your VAT is 21%, your real margin is lower than you think — and your break-even ROAS higher. It depends on whether you sell B2C or B2B.
The Testing Budget: How Much the Learning Phase Costs
The Meta algorithm needs data to learn how to target correctly. Specifically, it needs 50 conversions a week at the ad set level — only after reaching that number does it leave the learning phase and start truly optimizing. If you don't hit that number, the algorithm optimizes blind.
The formula for the testing budget:
Weekly testing budget = average CPA × 50
Example: Your average cost per order is 400 CZK (≈ €16) → the testing campaign needs at least 20,000 CZK (≈ €800) a week (or roughly 2,850 CZK (≈ €115) a day). If you don't know this, your testing campaigns will never step out of the shadow of vague results.
For creative testing, Foxwell Digital recommends these volumes based on your monthly spend level:
| Monthly spend | New creatives/month | Active creatives |
|---|---|---|
| up to 50,000 CZK (≈ €2,000) | 4–8 | 2–3 best |
| 50,000–150,000 CZK (≈ €2,000–6,000) | 8–16 | 3–5 best |
| 150,000 CZK+ (≈ €6,000+) | 16–30+ | 5–8 best |
One important thing about targeting: in the testing phase, use broad targeting (no detailed targeting). Meta's internal data shows that broad targeting has a 113% better ROAS than lookalike audiences. Your job in the testing phase is the creative — let the algorithm find the audience itself.
The Scaling Budget: The 20% Rule and Advantage+
Is your campaign running above break-even ROAS? Great — now you want to scale it. But how fast and by how much?
The golden rule we never break: increase by no more than 20% every 3–4 days. The reason is algorithmic — Meta reads a bigger jump as a major change, resets the learning phase and performance drops. Slow, gradual increases let the algorithm keep optimizing without interruption.
As for the structure of scaling campaigns: Advantage+ Shopping Campaigns (ASC) consistently outperform manually structured campaigns. Meta's internal data shows an average ROAS of 4.52 for ASC versus 3.70 for manual campaigns — a difference of 22%. ASC manages targeting, placements and creative rotation on its own, so all you have to do is upload your best creatives and set a daily budget.
The right account structure for scaling:
1 ASC campaign for your main product or category — 70% of the budget goes here. The Meta algorithm itself distributes spend between cold audiences and retargeting based on performance.
Separate branded and non-branded — if you sell under your own brand, put branded traffic into a separate campaign. It has a different CPA, a different conversion path and a different audience.
No microscopic ad sets — every ad set needs those 50 conversions/week to optimize effectively. Better to have fewer ad sets with a higher budget than dozens of small campaigns where none reaches the learning phase.
The Magic 70/20/10 Ratio — the Exact Table
This ratio isn't made up. It comes from analyzing hundreds of Meta accounts across various agencies (Foxwell Digital, Triple Whale, Northbeam). And it works across industries and budget sizes.
| Campaign type | Budget share | Purpose |
|---|---|---|
| Scaling | 70–80% | Scaling proven winning creatives to cold audiences. This is where the money is made. |
| Testing | 10–20% | Finding new winners — new creatives, formats, hooks, product angles. The pipeline for future scaling. |
| Retargeting | max 10–20% | Closing customers who are hesitating. Great ROAS, but a small audience — it mustn't get too large a share of the budget. |
Why retargeting at 20% maximum? Meta data shows that retargeting campaigns achieve an average ROAS of 3.61 — which sounds great. But this audience is small. If you pump too much money into retargeting, the display frequency quickly shoots up (over 7–10× in 30 days), the audience gets saturated, CTR drops and performance collapses. People start hiding you from their feed or blocking you outright.
Budget by E-shop Stage — Concrete Numbers
The 70/20/10 ratio applies universally, but the absolute numbers differ depending on the stage the e-shop is in. Here are ballpark budgets for the typical stages:
| Stage | Monthly budget | Strategy |
|---|---|---|
| Start-up | 10,000 CZK (≈ €400) | 100% testing. The goal isn't profit, but data. Find 2–3 creatives that work above break-even ROAS. |
| Ramp-up | 30,000 CZK (≈ €1,200) | 60% scaling of proven creatives, 30% testing, 10% retargeting. You're building a core library of winners. |
| Growth | 100,000 CZK (≈ €4,000) | 70% scaling (ASC), 20% testing (new formats, videos, UGC), 10% retargeting. Scaling rule: 20% every 3–4 days. |
| Scale-up | 300,000 CZK+ (≈ €12,000+) | 75% scaling (multiple ASC campaigns), 15% testing (a systematic creative sprint), 10% retargeting + upsell. |
Seasonal Budgeting — 2026 Calendar
CPM (cost per thousand impressions) on Facebook changes dramatically over the year. If you budget the same, linearly, all year round, you needlessly overpay in the expensive months and miss out on the cheap opportunities.
| Period | CPM level | Recommended strategy |
|---|---|---|
| Q1 (January–March) | Low | The cheapest advertising of the year. The ideal time to test new creatives and build audiences at low cost. |
| Q2 (April–June) | Medium | You start scaling the winners from Q1. Spring brings increased shopping activity, especially Easter and May. |
| Q3 (July–September) | Lower | Summer — lower CPM, but lower shopping activity too. A good time to test. September: the pre-season starts, raise the budget. |
| Q4 (October–December) | Highest | The most expensive advertising of the year. Don't use Q4 for testing — only proven winners. Max spend in November/December. |
TIP for November: Black Friday / Cyber Monday is the most expensive week of the year — CPM can be 3–5× higher than average. If you don't have strong creative and a proven audience, it's better to step back and let the big players overpay in the auction. In November, small e-shops often lose more than they make.
