By Robin Morgan, Co-Founder of intellijend
Here's the uncomfortable arithmetic. Every single day, around 106,000 new tracks are poured onto streaming services, a number that climbs every year like it's being chased. By the end of 2025 there were roughly 253 million songs sitting on these platforms.
Take a breath and look at that pile, because here's what became of it: 88% of those tracks were streamed fewer than a thousand times last year, and 120 million of them, nearly half, got ten plays or fewer. Ten. You could manage that by accident with your own thumb.
Meanwhile, up at the summit, exactly 29 tracks crossed a billion streams in 2025, two fewer than the year before.
So that's the shape of modern music: an ocean of the unheard, and a teaspoon of the colossal. The middle has quietly been deleted.
Now add the part that should make your eye twitch. Over on Deezer, around 75,000 of those daily uploads, about 44% of everything arriving, are now fully generated by AI. More than two million machine-made songs a month. In April 2026, an AI-generated track sat at number one on iTunes in five countries at once. When Deezer asked listeners to pick the real thing from the fake, 97% of them couldn't.
The machines aren't just out-producing you on volume. They're competing on quality, and most people can't hear the join. This is why "good" no longer rescues anyone. For most of music history, making something genuinely good was the hard part. That world is gone.
Good is now the cost of entry, not the prize, what gets you into the lottery, not what wins it.
But here's the good news, hiding in plain sight inside those same numbers. AI may be 44% of what gets uploaded, but it's still only 1 to 3% of what actually gets streamed, and 85% of even that is flagged as fraud. The machines can flood the landscape, but they can't make a single human being care.
That gap, between making something and making someone care, is the whole game now. And it's a game humans still win. You just have to actually play it.
I know this because I watch it in real time. I'm one of the co-founders of intellijend, a music marketing platform that runs campaigns for over 6,000 artists, from bedroom acts to names like Milky Chance, Ben Böhmer and BLOND:ISH, with more than $1M in marketing spend moving through the platform every month. We also recently launched rosterOS for the people juggling whole rosters: labels, managers, agencies.
The point of all this: it means I get to watch what actually makes a track break, in real time. And it's almost never about whether the song is good.
Is There a System Behind All This?
I won't bury the lead. Yes, there's a system, and the cheerful news is that it's learnable. Here are some things we've learned while watching campaigns come together in real time:
Artists that build their song to be heard get heard.
You get about fifteen seconds before a stranger's thumb decides your fate, so earn them. Hook early, keep the intro short, stay under roughly four minutes, write a melody someone can hum after one listen. This isn't dumbing down, it's remembering you're up against 106,000 tracks that landed today.
There's a number quietly running your career, respect it.
Spotify gives every track and artist a hidden "popularity score," 0 to 100. It isn't in Spotify for Artists, but it's pullable through Spotify's developer API, and it's the thing deciding whether you get fed into Discover Weekly, Release Radar and the rest, which is where the bulk of streams actually live. We went through a year of data on a thousand-plus artists to find what moves it. The answer is velocity: not how many streams, but how fast.
A burst of real streams in a short window spikes the score, the spike holds, and the thresholds come in tens, a track creeping from 29 to 30 suddenly gets shoved into far more algorithmic real estate. Each track-level spike also drags up your artist score, and that's the one that compounds.
Ads on Meta that perform well tell the rest of the internet what's up.
Velocity comes from people, quickly. Not bots, not a TikTok dance you hate. The cleanest way to summon the right humans on demand is Meta Ads, the same machine pretty much every business uses to acquire attention online, pointed precisely at the genre, country and age most likely to love you. This works, and it works best when you can attract real viewers on those ads, not bots.
+Read more: "How Musicians Leave Money on Their Own Websites"
Now how do we actionize it all? Here's how.
My 8 Best Tips for Building a Strong Campaign
1- The best ad doesn't look like an ad.
Nobody opens Instagram hoping to be sold to. Overproduction is the single most common mistake, a glossy, obviously-an-advert clip gets skipped on sight. Shaky, native, slightly imperfect phone footage beats it, because it feels human and blends into the feed. Plus, you don't want to be competing with multinational brands.
2- Win the first three seconds or lose everything.
