Targeting Mistakes That Kill Music Ad Conversions

Before you touch your audience, make sure targeting is actually the thing that is broken. Open Ads Manager and ask one question: are you getting conversions or not? If you are getting click-outs but Spotify stays flat, that is not a targeting problem. That is a song or expectation problem, and it lives in a different guide on tracked click-outs that do not turn into streams.

And if you are getting no clicks at all, the cause is usually the creative, not the audience. The video is what stops the scroll. You cannot out-target an ad nobody wants to watch. So fix the creative first, then come back here for the audience. This guide assumes the ad is decent and the cost is still too high or the clicks are too few.

This is the big one. Artists come from a world where smarter targeting meant tighter targeting, so they stack five interests, add an age range, pick a handful of countries, and end up with an audience so small Meta cannot do its job. Modern Meta rewards the opposite. The algorithm finds the people who click out for you, but only if you give it room to look.

I have run broad Advantage+ Audience for over a year and it performs at least as well as hand-picked interests, and it keeps getting better. Often I leave interest targeting blank or just add Spotify and let it run. The fix here is the easiest one on this whole page: widen the audience. You do not need to start over. Duplicate the ad set, open the targeting up, and let the new one learn next to the old one.

For the full picture on choosing audiences, including lookalikes, interests, and when broad wins, see the guide to audience targeting for music ads.

Where you advertise changes both the cost and the quality of what you get. Tier 3 countries run much cheaper, so it is tempting to point the campaign there and watch the conversion cost drop. But those clicks tend to be lower value and behave differently once they reach the DSP. A cheap conversion from a tier 3 audience is not the same as a cheap conversion from a tier 1 one.

If you did not set a country list, or you set one casually, check it. You want to be buying the listeners you actually want, judged against the cost of that specific audience, not chasing the lowest number on the screen. This is a setting you can change mid-campaign without rebuilding anything.

Lookalike audiences are only as good as the data you build them from. A lookalike off a tiny pixel, or off a fan list that is old or only loosely related to the song you are promoting, gives Meta a weak seed to copy. The result is an audience that looks targeted but does not convert any better than broad, and sometimes worse.

If a lookalike is underperforming, do not keep nursing it. Test it head to head against a broad ad set and let the numbers decide. In my experience broad holds up against most indie lookalikes, because the source data underneath them just is not big or clean enough to beat the algorithm at its own game. The full how-to on building lookalikes from the right source is in the targeting pillar.

Piling interest on top of interest feels like precision. In practice it usually just shrinks the pool and fights the algorithm. The other trap is that Meta removed the ability to target fans of specific artists, so the granular targeting people remember is mostly gone. You get broad genres like jazz music now, not Pat Metheny. At least one buyer I trust noticed his results drop after that change.

Chasing the old narrow targeting is a dead end, because it no longer exists. The honest move is to stop trying to replace it with interest stacks and put that energy into the creative instead. One or two interests at most, or none. Let the video find the audience.

The pattern under most of these mistakes is the same: not trusting the algorithm. There is a well-known test where the same set of ads was run as 84 separate ad sets at a few dollars each, and again piled into just two ad sets. Both campaigns converged on the same winning ad. The manually split-out version just spent more money to get to the same place. The ad decided the winner, not the targeting structure.

So the move is to give Meta enough creative to test and a broad enough audience to test it on, then get out of the way. Let the pixel gather data before you judge anything. Most first campaigns lose money while you learn, and that early spend is the tuition for finding the ad that works. Once it finds the winner, you scale that, not a clever audience.

When you widen placements, your cost per conversion often drops, and that looks like a win. Sometimes it is not. Certain placements reliably bring the number down but do not show up as real activity in Spotify for Artists. I do not claim to know exactly why, and you should not either. What I do is watch whether the cheap conversions actually move the streams. If they do not, the cheaper number is a mirage.

Fixing your targeting can lower your cost and put your ad in front of more of the right people. It cannot promise streams. A conversion is a tracked click-out to streaming, not a guaranteed play. If you widen the audience and your conversions get cheaper but Spotify stays flat, that is a signal about the track or the offer, not the targeting.

One more honest note: ad-driven traffic is rented, not owned. Turn off the campaign and the activity drops. The goal of good targeting is not just cheap clicks today, it is finding listeners who come back on their own after the budget stops.