Forget interest targeting. In 2026 your creative is the new targeting

Here's a number that should change the way you think about Facebook ads: creative variance explains 3–4× more of the difference in campaign performance than targeting structure does. In other words — what you show in the ad has three to four times more impact than who you show it to.

And yet most advertisers spend hours fine-tuning audiences and interest groups. And the creative? They "somehow put it together".

In our previous article on Meta Andromeda, we broke down how Meta's new AI brain works under the hood — Entity ID, the GEM model, the whole machinery. Today we'll look at the practical side: what all of it means for your targeting. What Meta gradually shut down, why it's doing it, and above all — what to do instead. Including the situations where broad targeting doesn't work (because it doesn't always work, and anyone who tells you otherwise is either lying or has a budget of 200,000 CZK (≈ €8,000) a month).

What's in this article:

  1. Why interest targeting is dying (and what killed it)
  2. How creative takes over the role of targeting
  3. A practical framework: creative targeting in practice
  4. When broad targeting doesn't work (and what to do instead)
  5. Retargeting in the Andromeda era
  6. Auditing your targeting — a 10-point checklist
  7. AI tips for creative targeting
  8. FAQ

Summary for those who don't have time to read the whole article

Why interest targeting is dying (and what killed it)

Timeline: what Meta gradually shut down

This wasn't one big change. Meta killed interest targeting gradually, piece by piece, so that as few advertisers as possible would notice:

See the pattern? Meta gradually took away your manual targeting tools and replaced them with automation. And the numbers prove it right.

Why Meta does it (and it has the data)

It's not that Meta is evil. It's that its AI is better at targeting than you are.

Meta's internal data shows that after detailed targeting exclusions were removed, the average cost per conversion fell by 22.6%. Read that again: less manual targeting = cheaper conversions.

Advantage+ Audience (Meta's automated targeting) achieves:

And Advantage+ Shopping campaigns? An average ROAS of 4.52× versus 3.70× for manual campaigns. That's 22% more.

Why? Because Andromeda sees signals you don't. Browsing history, purchase behavior, content interactions, time spent on pages. And it processes them for billions of users simultaneously, with 10,000× the capacity of the previous system.

Your manual setting of "women, 25–45, interested in fitness"? To Andromeda that's just a soft suggestion. The system takes a look, nods, and then does what it sees fit.

What it means for you in practice

Interests still exist in Ads Manager. You can fill them in. But their impact is a fraction of what it used to be. Your interest targeting is no longer a command — it's a suggestion the system may or may not take to heart.

The only thing you truly control is the creative. And that's exactly what the rest of this article is about.

How creative takes over the role of targeting

5 signals Andromeda reads from your creative

When you upload an ad, Andromeda doesn't just scan it for content moderation. It scans it to understand who to show it to. It looks for five key signals:

  1. Thumb Stop Rate — does the user stop scrolling? (Visual appeal, contrast, novelty)
  2. Hook Rate (first 3 seconds) — does it grab them? (Strength of the opening message)
  3. Hold Rate — how much of the video do they watch? (Content quality, relevance)
  4. Engagement signals — do they like, comment, share, save? (Resonance with the target audience)
  5. Conversion signals — do they buy? Fill in a form? (Commercial relevance)

Based on these signals, the system learns in real time what kind of people your ad "clicks with". And then it delivers it to more people like them. Regardless of the interests you set.

How different types of creative automatically reach different segments

This is the magic part. You don't have to tell the system who to show the ad to. You just give it the right ad and it finds those people itself.

A practical example — let's say you sell anti-allergy bedding:

Type of creative

Who it automatically reaches

UGC video from a mom: "My son finally sleeps through the night"

Moms of allergic children who are dealing with sleep quality

Infographic about dust mites with numbers

Analytically minded people who need data

Before/after video with a microscope

Visually oriented people looking for proof

Carousel "5 steps to better sleep"

People in the research phase looking for a how-to

Review from an allergist

People who trust authorities and experts

Five creatives = five different segments. Without a single setting in your targeting. Andromeda reads the content of each ad and delivers it to the people it's most relevant for.

Two people with the same demographics (both 35, Prague, male) can convert from completely different creatives. One responds to data and numbers, the other to an emotional story. In the past you couldn't influence that — both saw the same ad. Today Andromeda shows each of them the right one.

