Broad targeting vs lookalike targeting for video ads comes down to one trade-off: broad targeting hands the algorithm your creative and a wide audience and lets it find buyers on its own, while lookalike targeting gives the algorithm a head start by pointing it at people who resemble your existing customers. Neither one wins every account. The right choice depends on how much customer data you have, how strong your creative is, and what stage your ad account is in.
Broad Targeting vs Lookalike Targeting for Video Ads: The Short Answer
If you have little to no purchase data yet, start with broad targeting and let the video ad itself do the qualifying. If you already have a list of past purchasers, add-to-cart events, or email subscribers, a lookalike audience built from that list is usually worth testing alongside broad. The comparison isn't really 'which is better' in the abstract — it's 'which one matches the data and creative I have right now.'
Both approaches run through the same ad delivery system on Meta, TikTok and YouTube. The platform still looks at signals like watch time, clicks and conversions to decide who to show your video ad to next. What differs is the starting point: broad gives it almost no hint, lookalike gives it a hint based on people who already behaved like a customer.
What Is Broad Targeting?
Broad targeting means you set minimal or no detailed audience filters — no interests, no custom audiences, often just an age range, a country, and a gender if relevant. The ad platform's delivery system is left to find people likely to respond, using the video ad's performance signals (watch time, clicks, add-to-carts, purchases) as its only guide.
Broad targeting works best when:
- Your account is new and has little historical purchase or engagement data to build a lookalike from.
- Your product has a wide potential audience rather than a narrow niche.
- You can produce several distinct video ad concepts to test, since broad audiences need the creative to do more of the qualifying.
- You want to avoid audience overlap issues that can happen when you run many small, overlapping interest-based or lookalike sets at once.
What Is Lookalike Targeting?
Lookalike targeting (sometimes called a similar audience) starts from a source list you provide — past purchasers, website visitors who added to cart, email subscribers, or people who watched a chunk of a previous video ad — and asks the platform to find people who share similar characteristics with that list within a given country or region.
Lookalike targeting works best when:
- You have a source list with enough people in it to be statistically meaningful — a handful of customers isn't enough signal.
- Your source list is specific: a purchaser list usually outperforms a generic 'all website visitors' list because it reflects people who actually converted.
- You want to extend proven buying patterns into new, similar prospects rather than starting from zero.
- You're scaling a product that already has a track record, and you want to keep the creative doing less of the heavy lifting on audience qualification.
Broad vs Lookalike: Side-by-Side Comparison
| Factor | Broad Targeting | Lookalike Targeting |
|---|---|---|
| Data needed to start | None or minimal | A source list of existing customers or engaged users |
| Where the 'smarts' live | Mostly in the video ad and the platform's delivery system | Partly in the source audience, partly in the ad |
| Best account stage | New accounts, limited history | Accounts with some purchase or engagement history |
| Creative demands | Higher — needs strong hooks and variety to qualify viewers | Still important, but the audience is pre-filtered |
| Risk of audience overlap | Lower, since there's no narrow set to overlap with | Higher if you run several similar lookalike sizes at once |
| Typical use case | Early testing, broad-appeal products | Scaling proven products, warm-list extension |
When to Use Broad Targeting
Use broad targeting as your default starting point if your ad account is new or your product hasn't built up a purchase history yet. Broad removes the guesswork of picking interests or audience sizes and puts the pressure where it belongs: on the video ad. If the hook, problem statement, and offer are clear in the first few seconds, the algorithm has something real to learn from.
A useful rule of thumb to test: when running broad, plan for a handful of distinct creative concepts (different hooks, different angles — problem-first, before-and-after, offer-first) rather than one video with small copy tweaks. Broad audiences are diverse, so one creative angle rarely resonates with everyone in it. If you're not sure how long a hook should run before it loses people, our guide on the ideal hook length for video ads is a good starting point.
When to Use Lookalike Targeting
Lookalike targeting earns its place once you have a source list that's both large enough and specific enough to be useful — think past purchasers or people who completed checkout, not just anyone who ever visited your site. A purchaser-based lookalike tends to carry more predictive signal than a broad website-visitor list, because it reflects people who already decided to buy, not just people who clicked through.
When testing lookalikes, keep these points in mind:
- Test one lookalike size at a time against broad rather than running five overlapping lookalike percentages simultaneously — overlap can split your results without telling you much.
- Refresh the source list periodically as your customer base grows, especially after a product launch or seasonal spike that shifts who's actually buying.
- Pair the lookalike with creative that still earns attention on its own. A pre-filtered audience still needs a reason to stop scrolling.
