If the last time you set up a Facebook ad was a year or two ago, brace yourself — today almost everything is different. In 2025, Meta rolled out a new AI system called Andromeda that fundamentally changed the way ads work. The old approach — dozens of small campaigns, hand-picked interests, 2–3 ads per set — now actively hurts performance.
How much has it changed? At nanoSPACE, an e-shop selling bedding and cosmetics, we rebuilt the campaign structure along the new rules. The result: a 25% cut in budget while keeping revenue flat. ROAS rose by 30% — from 3.73× to 4.87× on profit. Year over year, the e-shop's total revenue grew by 59%.
And it wasn't a change of creatives or products. It was a change of structure and approach — fewer campaigns, broader targeting, more creatives, and room for Andromeda.
In this article we'll show you what has changed, what your campaign structure should look like in 2026, and what you should definitely stop doing.
The short version, for those who don't have time to read the whole thing
- Andromeda changed the rules of the game. Meta's new AI system, with 10,000× more capacity, decides for itself who sees your ad. Creative = targeting. Hand-picking interests no longer makes sense.
- Fewer campaigns, more creatives. The recommended structure: 3–4 campaigns (testing, scaling, Advantage+, retargeting). 8–15+ distinctly different ads in each set. Meta recommends up to 150 ads per set.
- Advantage+ Shopping lowers CPA by up to 32%. A fully automated sales campaign where Meta runs everything. ROAS of 4.52× on average. For e-shops, it's a must.
- Static images still generate 60–70% of conversions. You don't have to go all-in on video. But you do need varied creatives — UGC, product photos, carousels, founder content.
- nanoSPACE case study: consolidating campaigns + separating products by margin + broad targeting = −25% budget, +30% ROAS, +59% total revenue year over year.
What Andromeda is and why it changes everything
Andromeda is the new AI brain of Meta's advertising system. Its global rollout was completed in October 2025. It sounds like a technical detail, but its impact on how you advertise is profound.
How did it work before? You manually chose who you wanted to show the ad to — women, 25–45, interested in fitness. The system then went looking for exactly those people.
How does it work with Andromeda? Three steps happen:
1. Andromeda reads your ad — image, video, text, sound. It understands what the ad is about.
2. Andromeda understands people — it knows what each user does, what they buy, what content they follow.
3. Andromeda matches — it connects the ad with the people most likely to respond to it.
What does that mean in practice? Four fundamental changes:
Creative = targeting. Instead of hours spent configuring interests, put your time into making varied ads. If you have a video about anti-allergy bedding, Andromeda will show it to people who care about allergies and sleep quality. On its own. More precisely than you ever could by hand.
Fewer campaigns, more creatives. Meta now recommends 8–15 or more ads in a single set (previously a max of 6). The limit is up to 150 ads per set. Andromeda needs a large pool of creatives to choose from.
Broad targeting is the foundation. Hand-picking interests constrains the algorithm. You're telling it: "Only look here." When it might have found great customers elsewhere. Advertisers who switched to the new approach report 20–35% higher ROAS.
Consolidation = better results. Instead of 10 small campaigns on small budgets, run 2–3 bigger ones. More data in one campaign = faster learning = smarter algorithm decisions.
What your campaign structure should look like in 2026
Forget about 10–15 campaigns. In the Andromeda era, you need 3–4:
Campaign | Purpose | Budget | Number of ads |
Testing (ABO) | Testing new creatives | 10–20% | 3–6 per set |
Scaling (CBO) | Scaling the winners | 40–50% | 15–30+ |
Advantage+ Shopping | Automated sales | 30–40% | 10–150 |
Retargeting | Winning back the hesitant | 5–10% | 6–12 |
Testing campaign (ABO)
Your lab. This is where you test new creatives. You use ABO (budget at the ad set level) so you control how much money goes into each test. Targeting: broad (your whole market, no interests). Let Andromeda find the right people on its own based on the creative.
Scaling campaign (CBO)
The money machine. This is where you move the winning creatives from testing. CBO (budget at the campaign level) — Meta decides where the money goes. 15–30+ ads per set. Broad targeting.
Advantage+ Shopping (ASC)
A fully automated sales campaign. Meta runs everything — targeting, bidding, placements, creatives. The data shows: CPA up to 32% lower, CTR 11–15% higher, average ROAS of 4.52×. For e-shops, ASC is a must in 2026.
The minimum: a budget of €100 a day, Conversions API (CAPI) set up, and a product catalog in Meta Commerce Manager.
Retargeting campaign
Winning back people who visited the site, added to cart, but didn't buy. Here you can target more narrowly — specific audiences (website visitors from the last 7/14/30 days, cart abandoners). 6–12 ads aimed at overcoming those final doubts.
What not to do in 2026
Discussions on Reddit and LinkedIn and agency case studies all point to the same conclusion: these practices are counterproductive:
- Lots of campaigns with the same objective and similar targeting. The campaigns cannibalize each other in the auction — competing for the same people and driving up their own CPM.
- 5–10 ad sets with different interests or lookalikes. In the Andromeda era, that's pointless. One set with broad targeting and good creatives beats ten sets with manual targeting.
- Fake variety in creatives. 20 ads that differ only in button color or a single word. Andromeda needs genuinely different creatives with different angles.
- Big budget jumps. +100% overnight = a restart of the learning phase. Raise it gradually, by no more than 20–30% every other day.
- Frequently switching campaigns on and off. Every shutdown resets the learning. The campaign then starts again from scratch.
- The "Boost post" button. It looks tempting, but you have no control over targeting, bidding, or format. Always promote through Ads Manager.
Creative: the most important thing in 2026
Andromeda decides who sees your ad based on its creative. That means the quality and variety of your creatives matter more today than anything else.
