You ran Meta ads for a year. You found that a 1% Lookalike off your purchaser list was your strongest cold audience, every time. So when you opened a TikTok account, you cloned the move — built a Lookalike, set it live, expected the same magic. The cost per result came back ugly. Here's the thing: the audience isn't broken. TikTok just doesn't lean on Lookalikes the way Meta does, and treating the two as interchangeable is exactly why your first TikTok campaign underperforms.
How TikTok Lookalikes actually work
A TikTok Lookalike Audience takes a source — your customer list, Pixel events, or people who engaged with your TikTok content — and finds users who behave similarly. On paper, that's the same idea as Meta.
The difference is how hard the platform leans on it. Meta delivery is audience-led: you tell it who, it goes and finds them. TikTok delivery is content-led. The algorithm reads your video, watches who finishes it, and goes hunting based on that. Your audience is a starting hint, not a fence.
TikTok gives you three expansion sizes: Narrow (roughly the closest 1–3% of the country's user base), Balanced, and Broad. Narrow is the tightest match. Broad can run past 10%. Most founders pick Narrow because it feels precise — and that instinct is usually wrong on TikTok.
How to build one
- Go to Assets → Audiences → Create Audience → Lookalike Audience.
- Pick a source. Best to worst: customer list (500+ rows), Pixel
CompletePaymentevent, then TikTok video viewers or engagers. - Set location to Thailand.
- Choose Balanced unless you have a strong reason not to.
- Name it so future-you knows the source — e.g.
LAL_Balanced_PurchaseList_TH.
Your source needs at least 1,000 matched users for a usable Lookalike. A 200-person customer list won't cut it. TikTok needs volume to find a real behavioral pattern, and a thin source produces an audience that's basically noise.
Where TikTok and Meta Lookalikes diverge
1. Source quality matters more on Meta
A Meta Lookalike off a 1,000-purchaser list is reliably its strongest cold audience — that's the whole reason you trusted it. On TikTok, the same list often beats a Broad, interest-free setup by only a slim margin. The creative carries far more of the weight.
2. TikTok refreshes intent faster
A Meta Lookalike built off a list from 12 months ago still performs fine. TikTok user behavior shifts week to week, driven by whatever's trending. A stale source ages worse — rebuild it every 60–90 days.
3. Match rates are lower in Thailand
Thai phone numbers and emails match at a lower rate on TikTok than on Meta. More Thai users sign up to TikTok with a number that was never in your CRM. Expect a 1,000-row list to match 400–600 users, not 800+. That smaller matched pool is another reason Narrow expansion struggles here.
The trap nobody talks about
Founders build a TikTok Lookalike, watch it lose to Broad targeting in week one, and conclude "TikTok ads don't work for us." The real problem is that they judged the audience on a tired creative.
On Meta, a great audience can rescue a mediocre video. On TikTok it can't. If your hook doesn't earn a 3-second hold, no audience setting saves it. So flip the order: test creative first, audience second — the opposite of how you'd sequence a Meta launch.
Quick reference
| Situation | Audience to use |
|---|---|
| Brand new TikTok account, no Pixel data | Broad (no Lookalike) + strong creative |
| Customer list 1,000+, proven on Meta | Lookalike, Balanced expansion |
| Tiny list (under 500 matched) | Skip Lookalike — use interests or Broad |
| Scaling a proven creative | Broad — let the algorithm widen |
What to do next
Before you copy an audience from Meta to TikTok, run the question through a blueprint. The Targeting section flags when your source is too small to build a reliable Lookalike, and recommends Broad vs Lookalike based on how much data your account actually has — so you don't burn ฿2,000 proving a Lookalike you never had enough data to build in the first place.
