How to Target the Right Audience Without Overspending
Why narrower targeting usually wastes a small ad budget instead of saving it, and the real audience-building steps that make Meta's algorithm work for you instead of against you.

Older advice says to target as narrowly as possible to avoid wasting money. On Meta's current system, the opposite is usually true. The algorithm needs enough room and enough data to find the people who'll actually respond, and squeezing the audience down too tight starves it of exactly that.
Step 1: Start broader than feels comfortable
Rather than stacking five interest layers to narrow the audience down, start with one clear, relevant targeting parameter and let the system's own optimization do the rest of the narrowing based on real response data, not guesses about who "seems right."
Step 2: Use Detailed Targeting to nudge, not restrict
Add one or two relevant interests or behaviors to point the algorithm in the right direction, rather than layering in requirement after requirement until the audience shrinks to almost nobody.
Step 3: Build a Custom Audience from your own data
In Ads Manager, go to Audiences > Create Audience > Custom Audience. Build one from people who've already shown real interest: website visitors (via your Pixel from the previous lesson), your customer list, or people who've engaged with your Page or Instagram. This is consistently your highest-quality audience, since they've already interacted with you.
Step 4: Build a Lookalike Audience
Once you have a Custom Audience of at least a few hundred people, use Create Audience > Lookalike Audience and choose a 1-3% match. This tells Meta to find new people who resemble your actual best existing audience, far more targeted than guessing at interests from scratch.
Step 5: Exclude the audiences you don't want to pay to reach again
Under Audiences, exclude your existing customers from acquisition campaigns aimed at finding new people, otherwise you're paying to advertise to people who've already bought.
Step 6: Change one variable at a time
If you adjust targeting, creative, and budget all at once and performance shifts, you won't know which change actually caused it. Adjust one variable, give it time to show real results, then adjust the next.
Step 7: Let the learning phase finish before judging anything
Meta's system needs roughly 50 optimization events (clicks, leads, or purchases, depending on your goal) in a rolling week before it exits what's called the learning phase and starts performing consistently. Editing the ad repeatedly during this window resets that clock, give it real room before deciding whether it's working.
Common mistakes
- Layering in so many interests and behaviors that the audience shrinks to almost nothing.
- Judging performance while the campaign is still in the learning phase.
- Forgetting to exclude existing customers from acquisition-focused campaigns.
- Changing targeting, budget, and creative simultaneously, making it impossible to know what actually caused a result.
- Ignoring Lookalike Audiences entirely as "too advanced" when they're one of the simplest quality upgrades available.
Your first step
- Simplify your current targeting down to one or two genuinely relevant parameters.
- Build a Custom Audience from your website visitors or customer list.
- Build a 1% Lookalike Audience from that Custom Audience.
- Exclude existing customers from any new-customer campaign.
- Let any change run at least a few days before judging it.