How Creative Changes Depending on Targeting
The targeting choice should influence how you brief your video ad, not just where you aim it.
For broad targeting, the ad needs to qualify the viewer itself. Open with a hook that states the problem or outcome plainly within the first few seconds, since you can't rely on the audience already being pre-interested. Vertical, 9:16 formats built for mobile feeds tend to perform better here because they match how people are already scrolling — our breakdown of vertical video format for ads covers the specs if you want a refresher. Plan for several creative variations so the algorithm has options to test across a diverse audience.
For lookalike targeting, you can afford to assume some baseline familiarity or intent, especially if the source list is purchasers. The ad can lean more on proof (reviews, before-and-afters, demonstrations) and a clear offer, since the viewer is statistically closer to someone who already converted. That said, don't drop the hook entirely — a lookalike audience still scrolls fast and still needs a reason to stop.
In both cases, captions matter because many viewers will have sound off by default; design the ad to work on mute, with word-by-word captions carrying the message if the audio isn't on.
A Simple Test Plan to Decide for Your Account
Rather than debating broad vs lookalike in theory, run a structured test:
- Split budget evenly between a broad ad set and a lookalike ad set (if you have a usable source list), using the same video ad creative in both to isolate the targeting variable.
- Let each ad set run long enough to exit the learning phase before comparing results — judging after a few hours tells you almost nothing.
- Compare cost per result, not just raw volume, and watch for early signs of audience overlap if you're also running other ad sets.
- Once you have a signal, double down on the stronger performer but keep testing a small broad budget in the background — audiences and algorithms shift, and what worked last quarter may not hold.
If both approaches perform similarly, that's a useful outcome too — it usually means your creative, not your targeting, is the bigger lever to pull next.
How FrameNotion Fits In
Whichever targeting approach you're testing, the bottleneck is usually the same: having enough distinct, decent video ad creative to actually run the test. FrameNotion was built for exactly that gap. Paste a product link and FrameNotion AI writes and renders a custom 30-second vertical video ad from scratch — hook, problem, benefit, proof, offer, call to action — with an AI voiceover, music cut to the beat, sound effects and word-by-word captions, built to work on mute too. A finished ad takes about 10 to 20 minutes, which makes it realistic to produce the handful of creative variations a broad-targeting test needs, or to refresh the proof-and-offer angle for a lookalike audience built from purchasers.
Every ad also comes in 4:5, 1:1 and 16:9, so the same creative can run across Meta, TikTok and YouTube Shorts without starting over for each placement. FrameNotion doesn't publish ads or report on their performance — you'll still set up the campaign and read the results in your ad platform — but it covers the part of the process that usually takes the longest: going from a product page to a finished, platform-ready video. Browse example ads made with FrameNotion or see how it works to get a feel for what's possible, and check pricing — the first ad is free, no card required.
Final Take
Treat broad targeting and lookalike targeting as two tools for two different data situations rather than competing philosophies. New accounts with little history lean on broad and strong creative. Accounts with a real purchaser list lean on lookalikes to extend what's already working. The accounts that scale fastest usually test both at once, track results over a full cycle, and keep feeding the test with fresh creative — because no targeting setting fixes a video ad that fails to earn attention in the first few seconds.
Frequently asked questions
Can I run broad targeting and lookalike targeting at the same time?+
Yes, and many accounts do. Running both with the same creative lets you compare which one pulls a better cost per result for your specific product, as long as you watch for audience overlap between the two ad sets.
How big does a source list need to be before a lookalike audience is worth testing?+
There's no universal number worth stating as fact, but very small lists (a handful of people) rarely give the algorithm enough signal. Treat list size as something to test: if a lookalike built from a small list underperforms broad, your list likely needs to grow before trying again.
Does broad targeting work for niche products?+
It can, but niche products often need the video ad itself to do more filtering since there's no interest layer narrowing the audience. Clear problem-first hooks and specific proof in the creative help a broad audience self-select faster.
Should I use a purchaser list or a website-visitor list for a lookalike?+
A purchaser-based list generally carries stronger signal because it reflects people who already converted, not just people who clicked. If you only have visitor data, it can still be worth testing, but expect it to perform differently than a purchaser-based lookalike.
Does the video ad itself need to change based on targeting type?+
Yes. Broad audiences benefit from a stronger qualifying hook and more creative variety since there's no pre-filtering. Lookalike audiences, especially purchaser-based ones, can lean a bit more on proof and offer since the viewer is statistically closer to an existing buyer.