Some version of this story plays out every day.
An independent artist has a new single. They've heard Facebook and Instagram ads can drive streams. They set aside $200, make an ad with the cover art, link it straight to Spotify, target "people interested in indie music," and hit publish.
A week later: about 2,000 link clicks. 12 Spotify saves. $0.42 in streaming revenue.
They decide Meta ads don't work for music. I understand the feeling. But the platform did exactly what it was asked to do. It was just asked to do the wrong thing.
This is the playbook I'd hand that artist: why the first try fails, the two-phase setup that actually finds listeners, what to spend when you have somewhere between zero and 10,000 monthly listeners, and how to tell whether any of it worked.
First, the part nobody wants to hear
Ads do not make people like your song. They find the people who already would, faster.
If the first few seconds don't hook a total stranger, no budget saves it. You're paying Meta to show a lot of people something they'll scroll past. So run ads on your strongest song, not your favorite one. Those are often not the same song, and yes, that hurts.
If you honestly can't tell which song is the strongest, look at your own numbers first. The Read is a free one-page read of your streaming data, and it tells you whether your problem looks like the song or the reach before you spend anything.
Where the $200 actually went
The $200 campaign above was optimized for link clicks. That tells Meta's algorithm, "find me people who click links." It does, beautifully. It finds the cheapest link-clickers it can.
People who click links are not the same as people who listen to music. Plenty of those clicks are accidental taps on a phone. Some come from Audience Network, ads shown inside third-party apps, where a "click" can be a misfire on a tiny screen. The rest are people who were mildly curious and bounced in three seconds.
Now picture the trip from ad to fan. See the ad. Tap it. Open Spotify. Press play. Listen past 30 seconds, which is when Spotify counts a stream. Save it. Come back later. Every step loses people, and optimizing for the first step fills the funnel with people who were never going to reach the bottom.
You paid for clicks and wanted fans. Those are different purchases.
Stop hitting Boost
The Boost button in the Instagram app is the polite way to hand Meta your money and get nothing back. It picks the goal for you, optimizes for cheap engagement from anyone, and forces a "View Profile" button instead of anything useful.
Use Ads Manager. It looks like it was designed in 2013 and it's annoying for about a week. Then it's the only thing you'll use, because it's the only place where you choose what you're actually paying for.
The two-phase strategy
This is the core of the playbook. Don't try to sell the song to cold strangers. Find the people who stop for it first, then ask them to listen.
Phase 1: cheap video views to find your people
Take your best 15 to 30 second vertical clip. The hook, the drop, a striking lyric moment. Sound on.
In Ads Manager, pick the Awareness objective with ThruPlay as the goal, set the call to action to "No Button," and spend $5 to $10 a day. You aren't selling anything yet. You're paying small amounts to learn who actually stops scrolling.
Two settings matter more than they look:
Manual placements, Audience Network off. Automatic placements pull cheap junk views from random apps, and those people retarget terribly in Phase 2. Slightly pricier views now buy you a much better warm audience later.
Several hooks, not one. Your creative drives most of the cost, far more than targeting. Make three to five different openings, let them run a few days, and cut the losers. The first three seconds decide almost everything.
On targeting, there are two honest camps. One says go broad and let the clip find its audience. The other layers interests: the audience has to match the culture (say, "Rave"), the sound ("Techno" or "House music"), and a streaming platform ("Spotify" or "Apple Music"). One electronic producer who shared their breakdown went from roughly 12 Instagram followers in January to over 2,300 by March that way, while hitting 2,000+ streams on each release.
My take: at tiny budgets, a layered audience is a sensible start, because it stops Meta from spending your first days on whoever is cheapest to reach. Run a broader set next to it if you can afford it and let the numbers pick.
Expect thousands of views and almost no likes. That's correct for a views campaign. Watch hold rate instead: plays at 50 percent compared to 3-second plays. If a good share of people reach the middle, your hook works and your warm audience is growing.
Phase 2: retarget the people who watched
Build a custom audience of everyone who watched at least 50 percent of your Phase 1 clip. These people already told you they like the sound.
Run a "stream it now" or pre-save ad at them, pointing to your smart link, not straight to Spotify. Three reasons:
Your pixel lives on the smart link. Spotify won't run your pixel. Your landing page can, which means Meta can see who actually clicked out to a streaming app.
Listeners choose their platform. Not everyone uses Spotify.
The extra step is a filter, not a leak. Someone who saw the ad, tapped, landed, and picked their app made several small decisions to get there. Those are the people who save.
Then optimize the campaign for that click-out event, not for reach, engagement, or plain traffic. Now Meta hunts for people who leave and go listen, not people with a nervous thumb.
One trap: conversion optimization needs volume. Meta wants roughly 50 conversion events a week to learn. At $10 a day and $11 per result, that's about six results a week, and the algorithm guesses forever. Below about $15 to $20 a day, optimize for Landing Page Views instead, and move to the click-out event once you have the volume to feed it.