The 80/20 rule of advertising in 2026

Research shows that 80% of an ad's performance comes from the visual (the creative) and just 20% from the copy plus settings. And yet — in my experience — most advertisers invest 80% of their time in settings and 20% in the creative. Exactly backwards.

In 2026 that's suicidal. If you have to choose between an hour of tuning audiences and an hour of making a new creative, always pick the creative. The impact is 3–4× greater.

💡 **TIP:** Run a simple test. Open Ads Manager, look at the last 30 days and sort your ads by ROAS. I bet the gap between your best and worst creative is far bigger than the gap between your best and worst audience.

A practical framework: creative targeting in practice

3-Layer Creative Testing Model

When creative = targeting, you need a systematic approach to testing. Not "let's make 5 ads and see". This is the 3-layer model we use:

Layer 1 — Concept Testing (what you say)

Test fundamentally different angles:

Layer 2 — Hook Testing (how you say it)

On the winning concept, test different hooks:

Layer 3 — Format Testing (how it looks)

On the winning concept + hook, test formats:

Go step by step: first find the winning concept, then the winning hook, then the winning format. Not everything at once — otherwise you won't know what worked.

The 3×3 testing method

A simpler alternative for smaller budgets:

3 hooks × 3 formats × 3 copy angles = 27 combinations

In practice you don't have to test all 27. Pick the 9 most promising — three hooks in three formats. Run them for 5–7 days with an even budget. Evaluate on conversions (not likes!). Scale the winners.

How many creatives, how often, what hit rate to expect

Here come the uncomfortable numbers:

Sounds like madness? Partly, yes. But here's the reality: Andromeda "chews through" creatives faster than the old systems. A top performer lasts 2–4 weeks. Then creative fatigue sets in — CTR drops by 35–55%, CPC rises by 20–40%. And you need fresh blood.

German agencies put it bluntly: "Anyone testing fewer than 20 new assets a week loses touch with the market." For the Czech market, with smaller budgets, it'll be less, but the principle holds — you need a pipeline for producing creatives.

💡 **TIP:** You don't have to make everything from scratch. UGC videos shot on a phone cost a fraction of professional production and, on average, achieve a 30% lower CPC. You can assemble a carousel from existing photos. You can cut Reels from a longer video. A system for production > one-off creation.

When creative "dies" and how to spot it

Watch three metrics:

  1. CTR drops by 35–55% compared with the first days → the creative is losing its appeal
  2. CPC rises by 20–40% and frequency exceeds 2.5–3.0 → people are seeing it too often
  3. Conversion rate drops by 25–45% → the creative is no longer selling

When you see two of the three signals, it's time to swap the creative. Not the campaign — the creative. Add a new one to the ad set, pause the old one. Don't do it after 2 days — give it at least 7–10 days. But after 3–4 weeks even the best ad starts to fade.

When broad targeting doesn't work (and what to do instead)

This article would be unfair if I just told you "go broad and it'll be fine". It won't. Not always. Here are five situations where broad targeting fails:

1. New accounts with no conversion history

The Pixel has no data. Andromeda doesn't know who your customer is. When you say "find me buyers" and the system doesn't have a single conversion to learn from, it's searching blind.

What to do: Start with manual targeting (interests + a lookalike from your email database). Gather at least 50 conversions. Then switch to broad. It's a starting phase, not a permanent solution.

2. Small budgets (<50 conversions/week)

The algorithm needs data to learn. Below 50 conversions a week per ad set you won't get out of the learning phase. And in the learning phase, performance is unstable and expensive.

What to do: Reduce the number of ad sets. Instead of 5 ad sets at 500 CZK (≈ €20)/day, run 1–2 ad sets at 1,500 CZK (≈ €60)/day. A smaller structure with higher data density. See the small-budget framework from the Andromeda article.

3. Niche B2B products

Selling enterprise software for the pharmaceutical industry? Broad targeting will find you people who love to click — not the IT directors of pharma companies.

What to do: A Search-to-Social strategy. Capture high-intent users via Google Search (they're actively looking for a solution) and retarget them on Meta. Plus custom audiences from your CRM — upload your contacts' hashed emails. And creative as a filter: a video addressing the specific problem of IT directors → Andromeda shows it to exactly those people, with no interest settings.

4. Niche consumer products

Selling an ultra-specific product that 0.01% of the population needs? Broad targeting will be expensive, because the system has to search a huge database for a very small target group.

