4.87× profit ROAS (nanoSPACE)
50 conversions/week for the learning phase
4 phases from zero to scaling

October 2019. The nanoSPACE store was selling air purifiers and humidifiers with monthly revenue of around 3 million CZK (≈ €120,000). Founder Dávid Kopecký knew the product had potential — but he didn't know how to reach the people already searching for it, or the ones who weren't searching yet but needed it.

Today nanoSPACE generates over 150 million CZK (≈ €6M) a month and ranks among the largest e-shops in its segment in the Czech Republic and Slovakia. And Facebook advertising was one of the key engines of that growth.

But there are no secret tricks or magic strategies behind it. There's a system — four clear phases that we walked through, again and again, with every new client. And in this article, I'll show you the entire system.

A summary for those who don't have time to read the whole thing: Successful Facebook advertising for e-commerce rests on four phases — testing, optimization, scaling, and long-term strategy. Key takeaways: the learning phase needs 50 conversions per week; creative matters more than targeting today (Meta Andromeda); measure ROAS on profit, not revenue; the Conversions API (CAPI) is a must for accurate data; Advantage+ Shopping campaigns can dramatically cut the effort of scaling.

Why e-shops make money on Meta

Let's start with numbers. According to Statista, more than 2 billion people actively shop through social media. Meta's platforms — Facebook and Instagram — hold a dominant position in this space. And industry benchmarks say this: the average ROAS for e-shops on Facebook sits around 2.5–4×. But those are averages.

In specific segments, the picture looks different. Beauty and cosmetics average a ROAS of 2.31×. Home & Garden, like nanoSPACE? An average of 3.86×. And the best performers — the ones who work systematically — reach 6× and beyond.

Why does Meta work so well for e-shops? Because it can work with the entire purchase funnel. Google captures existing demand — people who already know what they want. Meta creates demand. It shows your product to people who didn't know about it yet — but the moment they see it, they want it. And then the whole funnel — from first touch to repeat purchase — lives on a single platform.

The Czech market? CPM (cost per thousand impressions) runs around 80–150 CZK (≈ €3–6) depending on the segment and season. CPC (cost per click) is typically 3–8 CZK (≈ €0.12–0.32). Those are still favorable prices compared to Western Europe — and the window is slowly closing.

Phase 0 — Before you spend a single koruna

Plenty of e-shops skip this phase and then wonder why their campaigns don't work. But successful Facebook advertising isn't built on the right targeting or the right creative — it's built on the right foundations. Without them, you're throwing money away.

Here's what needs to be working before you put your first euro into advertising:

What you need Why it's critical
Conversions API (CAPI) Without server-side tracking, you lose 20–40% of conversion data to blockers and iOS 14+. Meta can't see the results, and the algorithm optimizes blind.
A quality XML feed Dynamic product ads are the most effective format for e-shops. A bad feed = bad ads = higher acquisition costs.
Margins by product A 3× ROAS sounds great — but not if you sell at a 15% margin. You need to know at what ROAS you turn a profit and at what ROAS you don't.
A clear USP Your creative communicates your advantage. If you don't know why a customer should buy from you rather than a competitor, you don't know what to put in your ads either.

One number to remember: e-shops without CAPI set up properly typically measure 30% fewer conversions than they actually get. The algorithm then optimizes on bad data — and the results reflect it.

Phase 1 — Testing (2–4 weeks)

This is where the success of the whole strategy is decided. Phase one has a single goal: find out what works. Not to earn, not to scale — to test.

Start with broad targeting. No detailed interests, no lookalike audiences — just age and gender, if that's relevant to your product. Why? Because Meta Andromeda — Facebook's new ad system launched in 2024 — is now so sophisticated that creative is effectively your targeting. The algorithm analyzes who responds to the ad and finds more people like them on its own. The less you tie its hands with targeting, the better it works.

The theory behind this is called "creative as targeting" — and it's a fundamental shift from the 2020 approach, when you carefully assembled interest-based audiences. Today the most important question is: what story does the creative tell, and who does it attract?

What to test in phase one:

The key milestone of phase one: the learning phase. Facebook needs at least 50 conversions per week per ad set to exit the learning phase and start optimizing. Below that threshold there isn't enough data and results are random. If your budget doesn't allow for that, optimize for add-to-cart instead of purchase.

Phase 2 — Optimization (weeks 4–8)

The data is on the table. Now comes the work of using it — and this is where most advertisers make their first big mistake: they make too many changes too fast.

