Your Distributor Doesn’t Matter Until It Really, Really Matters

Choose wisely. Distributors are no longer a utility. Photo Credit: Michael J. Whalen

Pay attention. It’s your music and they handle a lot.

I have been actively looking for a new music distributor, which is one of those jobs that sounds simple until you actually start doing it. I compared services, read terms, looked at pricing, thought about my catalog and tried to figure out which company I actually wanted sitting between my music and the streaming platforms.

In the end, I landed on AWAL, which is owned by Sony Music. I already knew AWAL from my years running Real Music, so I wasn’t starting from zero. There was history there, some familiarity with the company and at least a sense of what I was getting into. That mattered more to me than I expected because after looking closely at digital distribution in 2026, I kept coming back to one word that barely appears on comparison charts: trust.

We spend a ridiculous amount of time comparing whether one distributor costs a few dollars more than another. We study royalty percentages, automated splits, YouTube Content ID, analytics dashboards and whether UPC codes are included. Meanwhile, we casually hand these companies our masters, metadata, royalties and increasingly the authority to decide whether our music is legitimate.

That seems like something worth thinking about.

Ari Herstand maintains an exhaustive and genuinely useful comparison of digital music distributors at Ari’s Take. He looks at 18 companies and goes deeply into pricing, commissions, features, royalty splits and just about every other measurable variable. I don’t want to recreate that article. Instead, I want to look more closely at five distributors that represent very different approaches to the business: DistroKid, TuneCore, Symphonic, LANDR and CD Baby.

What interests me isn’t simply which one is cheapest. It’s what these companies have quietly become while musicians were busy treating distribution like plumbing.

For years, that was basically what it was. Upload a WAV file, upload the artwork, enter the metadata, choose a date and wait for the music to show up on Spotify, Apple Music, Amazon and everybody else. Digital distribution became cheap and easy enough that most musicians stopped thinking very much about it.

That was a genuine victory. For most of recording history, getting music into the marketplace required labels, physical distributors, warehouses, manufacturing, salespeople and money. Suddenly an independent artist could put a record on the same digital shelf as the biggest artists in the world for less than the cost of dinner.

Unfortunately, while we were congratulating ourselves on eliminating the gatekeepers, the job changed.

Your distributor may now be involved in deciding whether your recording contains unauthorized artificial intelligence, whether you really own what you are uploading, whether your streaming activity looks fraudulent and whether a copyright claim against you has merit. It may fingerprint recordings, freeze revenue, reject releases, remove tracks or terminate accounts. Some distributors are even asking questions about whether the AI tools used during production were trained on properly licensed material.

That is a pretty dramatic promotion from “company that sends my .WAV to Spotify.”

I wrote previously about aggregators becoming copyright cops in the age of AI, and the more I look at the distribution business, the more I think AI is accelerating a change that was already underway. These companies aren’t just delivering music anymore. They are becoming part of the enforcement layer of the streaming economy.

And the streaming services are giving them one hell of a job.

There Isn’t Even One Rulebook

A recent Billboard roundup of how the major music platforms are handling AI sent me down another rabbit hole, and the most interesting thing wasn’t any single policy. It was how completely different the policies are.

Spotify, for example, is introducing badges for artist profiles whose public identity appears to be an AI-generated photorealistic person. Those AI Personas are generally excluded from editorial and algorithmic recommendations unless the listener has intentionally engaged with them. Spotify is careful to point out that this badge is about who the artist appears to be, not necessarily whether the music itself was AI-generated. It has separate systems for disclosing AI involvement in recordings.

YouTube asks its music partners to classify delivered content as Fully Gen AI, Partly Gen AI or No Gen AI. A bass line generated by AI underneath human musicians can count as partly AI. Automated mastering, on the other hand, doesn’t. If the distributor supplies no AI information, YouTube says it can use other signals and make its own determination.

Apple Music is going in still another direction. It has added metadata for AI transparency and says content providers will be required to tag material where AI created a material portion of the content. Those disclosures are expected to become visible to listeners as “Made With AI” labels later in 2026. In practical terms, Apple is pushing much of the responsibility upstream to the record companies and distributors delivering the music.

TIDAL is more aggressive. It identifies and labels recordings it determines are wholly AI-generated, allows listeners to block those recordings and currently makes wholly AI-generated recordings ineligible for royalties. Most interesting for this article, TIDAL explicitly says it expects distributors to identify AI-generated content before it reaches the platform.

Bandcamp has made the philosophical decision much simpler. Music generated wholly or substantially by AI isn’t permitted.

And then there is Deezer, which apparently looked at this entire mess and decided to build a detector.