What to do: A combination — broad targeting with the creative as a filter (specific creative naturally filters out irrelevant people) plus custom audiences and lookalikes from your existing customers.

5. Missing CAPI and quality pixel data

Without the Conversions API (server-side tracking) you lose 30–50% of conversion data. Andromeda learns from incomplete information and optimizes for a distorted reality. An Event Match Quality below 6.0 out of 10 = a problem.

What to do: This is priority number one. Before you deal with targeting, fix your tracking. Pixel + CAPI + Event Match Quality above 6.0. Without it, any targeting — broad or manual — is like navigating with a broken compass.

Transition strategy: from manual to broad

If you've been targeting manually so far and want to move to broad, don't do it overnight. The proven approach:

  1. Weeks 1–2: Keep your existing campaigns running + launch a new one with broad targeting (CBO, 20% of the budget)
  2. Weeks 3–4: Compare performance. If broad holds a similar CPA/ROAS, raise it to 50%
  3. Weeks 5–6: If broad works, move most of the budget over. Keep manual running as a safety net
  4. Week 7+: Gradually wind down the manual campaigns if broad is clearly winning

The key: don't pair broad and manual campaigns on the same audience — they'll compete against each other in the auction.

Retargeting in the Andromeda era

Automatic retargeting inside broad campaigns

A surprising fact: Andromeda already retargets automatically. When you launch a broad campaign, the system itself recognizes users who have already interacted with your brand and prioritizes showing them the ad.

That doesn't mean separate retargeting is pointless. It means its role has changed.

When it still makes sense to run separate retargeting

Separate retargeting still makes sense in three situations:

1. Specific messaging for stages of the buying process

2. Different creatives for different stages

A broad campaign uses acquisition creatives (education, stories). A retargeting campaign uses conversion creatives (social proof, urgency, discounts). If you have both in one campaign, the system may mix them up.

3. Budget control

A retargeting audience is small and valuable. In a broad campaign the system may not give it priority. A separate campaign guarantees that retargeting gets exactly as much as you want (typically 10–20% of total spend).

First-party data — the new foundation of everything

With the end of cookies and restrictions on third-party tracking, first-party data (the data you collect yourself) is more valuable than ever.

CAPI (Conversions API) — server-side tracking, a must in 2026. Without it you lose 30–50% of conversion data and Andromeda optimizes on incomplete information.

Customer lists — your customers' hashed emails. Upload them to Meta Business Manager. The system uses them to train the AI on your actual buyers. The higher the quality of the list, the better.

Event Match Quality (EMQ) — a score of 0–10 that tells you how well Meta can connect a conversion event to a specific user. Target: 6.0 and above. Check it in Events Manager. Below 6.0 = your data is incomplete and optimization suffers.

💡 **TIP:** Lookalike audiences matter less in the Andromeda era than they used to — the system does the expansion automatically. But customer lists as seed data for training the AI matter more than ever. The more conversions you upload via CAPI, the better Andromeda understands who your customer is.

Auditing your targeting — a 10-point checklist

Open Ads Manager and go through these points. Give yourself a point for each "yes". Below 6 points, you have a problem.

  1. ☐ Do you have CAPI implemented? (Events Manager → Data Sources → check Server Events)
  2. ☐ Is your Event Match Quality above 6.0? (Events Manager → Overview → EMQ score)
  3. ☐ Have you uploaded a customer list? (Audiences → Custom Audiences → Customer List)
  4. ☐ Are your main campaigns running on broad targeting? (Location, age, language only — no interests)
  5. ☐ Do you have Advantage+ audience turned on? (Ad Set level → Audience → Advantage+ Audience)
  6. ☐ Do you have at least 8–15 conceptually distinct creatives per ad set? (Not variations — concepts)
  7. ☐ Are you adding new creatives at least once a week?
  8. ☐ Do your campaigns have enough budget for 50+ conversions/week/ad set?
  9. ☐ Are your retargeting audiences merged into a single ad set? (Not 5 ad sets at 200 CZK (≈ €8)/day)
  10. ☐ Are you leaving campaigns alone during the learning phase (the first 7 days)?

8–10 points: You're ready for 2026. Keep optimizing your creatives.

5–7 points: You have the foundations, but you're making unnecessary mistakes that cost you money.

0–4 points: Pause your campaigns and fix the fundamentals. Without them, any targeting optimization is pointless.