The rule of optimization: one change, one week. If you change the creative, the targeting, and the budget all at once, you'll never know what moved the results. And the learning phase resets with every major change.

What to do in the optimization phase:

Cut the weak creatives. Look at cost per purchase (or cost per add-to-cart during the learning phase). Turn off creatives performing well below average. Keep the ones that work running, and test variations of the winners.

Launch retargeting. People who visited the site but didn't buy — that's your warmest audience. Dynamic product ads show them exactly what they were browsing. Retargeting typically delivers 3–5× higher ROAS than prospecting campaigns.

Watch frequency. If the average frequency (how many times a single person has seen your ad) climbs above 3–4×, the creative is burned out. Results start to fall and costs start to rise. Time for new content.

TIP: The most common mistake in the optimization phase is rushing decisions. You see a bad day and immediately reach for changes? Stop. Facebook needs a full week of data — day-to-day swings are normal. Wait for a statistically relevant sample. Itchy-hands syndrome costs advertisers millions every year.

Phase 3 — Scaling (month 2+)

You've got winners. Creatives that work, an audience that responds, and a ROAS that turns a profit. Now you want more. And this is where the hardest part begins — because scaling on Facebook isn't linear.

Vertical scaling (increasing budget): The golden rule — a maximum of +20% budget at a time. Any more and the algorithm restarts the learning phase. Typical cycle: raise it 20%, wait 5–7 days, let the results stabilize, raise again. Slowly, but safely.

Horizontal scaling (duplicating ad sets): You copy a winning ad set and launch it as a separate campaign with the same or a higher budget. Each set builds its own optimization, which gives you more ad inventory without disrupting the existing campaign.

Advantage+ Shopping campaigns: In 2024 Meta significantly improved this automated format, where the algorithm decides everything itself — targeting, placements, creative combinations. For e-shops with enough conversion data and a quality feed, Advantage+ Shopping can substantially lower CPA while maintaining or increasing volume.

The critical mindset when scaling: forget ROAS as a percentage, focus on profit in real money. A 5× ROAS on 100,000 CZK (≈ €4,000) in revenue = 20,000 CZK (≈ €800) profit (at a 20% margin). A 3× ROAS on 500,000 CZK (≈ €20,000) in revenue = 30,000 CZK (≈ €1,200) profit. Is it clear now which campaign is better?

Phase 4 — Long-term strategy

The biggest threat to successful campaigns is called creative fatigue. An ad that converts brilliantly today will stop working in 6–8 weeks — simply because your target audience has seen it too many times. This isn't a problem; it's the reality of the platform.

The solution: a creative system, not ad-hoc production. Keep a pipeline of new creatives feeding into rotation. At least 2–3 new formats every month. Watch frequency, and when it crosses 3–4×, add new content before results drop.

Long-term strategy also has to address customer lifetime value (LTV). A customer who buys from you once has a different value than one who comes back regularly. If you know your average LTV, you can afford a higher CAC (cost of acquisition) — and scale more aggressively.

Connecting this to email marketing is essential at this point. Facebook brings the customer in — email keeps them. Automated flows after the first purchase, winback campaigns after 30/60/90 days without an order, cross-sell sequences — this is where a good e-shop becomes a great one.

nanoSPACE case study: From 3 to 150 million

Let's move from theory to practice. When we took over the nanoSPACE ad account, the situation was typical for a mid-sized e-shop: an account full of historically accumulated campaigns with no clear structure, tracking that missed a large share of conversions, and a frustrated owner who knew the platform should be working better.

Phase 0 took three weeks at nanoSPACE. We deployed CAPI via a server-side integration, cleaned up the XML feed (over 200 products with missing data removed), mapped margins by product category, and defined 4 key USPs for different customer segments.

Account restructuring: from 40+ campaigns down to 8 clearly defined ones. Each with a clear goal, a clear audience, a clear creative.

The result of the first restructuring was shocking — and at the same time typical: -25% ad spend while keeping the same revenue. Money was simply leaking into dead campaigns that no one had shut off.

After 6 months of systematic work, the results came in:

The most important takeaway from nanoSPACE: scaling into millions in revenue doesn't require magic tricks. It requires discipline — precise measurement, patience during the learning phase, systematic creative testing, and the courage not to cut a working campaign just because it personally bores you.

