Audience Targeting for Music Ads: Who to Show Your Ads To
- →There are three audience types: lookalike, interest/detailed, and broad. Each has a job.
- →Broad plus strong creative is the modern baseline. Meta's delivery uses your conversion signal to find listeners.
- →Narrow is a test, not the default. Run lookalikes and interest stacks against broad and keep the cheaper one.
- →Targeting predicts who is likely to be interested. It does not guarantee a stream, and it cannot save a weak video.
The three audience types
When you build a music ad, you are really answering one question: who does Meta show this to? There are three ways to answer it, and picking the right one early saves you a lot of wasted spend.
The first is a lookalike audience. You hand Meta a source, meaning your best existing people, and Meta finds new people who resemble them. Meta's own definition: a lookalike is a way to reach new people who are likely to be interested in your business because they are similar to your best existing customers. The keyword there is "likely." It is a prediction, not a list of confirmed fans.
The second is interest, or detailed, targeting. This is the manual approach. You tell Meta who to show the ad to by picking interests, behaviors, and demographics: fans of a related artist, a genre, a festival, a record label. You are writing the guest list yourself.
The third is broad. You give Meta little or no constraint, usually just a country and an age range, and let the delivery system find listeners using the conversion signal from your pixel and Conversions API. This is the Advantage+ audience approach Meta now pushes by default, and it leans on the same click-out event your smartlink fires.
When each one works
None of these is the right answer for every campaign. Here is how I think about which to reach for.
A lookalike is only as good as its source. That is the whole game. Meta wants a source of at least 100 people and recommends 1,000 to 5,000. But size is not the limiter, quality is. A clean source of 1,000 real fans out-targets a list of 50,000 random page likers every time. If you build off junk, you get a junk audience no matter what percentage slider you drag.
Interests have their own trap. "Likes artist X" means Meta thinks someone likes X, not that they will like you. It is a hypothesis about taste, so treat it like one. Stack a few related artists and genres to keep the audience large enough to deliver, then check it against cost per conversion. Go too granular, layering one tiny niche on another, and the audience shrinks until delivery struggles and the cost climbs.
The modern lean: broad plus strong creative
For years the advice was to hand-build tight audiences. That has flipped. Meta's own default is now broad, where the delivery system uses your conversion signal to find buyers instead of relying on an interest list you guessed at. From running these campaigns, broad often matches or beats narrow once your pixel and CAPI are firing a clean click-out event. That is a practitioner read, not an official Meta guarantee, but it is what I see with my own money.
The reason is simple. When the conversion event is honest, the algorithm has a real target to chase, and it finds people who actually convert better than I can guess them by hand. Hand-built audiences fight that. They tell Meta to ignore everyone outside a box you drew on a hunch.
All three approaches lean on the same thing: a conversion signal that actually fires. That is the layer VLVTN owns. The smartlink landing page sends one click-out event to the pixel from the browser and the same event from the server through the Conversions API, deduped with a shared event ID so it counts once. Without that signal, broad has nothing to optimize toward and a lookalike has no quality source to grow from. Good targeting depends on clean tracking.
How to choose without wasting spend
The order matters more than the audience. Here is the loop I run.
Build the creative first, usually a short vertical video of the song. Pick broad as your baseline audience. Optimize the ad set for your smartlink's click-out conversion event, not for clicks or reach. Let it run long enough to settle, then open Ads Manager and read cost per result. If you want to pressure-test an audience, spin up a lookalike off a real source or a stacked interest test next to broad, and let the cheaper one win.
Iterate on creative before you iterate on audience. The video is the bigger lever, almost always. A bad audience on a great video still finds people. A great audience on a video people scroll past just burns money faster.
Grade your cost per conversion→The honest part: targeting is a prediction
Every one of these methods is Meta guessing. A lookalike is a guess about who resembles your fans. An interest is a guess about taste. Even broad is the algorithm guessing who is most likely to click out based on past converters. Some guesses are wrong, and that is exactly why you read cost per conversion and let the data decide instead of falling in love with an audience.
Two things targeting cannot do. It cannot guarantee a stream. A targeted impression is not a listen, and a click-out is not a play, save, or follow. What happens after the click is on the song. And it cannot rescue weak creative. The best audience in the world loses to a bad video. Get those two truths straight and the rest of this gets a lot easier.
Once you have the audience picture, go deeper on each one. Here is how to build a lookalike that actually finds new listeners, how to pick interests that map to real fans, and the full case for broad over narrow.
Read next: Lookalike audiences for music ads, Interest targeting for musicians, and Broad vs narrow targeting.
Frequently asked
What is the best audience for music ads?
For most campaigns now, broad is the baseline. You give Meta a country and an age range, fire a clean click-out conversion event, and let the delivery system find listeners. Lookalikes and interest stacks are worth testing, but they win only when they beat broad on cost per conversion. The audience matters less than the creative and the conversion signal.
Are lookalike audiences still worth it for musicians?
Yes, when you have a good source. A lookalike is only as strong as the people you build it from. A clean source of real fans (people who clicked out to a DSP, watched most of a video, or a real email list) makes a useful lookalike. A junk source makes a junk one, no matter what percentage you pick.
Should I pick interests like related artists, or go broad?
Try both and read the numbers. Interests are a guess about taste, not proof of fandom. Liking artist X means Meta thinks someone likes X, not that they will like you. Start broad as the baseline, run an interest stack as a test, and let cost per conversion decide which one keeps your money.
Can good targeting fix a song people scroll past?
No. The best audience in the world loses money against a video people skip or a song they do not replay. Creative is the bigger lever. Targeting points Meta at the right crowd. It cannot make a track people do not want into one they do.
Does targeting guarantee streams?
No. Targeting predicts who is likely to be interested, it does not guarantee anything. A targeted impression is not a listen, and a click-out conversion is not a play, save, or follow. What happens on Spotify is on the song.
Bradley J Simons founded VLVTN and runs his own paid Meta and Spotify ad campaigns as the artist Babbage. He writes about paid music marketing from the buyer's seat, with his own money on the line.