AI tips for creative targeting

When creative = targeting and you need dozens of them a month, there's no doing it without AI in 2026. Here are some concrete tips:

Generating concepts with ChatGPT/Claude

Enter this prompt (adapt it for your product):

"I run an e-shop selling [product]. My target audience includes [personas]. Generate 10 conceptually distinct angles for a Facebook ad. Each angle must reach a different segment or address a different problem. For each angle, suggest: 1) the hook (the first line), 2) the main message, 3) the format (UGC/static/carousel/video), 4) what type of person it's likely to reach."

Meta Advantage+ Creative

Turn it on in your ad settings. Meta automatically adjusts creatives — crops, music, text variants. Advertisers using Advantage+ creative achieve, on average, a 22% higher ROAS. You don't have full control over the final look, but the results speak for themselves.

Creative production at scale

💡 **TIP:** Build yourself a "creative pipeline" — a fixed process that produces 3–5 new creatives every week. Monday: brainstorm concepts. Tuesday: shoot/photograph. Wednesday: edit. Thursday: upload to Ads Manager. Friday: launch the tests. Without a system you'll always be behind.

Conclusion

Let's be honest: losing control over targeting is uncomfortable for most advertisers. For years we built audiences, tuned interests, tested lookalikes. And now Meta is telling us to "leave it to us"?

Yes. That's exactly what it's saying. And the data proves it right — at least for most cases.

The key takeaway from this article? Stop optimizing who and start optimizing what. Your creative decides who sees the ad, how they react to it and whether they buy. Targeting in Ads Manager is just a light seasoning in 2026 — the main course is the creative.

Three things to do this week:

  1. Go through the 10-point audit — how many points do you have? (Be honest with yourself.)
  2. Launch one broad campaign — alongside your existing manual ones. 20% of the budget, 7 days hands-off. Compare.
  3. Make 5 conceptually distinct creatives — a different format, a different angle, a different persona. Use the AI prompt above as inspiration.

And if you find you don't have the capacity to produce dozens of creatives a month — there's no shame in asking for help. At LK Media, this is what we do for clients every day. From strategy through creative production to campaign management. Because in 2026 it's not about who's best at clicking around Ads Manager. It's about who understands their customers best and can turn that into an ad that resonates.

Frequently asked questions about Facebook ads

Is broad targeting or interests better on Facebook in 2026?

For most advertisers with a sufficient budget (50+ conversions/week) and quality tracking (Pixel + CAPI), broad targeting is clearly better. The data shows a 22.6% lower cost per conversion without manual targeting. Exceptions: new accounts with no conversion history, niche B2B products and very small budgets — there, start with manual targeting and move to broad gradually.

How does Advantage+ audience work?

Advantage+ audience is Meta's automated targeting. Your inputs (interests, demographics) serve only as suggestions — the system automatically expands beyond your settings to people who are likely to convert. It achieves a 7–15% lower cost per result across campaign types. In 2026 it's the default recommended setting for most campaigns.

How many creatives do I need when creative = targeting?

It depends on your budget: smaller advertisers (up to 20,000 CZK / ≈ €800 a month) need a minimum of 8–15 conceptually distinct creatives and at least 3–5 new ones per week. Mid-size advertisers (20,000–50,000 CZK / ≈ €800–2,000) should have 40–50 new ads a month. Large advertisers 100+. The hit rate is 5–10% — most creatives won't make it, and that's normal.

How do I set up retargeting on Facebook in 2026?

Andromeda retargets automatically inside broad campaigns, but separate retargeting still makes sense for specific messaging (cart abandoners, urgency). Merge all your retargeting audiences into a single ad set (web visitors + cart + IG engagers). Budget: 10–20% of total spend. Creatives: social proof, reviews, limited offers. And above all — have CAPI; without it you lose 30–50% of your data.

What is a lookalike audience and does it still work in 2026?

A lookalike audience is a copy of your best customers — Meta finds people with similar traits. In 2026 they're less important than they used to be, because Andromeda does the expansion automatically inside broad campaigns. But customer lists (the source data for lookalikes) are actually more important — they serve as training data for the AI. Upload your customers' hashed emails to Meta Business Manager.

Meta title: Facebook Ads targeting 2026: creative is the new targeting

Meta description: Why interest targeting on Facebook is dying and what to do instead. A practical framework, a 10-point audit and AI tips for creative targeting.