Lead with the best part of the song, make what they see match what they hear (techno belongs at a rave, not a beach), and keep it simple.
3- Lean on proven angles.
"RIYL", "For fans of [a huge artist you sound like]", "a song for overthinkers", a friend's genuine reaction, b-roll that matches the mood with your lyrics over it. There are a handful and they work across every genre. If you want more, check out our YouTube channel. On intellijend, you can view a bunch of resources on this.
4- Test relentlessly, because your taste is not the data.
This is the bit most artists skip, and it's the whole game. You don't know which ad will work, nobody does. So you run many: say ten creatives against four audiences, forty combinations. Kill the losers fast and pour budget into the winners. (Duplicate a winner rather than just cranking its budget, Facebook treats a sudden budget hike as a brand-new ad and the performance wobbles.)
5- Be ruthless and objective.
The numbers tell the truth, your ego doesn't. A play costing under ~30 cents is good, under 20 cents is an "evergreen" you can run for months. And the lovely part: every release teaches you who your people are, so it gets cheaper and easier over time.
6- On budget, think asset, not expense.
Roughly $300 a release is a real floor, $600 is a good number, $1,000+ is excellent. But don't picture money set on fire for a stream count. It's marketing and unlike most marketing, you keep the upside: real followers, fans who come back, and a tracking pixel that learns your audience and makes every future campaign cheaper. You're not renting streams. You're building a machine.
7- Front-load the spend, the bit most people miss.
Don't dribble the budget out evenly. Pile the bulk of it into the first few days after release. Remember move two: the score rewards velocity. The same money spent in a concentrated burst spikes your popularity score far higher than the same money spread thin and the higher the spike, the longer it holds and the more algorithmic pushes it triggers. In plain terms: front-loading turns the streams you paid for into a fuse that lights the streams you don't. Paid demand, igniting free demand.
8- Repeat this process every four to six weeks.
That's the engine. Each release spikes your track score, which lifts your artist score, so your next release starts from a higher base more traction out of the gate, and your back catalogue gets fed back into the machine too. It's not one viral miracle. It's a flywheel: release, spike, compound, repeat. Boring, maybe, but also exactly how it works.
The catch is that running it by hand the ad structures, the pixel, the landing pages, the daily optimizing, the front-loaded schedules, the forty combinations is a proper faff, because Meta built that machine for ecommerce teams, not for someone who'd rather be making music.
Where do we go from here?
I'm optimistic. For most of music history, the marketing infrastructure that labels guarded: the data, the testing, the coordinated launches, was simply out of reach for an independent artist. That wall is coming down. The same capabilities are getting cheaper, simpler, and more connected, and they're landing in the hands of people working with five-dollar budgets instead of five-hundred-thousand-dollar ones.
At intellijend we're building toward pulling the scattered pieces: campaigns, creative, data, into one connected system, and eventually beyond streaming into the rest of an artist's world: ticketing, merch, the whole picture. But honestly, whether you ever use what we built or not, the shift is bigger than any one company. The independent artist is finally inheriting the toolkit that used to define the majors.
So here's my pitch, and it's not really a pitch:
Stop releasing your music and hoping. Start launching it like it matters, because the tools to do that, at long last, are yours. Master that, and something rather nice happens: the muscle you build doesn't stay in your release strategy.
Once you start thinking like an entrepreneur, testing, reading the data, owning the outcome, that same instinct quietly spills into everything else. Your bookings. Your brand. Your merch, your fan relationships, the deals you say yes and no to. You stop waiting to be picked.
None of this asks you to become less of an artist. Make the song you'd make anyway. Just refuse to let it die in silence, because in 2026, waiting to be discovered isn't humility, it's the riskiest move on the board. Learn the boring parts. Do the slightly uncomfortable work.
The artists who win the next decade won't be the most talented; they'll be the ones who decided being good wasn't going to be enough, and got on with it.
Good music doesn't just find its audience. But you can go put it in front of them.
Robin Morgan is the Co-Founder of intellijend, a music marketing platform built to help independent artists and labels run smarter, data-driven campaigns. What started as a book, The Independent Artist Blueprint, co-authored with Nico Jend, has grown into a full suite of tools serving the modern music business, plus rosterOS, which powers roster management and content creation at scale for labels, managers, and agencies.