The 5 biggest mistakes e-shops make on Facebook

1. The blue button of doom (boost)

The "Boost post" button sitting right under an organic post is the fastest way to spend money for zero results. Boosted posts have no access to advanced audiences, conversion optimization, or pixel data. Always advertise through Ads Manager — no exceptions.

2. Itchy-hands syndrome

The campaign runs three days, the results aren't perfect — and you start changing targeting, cutting budget, turning off ads. The learning phase gets banned at the same time. The algorithm optimizes blind. Results get worse. You panic and make more changes. That's what a spiral you can't climb out of looks like.

3. Chasing ROAS instead of profit

ROAS is a ratio — and ratios seduce you into illusions. An e-shop with a 10% margin needs a 10× ROAS just to break even. An e-shop with a 40% margin can afford a 2.5× ROAS. Track gross profit after ad costs — MER (Marketing Efficiency Ratio) or blended ROAS. Not the number in Ads Manager.

4. Boring creatives

A creative that works today will stop in 2 months. Yet most e-shops leave the same ads running for months without refreshing them — and then wonder why results are sliding. Creative fatigue is the single biggest performance killer on the platform. Plan new production ahead of time.

5. Leaks on the website

A great ad brings people to the site — but the site doesn't close them. Slow loading (over 3 seconds = losing 50% of visitors), no mobile display, a complicated checkout, an untrustworthy look — these are leaks that advertising won't fix. Before scaling budget, always confirm that the site converts.

An AI tip for 2026

Meta Advantage+ Creative automatically tests different combinations of your creative elements and shows each user the variant they're most likely to respond to. In 2026 this feature is so advanced that ignoring it is a mistake.

Turn on Advantage+ Creative for all campaigns — but upload quality material. The algorithm is good at picking combinations, but it can't invent good creative for you.

And for writing ad copy? ChatGPT or Claude with this prompt generate consistently strong results:

"You are a Facebook ads copywriter. Write 5 variants of primary text for the e-shop [name] selling [product]. Target audience: [description]. Key USPs: [benefits]. Each variant uses a different angle — problem-solution, social proof, product benefit, urgency, curiosity. Each one has a maximum of 125 characters in the first sentence (the hook). Tone: friendly, no buzzwords."

Then review the output, adjust it to your tone of voice, and test. AI won't replace knowing your customer — but it dramatically speeds up generating variants to test.

Conclusion

Facebook advertising for e-commerce isn't a sprint — it's a marathon with systematic training. Results like nanoSPACE's didn't appear overnight. They came from consistently working through four phases: the right foundations, patient testing, disciplined optimization, and bold but controlled scaling.

Every phase has its rules. Every phase has the mistakes that can bury it. And every phase is manageable — if you know what you're doing and why.

One line to take away from this article: creative matters more than targeting today. Meta Andromeda will find the right people if you give it the right story. Focus on what you say — and the platform will figure out who to say it to.

Frequently asked questions about Facebook advertising for e-commerce

What's the minimum budget to start?

For meaningful testing, plan on at least 300–500 CZK per day (≈ €12–20). That gives you room to test 2–3 creatives and reach the learning phase (50 conversions/week). Below 200 CZK per day (≈ €8) the algorithm has nothing to learn from and results will be random. A smaller budget = slower data = slower decisions.

Do I need a page full of posts?

Not necessarily. A Facebook page serves primarily as the platform for launching ads — organic page reach is minimal today. What matters more is that the page looks trustworthy: a profile photo, a cover photo, complete contact details, and at least a few posts. Customers may check you out there.

What should I do when performance suddenly drops?

First check the technical side — is the Pixel firing, is conversion data coming through, is there a problem on the website? Then look at frequency — if it's above 3–4, the creative is burned out and you need new content. Only as a last resort should you consider changes to targeting or budget. 80% of performance drops are caused by creative fatigue or a technical issue, not bad targeting.

Is it worth targeting detailed interests?

In 2025+, less and less. Meta Andromeda — Facebook's new ad system — is so good at finding the right audience based on creative and conversion data that detailed interest targeting usually limits performance rather than helping it. Broad targeting (no restrictions, or just age/gender) combined with strong creative delivers better results over the long run.

How do I know whether poor sales are the ad's fault and not the website's?

Track add-to-cart rate and checkout abandonment rate alongside purchases themselves. If people click the ad (good CTR), add to cart, but don't buy — the problem is on the website (price, trust, payment methods, speed). If they don't even click the ad — the problem is in the creative or the targeting. Facebook Ads Manager shows you this data precisely.