Leave it alone for four to seven days
Every campaign has a learning phase where the numbers look ugly and jumpy. You'll want to kill it. Don't. Every edit resets the learning, and you pay for day one all over again. Give it four to seven days before you're allowed to have feelings about it.
If a campaign is stuck at $0 rather than just learning, that's a different problem with a different fix. Why your Meta ads won't spend walks through the un-stick checklist, pixel checks, and the starving-budget problem in detail.
Budgets that make sense at 0 to 10K listeners
This is where most advice gets vague, so let me be specific.
Under $100 total. Honestly, Meta ads probably aren't the move yet. That money usually does more as a better clip you'll use everywhere, or simply kept, while you put time into free channels: pitching your release to Spotify editorial through Spotify for Artists, posting content that shows the music (not just your lifestyle), and showing up in your genre's community.
$200 to $300. This is a real test. For example: Phase 1 at around $7 a day for a week (about $50), then Phase 2 at $15 to $20 a day for 7 to 10 days. That's enough data to learn whether strangers stop for the clip and whether warm viewers click out. It's not enough to "launch a career," and nobody honest will tell you it is.
$500 a month and up. This is roughly where conversion-optimized campaigns to colder audiences start to have enough volume to work properly, and where ads can become an ongoing habit instead of an experiment. Signs you're ready: you can commit that amount consistently, you've already seen a decent save rate from earlier tests, and you're willing to learn the setup or pay someone to.
Now the cost reality. Say a campaign gets about 2,000 streams from $400 to $500 of spend. That's roughly $0.20 to $0.25 per stream. Spotify pays somewhere around $0.003 per stream, so you're spending about 80 times what those streams earn back. Plug your own numbers into the streaming royalty calculator if you want to feel it properly.
So why do it at all? Because the streams were never the point. The fans are. You're paying to find people who save, follow, and come back for the next release. If you judge ads as a royalty machine, they will always lose.
What to measure
Ignore impressions, views, and click-through rate. They count eyeballs and taps, not people choosing to listen.
To judge the ad: cost per click-out. What it costs for a real person to leave your smart link for a streaming app. Rough benchmarks for indie budgets: around 80 cents is a normal first campaign, under 50 cents is strong, under 30 cents is what experienced people reach after testing a lot of creative, and well above $2 means something is broken. From your second campaign on, the number to beat is your own best. The Only Ad Number That Matters goes deeper on this.
To judge the song: saves in Spotify for Artists. Expect fewer streams than click-outs. Some people never press play, and Spotify only counts 30+ seconds from a logged-in listener. That drop is normal. What matters is what the people who do listen do next:
Save rate. If more than about 10 percent of the ad-driven listeners save the track, it's landing with the audience you're reaching. Under about 5 percent, you're probably reaching people who will listen once and never return.
Follower growth. Followers hear about every future release. That's the part that compounds.
Listening after the ads stop. If streams drop to zero the day you stop spending, you bought reach, not fans.
Where the new listeners are. 500 new listeners in Chicago are worth more than 5,000 scattered across 80 countries, if Chicago is somewhere you could actually play.
Then read the two numbers together. Cheap click-outs but flat saves means the ad works and the song isn't connecting with this audience. Expensive click-outs but strong saves means the song is fine and the creative needs work, which is the good problem, because creative is fixable.
And be patient with the verdict. Practitioners consistently see algorithmic pickup lag about two weeks behind the last ad dollar, so judge a campaign two to three weeks after you stop, not the day you stop.
If a campaign is still expensive after the learning window, don't add money. Swap the audience first (cheapest test), then try genuinely new creative. If both fail, the song isn't connecting with cold audiences right now, and the budget is better spent on your next release. That's the campaign doing its second job: testing whether the song has the pull to justify more money.
Honest expectations
No ad setup guarantees streams, saves, or followers, and anyone selling "1,000 streams for $20" is selling something other than fans. Bought plays from people who don't care about your music can confuse the signals Spotify uses to recommend you. If someone pitches you a promo package that sounds too good, run it through the free promo offer checker before you pay.
What a well-run first campaign can realistically give you: a pool of warm people who chose to listen, a clear read on whether your song works on strangers, and a cost-per-click-out baseline to beat next time. Not glamorous. It is how ad spend starts compounding instead of evaporating.
The whole method in one breath
Cheap views to find the people who stop. Retarget the ones who watched half. Send them to a smart link and optimize for the real click to Spotify. Leave it alone through the learning phase. Judge the ad by cost per click-out and the song by saves.
None of this is about spending more. It's about spending on the right thing, which is exactly the kind of judgment a manager would bring. That's what the Ads Coach inside Musuni is for. It coaches you step by step inside your own Ads Manager, walks the checklist when a campaign stalls, and reads your results with you. It's advisory only: it never spends your money or touches your budget. And Musuni's smart links already fire the pixel events you need, so you can point Phase 2 at one and actually see your click-outs.