TikTok Ads Targeting: Finding Your Audience on the Most Chaotic Platform

The Reality of TikTok Targeting (It's Not Meta)

Let me save you some frustration: if you're coming from Facebook or Google Ads expecting the same level of targeting precision, you're going to be disappointed. TikTok's targeting is like Meta's weird younger cousin — it's got potential, but it's still figuring things out.

I spent six months testing TikTok audiences for an e-commerce client, and here's what I learned: the platform's strength isn't targeting, it's distribution. TikTok's algorithm is scary good at finding people who'll engage with your content, even if the interest categories feel like they were written by someone who's never run an ad before.

That said, you can't just throw spaghetti at the wall. You need a strategy. Here's what actually works.

Real talk: I wasted $4,200 trying to replicate our Facebook targeting strategy on TikTok. It didn't work. TikTok plays by different rules.

Interest Targeting: Good Luck Finding Your Niche

TikTok's interest categories are... rough. Where Facebook gives you thousands of hyper-specific interests (remember when you could target "people who like organic dog food"?), TikTok gives you broad buckets like "Pets" or "Food & Drink."

Here's what you're working with:

The interest categories are getting better — TikTok added some useful ones like "Small Business Owners" and "Online Shopping Enthusiasts" in 2024 — but they're still way behind Meta.

What actually works:

• Stack 2-3 related interests instead of going narrow. Instead of just "Fitness," try "Fitness + Health & Wellness + Weight Loss" to give the algorithm room to find patterns.
• Test interests in separate ad groups. Don't lump everything together. Run "Skincare" separately from "Beauty Tools" so you can see what's driving results.
• Watch your frequency. TikTok's audiences are smaller than Facebook's. If you target too narrow, you'll burn through your audience in a week and your CPMs will skyrocket.

In one campaign, we tested "Fashion & Accessories" (broad) vs. "Streetwear" (narrow). The narrow audience had a 40% higher CPA because we exhausted it in 9 days. Broad won by a mile once the algorithm learned.

Custom Audiences That Actually Work

This is where TikTok gets interesting. Custom audiences aren't as robust as Facebook's, but they're surprisingly effective if you build them right.

Here's what you can create:

Website visitors (via TikTok Pixel) — Same as Facebook. Track page views, add-to-carts, purchases, etc.

App activity (if you have an app) — Track installs, in-app events, purchases.

Engagement audiences — People who watched your videos, liked posts, visited your profile.

Customer lists — Upload emails or phone numbers (match rates are around 30-50%, worse than Facebook).

The pixel is your best friend here. If you're not tracking events properly, you're flying blind.

What I recommend:

• Build a 30-day website visitor audience ASAP. Even if it's only 2,000 people, you can retarget them with different creative and see what sticks.
• Create separate audiences for high-intent actions. Don't lump "page viewers" with "add-to-carts." Split them so you can bid more aggressively on warm traffic.
• Use video engagement audiences for creative testing. If someone watched 75% of your video but didn't click, they're interested. Hit them again with a different CTA.

I ran a retargeting campaign for a SaaS client targeting people who visited the pricing page but didn't sign up. We used a simple "Still thinking about it?" video with a 20% discount code. CPA dropped 58% compared to cold traffic.

Want to see which audiences are driving real ROI? Start with AdsMAA's free audit — it'll show you exactly where your TikTok budget is going.

Lookalike Audiences: When to Use Them

Lookalike audiences on TikTok are hit or miss. They work, but you need patience and decent source data.

Here's the deal:

• TikTok says you only need 100 people in your source audience. That's technically true, but it's also useless. I've never seen a lookalike perform well with less than 1,000 people.
• 1% lookalikes work best. Don't get fancy with 5% or 10% unless you're working with a massive budget. Stick to 1% and let the algorithm do its thing.
• Quality of source matters more than size. A 1,000-person purchaser list will outperform a 10,000-person page visitor list every single time.

Best source audiences for lookalikes (ranked):

Purchasers (last 30-60 days) — This is gold. If you have it, use it.

High-value engagers — People who watched 95%+ of your video AND clicked through.

Add-to-carts — Decent intent signal if you don't have enough purchasers yet.

Website visitors — Broad, but works for top-of-funnel awareness campaigns.

I tested lookalikes for a D2C beauty brand. We built three versions:

• 1% lookalike of 800 purchasers → $32 CPA
• 1% lookalike of 5,000 video viewers → $61 CPA
• 1% lookalike of 2,500 website visitors → $54 CPA

The purchaser lookalike crushed it, but it took 8 days to start delivering consistently. TikTok's algorithm needs time to learn. Don't panic and turn it off after 48 hours.

What Actually Works (Real Campaign Data)

Here's what I've learned from running 40+ TikTok campaigns across e-comm, SaaS, and local businesses:

Targeting strategies that consistently perform: My go-to testing framework:

Week 1: Run auto-targeting with 3-4 different creatives. See what the algorithm likes. Budget: $50-$100/day.

Week 2: Take the winning creative and test it across 3-4 interest categories (separate ad groups). Budget: $30-$50/day per group.

Week 3: Build custom audiences from pixel data (visitors, engagers, add-to-carts). Retarget with new creative. Budget: $40-$80/day.

Week 4: If you have 1,000+ purchasers or high-intent actions, build a 1% lookalike and scale. Budget: $100-$200/day.

One campaign that killed it: We ran ads for a productivity app. Started with auto-targeting, saw strong engagement from 25-34-year-olds interested in "Entrepreneurship." We narrowed to that interest, layered in a lookalike of app installers, and CPA dropped from $48 to $29 over three weeks.

Things that don't work (trust me, I tried):

• Hyper-narrow interest targeting (you'll run out of people too fast)
• Customer list uploads under 1,000 emails (match rates are terrible)
• Lookalikes based on video views alone (engagement doesn't equal intent)
• Over-layering audiences (e.g., Interest + Lookalike + Custom = dead delivery)

The Honest Bottom Line

TikTok targeting isn't as sophisticated as Meta. It won't be for a while. But if you lean into what TikTok does well — algorithmic distribution, creative-first delivery, and engagement-based learning — you can find your audience and drive real results.

Stop trying to recreate your Facebook campaigns. Start thinking like a TikTok advertiser: test fast, trust the algorithm (a little), and let creative do the heavy lifting.

I've seen TikTok CPAs beat Facebook by 30-40% when the targeting strategy matches the platform's strengths. It's possible. You just have to stop fighting the chaos and work with it.

Tools like AdsMAA can help you track which audiences are actually converting across platforms, so you're not guessing based on TikTok's dashboard alone. The data layer matters more than the targeting options.

Now go test some audiences. And for the love of god, don't try to target "people who like avocado toast and also own a French Bulldog." TikTok doesn't work like that.