5 Budgeting Mistakes That Cost You Money
1. You mess with it every other day. Every campaign change — even a budget change of a few percent — restarts the learning phase. Optimal editing frequency: once every 3–4 days for scaling campaigns, once a week for testing ones. Discipline is the foundation of performance.
2. You scale from zero to a hundred. Doubling or tripling the budget overnight is a classic mistake. Meta treats it as a new campaign. The +20% every 3–4 days rule exists for a good reason. If you want to go faster, duplicate the campaign with a higher budget alongside the existing one.
3. You ignore break-even ROAS. Campaigns with a ROAS of 2.8 can be either very profitable (50% margin) or a total flop (20% margin). Without the break-even number you know nothing. Calculate it before you even launch your first campaign.
4. Zero money on testing. "One creative works for us, so why test?" — because creative fatigue will come. The average lifespan of a successful creative on Facebook is 4–8 weeks. If you don't have a pipeline of new creatives, one day you'll wake up with a collapsing ROAS and nothing ready to replace it.
5. Over-segmentation. 15 different ad sets for 15 different audiences, each at 500 CZK (≈ €20) a day. None reaches the learning phase. None learns to optimize. The result: an average of averages. Better 3 ad sets at 2,500 CZK (≈ €100) than 15 ad sets at 500 CZK (≈ €20).
An AI Tip for 2026: Let It Do the Math for You
If you want to quickly calculate your break-even ROAS and set the optimal budget split for your specific case, use this prompt in ChatGPT or Claude:
"I run an e-shop with products at an average price of [X CZK]. My gross margin is [Y%]. My average cost per order (CPA) from Facebook is [Z CZK]. My monthly budget for Meta advertising is [B CZK]. Calculate for me: (1) break-even ROAS, (2) the minimum daily budget for testing campaigns, (3) the split of the total budget by the 70/20/10 ratio, (4) the maximum possible increase per scaling step. Present the results as a clear table."
You'll get concrete numbers for your business — not generic recommendations. That's exactly what you need before every strategic discussion about your Facebook Ads budget.
Conclusion: Math Protects You from Emotions
Facebook Ads is an emotionally demanding platform. Results fluctuate from day to day. It's easy to panic after a bad week and start fiddling with your campaigns. Or, on the flip side, to lose your head over a good ROAS and overdo the scaling.
Math protects you from these reactions. If you know your break-even ROAS, you know when you have to act. If you know the 20% rule, you know how fast you're allowed to scale. If you have the 70/20/10 ratio, you know how to allocate your budget without overthinking it.
A system doesn't allow for bad decisions. And that's exactly what you want from an ad budget.
Frequently asked questions about the Facebook Ads budget
What should I do after raising my budget if performance drops?
After every budget increase the campaign goes back into the learning phase, which lasts 7–14 days. During it, performance may fluctuate. Don't react with further changes — let the algorithm stabilize. If after 14 days performance still doesn't reach break-even ROAS, check your creative and targeting, not the budget.
Daily or lifetime budget for Facebook Ads?
For testing campaigns we recommend a daily budget — you control your spend better and can more easily stop creatives that aren't working. For scaling campaigns, switch to a lifetime budget — Facebook then distributes it better according to performance over the week and takes advantage of seasonal opportunities (weekends, evenings).
Can I test with less than 500 CZK (≈ €20) a day?
Technically yes, but the results will be unreliable. The Meta algorithm needs enough data to learn correctly. With a daily budget below 500 CZK (≈ €20) the learning phase will take longer (or never complete at all) and the numbers will be distorted. If you have a limited budget, test fewer creatives at once — better 2 creatives at 500 CZK (≈ €20) than 10 creatives at 100 CZK (≈ €4).
When should I turn off a losing campaign for good?
Turn a campaign off if: (1) after completing the learning phase (50 conversions) the ROAS is still below break-even, (2) spend has exceeded 3× your average CPA without a single conversion, or (3) CPM is rising by more than 40% while CTR falls — a sign of audience saturation. Before turning it off completely, first try swapping the creative.
Why is retargeting capped at a max of 20% of the budget?
A retargeting audience is inherently limited — it's people who have already visited your site. If you put too much money into retargeting, you saturate this small audience, the display frequency shoots up and performance drops. Meta data shows that the optimal frequency for retargeting is 3–7× in 30 days. Above 20% of the total budget you'll almost always exceed it.