Your Meta ads are finally humming. Sales come in on a steady rhythm, your ROAS looks healthy, and the audience just feels right. So you work up the courage to push your budget from Rp5 million to Rp15 million a month — and everything flips. Cost per lead spikes, prospect quality drops, and closing grinds to a halt. This is the classic scaling nightmare that keeps so many SMB owners from ever growing their accounts. The interest audience that once felt "fat" starts to fatigue, and Meta simply runs out of quality people to show your ads to.

This is exactly where the Lookalike Audience becomes your secret weapon. Instead of guessing at interests, you tell Meta to go find the "twins" of your best customers across all of Indonesia. This guide walks through how to build the right lookalike audience so your reach widens without your lead quality falling through the floor.

What a Lookalike Audience Is — and Why It's Built for Scaling

A Lookalike Audience is a pool of new people whose characteristics closely match a source you hand to Meta. You give the algorithm an "example" — a list of buyers, people who reached checkout, or your most loyal video viewers — and Meta analyzes hundreds of signals (behavior, interests, spending patterns) to find other people whose profile looks most similar.

The difference from ordinary interest targeting comes down to who's doing the guessing. Interests are your guess: you assume your customers love "food", "travel", or "parenting". A lookalike flips the process — Meta draws its own conclusion about who resembles your buyers, based on real data. That's why Facebook lookalikes are the backbone of so many ad accounts that successfully scale to big budgets: they protect quality while expanding the pool.

The One Rule Above All: Source Data Quality

Hold on to one principle before you touch a single technical step: a lookalike is only as good as its source. If you build a lookalike from people who only clicked your ad but never bought, Meta will go find more people who love to click and never buy. Garbage in, garbage out.

So rank your sources from most valuable to most generic. From highest quality down:

  • Repeat / high-value customers — people who bought more than once, or spent a lot. This is gold.
  • All buyers (Purchase) — from the Pixel / Conversions API, or an uploaded list of emails and phone numbers.
  • Add to Cart / Initiate Checkout — high purchase intent, even if they haven't closed yet.
  • Qualified leads — for service businesses, people who filled out a form and are genuinely serious prospects, not random sign-ups.
  • Engagement — 75%+ video viewers, or your most active followers on Instagram/Facebook. This makes sense when buyer data is still thin.

How Much Data Do You Actually Need?

Meta requires a minimum of 100 people from a single country in your source. But honestly, 100 is too few for good results. Ideally, gather at least 500–1,000 quality contacts — the more you have, the more accurate the match. If your buyer list is still only a few dozen, don't force it. Lean on engagement or lead sources for now while you keep collecting real buyers.

How to Build a Lookalike Audience, Step by Step

  1. Open Audiences in Meta Ads Manager (hamburger menu → Audiences).
  2. Click Create Audience → Lookalike Audience.
  3. Choose your source: an existing Custom Audience (for example, "Buyers – 180 days") or a customer list you upload.
  4. Pick the audience location: select Indonesia. If your business is local to specific cities, you can narrow it further at the ad set level later.
  5. Set the audience size from 1%–10%: this controls how "similar" the new audience is. 1% = most similar (most accurate, narrowest), 10% = broadest (huge reach, looser resemblance).
  6. Click Create Audience and wait for Meta to process it — usually 30 minutes to a few hours.

Making Sense of 1% vs 10%

This part trips a lot of people up. The percentage refers to the percentage of the target country's population. For Indonesia, with tens of millions of Meta users, even a 1% lookalike is already millions of people — more than enough for most SMBs.

  • 1%–2%: most similar to your best customers, best-protected lead quality. Start here.
  • 3%–5%: a bigger pool with slightly lower resemblance. Great once your 1% starts to fatigue and you need more room to scale.
  • 6%–10%: very broad, loosest resemblance. Usually reserved for brands already running very large budgets.

The rule of thumb: start narrow, prove it, then widen. Don't fire straight into 10% hoping for volume — you'll just burn money.

A Real-World Scaling Structure

Picture a local skincare brand on a Growth plan managing Rp20 million/month in ad spend. At first, all of it sat in interest audiences, but cost per purchase (CPA) started creeping up. Here's how a scaling structure might look:

  • Ad Set A — Lookalike 1% Buyers: Rp250,000/day. Your lead racehorse, highest quality.
  • Ad Set B — Lookalike 1–3% Add to Cart: Rp200,000/day. Captures high purchase intent.
  • Ad Set C — Lookalike 3–5% Buyers: Rp150,000/day. Room to widen when you need volume.

With this structure, when you want to add budget you simply raise it gradually — say 20–30% every few days — on the ad set that's performing most consistently, rather than doubling it overnight. Increasing budget too aggressively forces Meta to "reset" the learning phase and can wreck your results. These numbers are illustrative only; your real figures depend on your product, margins, and creative quality.

Don't Forget the Creative

No lookalike, however powerful, will save an ad with weak visuals and copy. A lookalike widens who sees your ad, but what they see still decides whether they buy. As you scale, prepare several creative variations so this broad audience doesn't get bored fast — that's how you avoid ad fatigue.

Common Mistakes That Make Lookalikes Fail

  • Source data that's too mixed: blending premium buyers with random web visitors muddies the signal. Separate sources by quality.
  • Going too broad, too soon: launching straight at 5%–10% before proving 1% dilutes quality and wastes budget.
  • Never refreshing the source: a lookalike built once and left for a year drifts out of date. Rebuild it periodically as new buyers come in.
  • Forgetting to exclude existing customers: if you're chasing new buyers, exclude your current customer list so you don't pay to reach people who already bought.
  • Stacking too many audiences in one ad set: overlapping lookalikes compete against each other in the auction and inflate your costs.

The Bottom Line

Lookalike audiences let you scale without gambling on interests — but only when they're fed high-quality data, started narrow, and paired with strong creative. Build from your best customers, expand in stages, and let performance, not impatience, decide when you widen. Remember, too, that in Aira Tech's model your ad budget is paid directly from your own Meta ad account to Meta — we never hold your spend, so you keep full control over every rupiah.

Want your scaling handled with a data-driven system instead of guesswork? Explore our Meta Ads services, see our management packages and pricing, or build a plan tailored to your business. And keep sharpening your strategy on the Aira Tech blog.