What works
- Static images — still generate 60–70% of conversions. Don't underestimate them.
- UGC (user-generated content) — authentic videos from customers outperform polished brand spots.
- Short vertical video (Reels 9:16) — 10–30% cheaper CPM than the Feed. Essential for reaching a younger audience.
- Carousel — the best conversion rate of any format. Great for showcasing a product range.
- Founder content — the founder on camera, authentic, shot on a phone. It builds trust.
How many creatives you need
For a mid-sized e-shop: 40–50 new ads a month. It sounds like a lot, but that includes variants — different copy on the same image, a different hook on the same video. Refresh your creatives when CTR drops below 1.5% or CPM jumps by 25%+. Typically after 5–7 days.
How to split your creative budget: 60% proven, 30% testing, 10% seasonal.
Case study: nanoSPACE — how restructuring cut the budget by 25%
nanoSPACE.cz is a Czech e-shop selling bedding for allergy sufferers, waterless cosmetics, and cleaning products. It operates on the Czech and Slovak markets on the Shoptet Enterprise platform.
The problem
The e-shop had lots of smaller campaigns, each living its own life. The budget spread out with no clear priority. Products with different margins (a €4 toothpaste vs. a €32 cream) sat in the same campaign — the algorithm favored the cheaper conversions. Andromeda didn't have enough consolidated data.
What we changed
1. We consolidated campaigns into two main CBO campaigns — one for cheaper products (cosmetics, cleaning), another for more expensive ones (bedding).
2. We separated products by margin, not by category — so the algorithm wouldn't prioritize cheap conversions.
3. We gave Andromeda room — broader targeting (broad CZ/SK), fewer constraints, more data in each campaign.
4. We identified hidden gems — creatives with high ROAS on a small budget (cellucream ROAS of 10.29×).
The results
Metric | January 2026 (before) | February 2026 (after) | Change |
Ad spend | 307,000 CZK (≈ €12,300) | 230,000 CZK (≈ €9,200) | −25% |
Revenue from ads | 1.15M CZK (≈ €46,000) | 1.12M CZK (≈ €45,000) | −2.6% |
ROAS (on profit) | 3.73× | 4.87× | +30% |
Purchases from ads | — | 600 | — |
On a 25% smaller budget we kept revenue almost unchanged. Efficiency rose by 30%. Year over year (February 2025 vs. February 2026), the e-shop's total revenue grew by 59% and its total margin by 50%.
Key takeaways
- Structure matters more than budget. Most e-shops look for salvation in a bigger budget. Often, all it takes is restructuring the campaigns.
- Andromeda needs room. Fewer campaigns with consolidated data = better decisions.
- Separate products by margin, not by category. Otherwise the algorithm will favor cheap conversions.
- Look for hidden gems. Creatives with high ROAS on a small budget are candidates for scaling.
How Meta Ads and Google Ads work together
A common question: "Should I put my money into Meta or into Google?" The right answer: both. Because they do different things:
Meta Ads | Google Ads | |
**Role** | Creates demand (discovery) | Captures demand (intent) |
**How it works** | Shows ads to people who match a customer profile | Shows ads to people who are actively searching |
**CPC** | About €1.00–1.60 | About €0.20–1.00 (Search) |
**Strength** | Visual storytelling, brand building, retargeting | High purchase intent, Shopping |
Meta warms up the audience — people see your product and remember it. Google then catches them — when those same people search for your product or category. The best e-shops run both at once.
To start: if you have a limited budget, begin with Google Ads (Search + Shopping) — that's where the direct purchase intent is. Add Meta once Google is running profitably and you want to reach new people.
Conclusion
Facebook advertising in 2026 is fundamentally different from two years ago. Andromeda has taken over control of who sees your ad. Your job is no longer to configure interests and demographics — your job is to create great creatives and give the algorithm room.
If your account has 10+ campaigns with the same objective, hand-picked audiences, and 3 ads per set — it's time for a change. Consolidate, broaden targeting, add creatives. The nanoSPACE results show it works.
Want help restructuring your Meta Ads? Email info@lkmedia.cz — we'll take a look at your account and propose a structure that makes sense for your e-shop.
Frequently asked questions about Facebook advertising
How much does Facebook advertising cost in 2026?
In Europe, CPM (cost per 1,000 impressions) ranges between €6 and €12. CPC (cost per click) is typically €0.60–1.40 for e-commerce. Average CPA depends on the industry — for e-shops it runs between €20 and €60. For comparison, US CPM is around €23, so the European market is significantly cheaper.
How much should I invest in Meta Ads?
The minimum for Advantage+ Shopping: €100 a day. For a testing campaign: €20–40 a day per ad set. Overall, we recommend starting at €120–200 a day (€3,600–6,000 a month) for an e-shop that wants meaningful results. Below €60 a day, the algorithm doesn't have enough data to learn.
Is broad targeting or interests better?
In 2026, broad targeting usually wins — Andromeda finds the right people on its own based on your creative. Interest targeting makes sense for brand-new accounts with no data, for very specific niche products, or on small budgets. If you have a conversion history and CAPI set up, try broad.
How many creatives do I need?
For a mid-sized e-shop: 40–50 new ads a month. Keep 8–15+ distinctly different creatives in each ad set. Refresh them when CTR drops below 1.5% or CPM jumps by more than 25%. A creative typically "burns out" in 5–7 days.
Should I use Advantage+ Shopping?
If you have an e-shop with a product catalog in Meta Commerce Manager and CAPI set up — absolutely yes. Advantage+ Shopping delivers an average ROAS of 4.52× and lowers CPA by up to 32% compared with traditional campaigns. It's the most effective format for e-commerce on Meta in 2026.