At its June 2026 peak, Deezer says more than 50% of all new music delivered to the platform was fully AI-generated, roughly 90,000 tracks per day. Yet those recordings represented only about 1% to 3% of actual listening. Even more astonishing, Deezer says up to 85% of the streams on fully AI-generated recordings during 2025 were fraudulent. It excludes detected AI music from algorithmic recommendations and editorial playlists and has begun taking down AI material associated with streaming fraud.

Half the supply. Almost none of the actual demand.

That explains a lot.

The streaming services aren’t suddenly getting aggressive about AI because the CEOs all woke up with an overwhelming concern for the sanctity of human creativity. They are being buried in the stuff. AI has turned what was already an enormous content-ingestion pipeline into something approaching an industrial waste-management problem.

Now imagine being the distributor in the middle of all of this.

You don’t have one rulebook. Spotify cares about artist identity and AI disclosure. YouTube wants levels of AI involvement classified. Apple wants AI transparency metadata. TIDAL may make the music non-monetizable. Deezer runs detection technology and removes AI material from recommendations. Bandcamp may not want the recording at all.

The distributor has to take whatever the musician tells it — “I used AI a little, but not like THAT” — and somehow translate it into several different technical, copyright and policy regimes.

That isn’t aggregation anymore.

It’s compliance.

There is another part of this that should make musicians even more interested in who is handling their distribution. TIDAL’s own terms acknowledge that AI-detection technology can produce false positives and false negatives, and it provides an appeal mechanism for recordings that artists believe were wrongly classified. Spotify similarly allows appeals when it applies an AI Persona designation itself.

We are officially entering an era where machines will be used to determine whether musicians used machines, and everybody involved already acknowledges that sometimes the machines will be wrong.

At that point, customer service stops being a minor feature.

Distrokid: Just Get the Music Out

DistroKid still comes closest to the original promise of digital distribution. It is inexpensive, fast and designed for artists who release a lot of music. For prolific independent musicians, that remains extremely attractive.

Its position on AI is also relatively easy to understand. AI-created music is allowed as long as you have the necessary rights, aren’t impersonating somebody without permission, aren’t infringing another artist’s work and aren’t mass-producing junk to manipulate streaming platforms. The basic philosophy seems to be that DistroKid doesn’t particularly care how you made the music as long as you aren’t stealing or cheating.

I appreciate the simplicity.

Things get more complicated when artificial streaming enters the picture. If a DSP determines that activity around your music is fraudulent, royalties can disappear and releases can be removed. The uncomfortable part is that musicians don’t always know where suspicious activity came from. You could hire a promotion service promising legitimate playlist exposure and discover later that some of its traffic triggered a fraud system.

Suddenly the distributor you barely thought about when you signed up becomes very important.

DistroKid also raises a larger catalog question because its normal subscription model means artists need to understand what happens if they eventually stop paying. There are mechanisms for keeping individual releases available, but the larger issue applies well beyond DistroKid: musicians spend enormous amounts of time thinking about the next release and surprisingly little time thinking about what happens to thirty years of music when they retire, lose interest, change credit cards or die.

A serious catalog is an asset. We should probably treat it like one.

DistroKid remains an excellent fit for a high-output independent artist who wants inexpensive, efficient distribution. I would just spend less time obsessing about how quickly the release gets delivered and a little more time understanding what happens when something becomes complicated.

Tunecore: Welcome to Compliance

TuneCore competes in much of the same unlimited-release territory, but its approach to AI makes the changing role of distributors particularly obvious.

TuneCore’s current policy says music involving generative AI can be eligible for distribution only when the underlying AI model relies on properly licensed training material. The policy can apply even when AI was only one element of the creative process.

Think about what that means for an independent musician. You use a software tool while producing a track and now you may need to understand how the company that developed that tool trained its model. Was the material licensed? Were rights holders compensated? Was the dataset opt-in? Is there documentation you can actually understand?

You opened the software because you wanted to finish a song. Now you are conducting due diligence on machine-learning datasets.

I understand TuneCore’s position. Consent, compensation and rights protection are reasonable principles. The problem is that the burden eventually lands on musicians who may have no realistic ability to independently audit how a complex AI model was built.

This is where the distributor stops looking like a delivery company and starts looking like a compliance department. It isn’t merely asking whether you own the recording. It may also want assurances about the technological process that produced it.

That requires a different kind of trust. You need to believe that the rules are clear, consistently applied and sophisticated enough to catch bad actors without turning legitimate artists into collateral damage.

Symphonic: Distribution Starts Looking Like Infrastructure

Symphonic may be the traditional independent distributor I find most interesting because it increasingly looks like something larger than a place to upload records. Its services include the kinds of royalty splitting, recoupment and rights-management tools that become genuinely useful when producers, featured artists and collaborators are involved.

Its approach to AI also gives away where the business is headed. Symphonic talks about disclosure, rights verification, fingerprinting, detection and provenance technology. Provenance is particularly revealing because it means the distributor isn’t simply interested in what the recording is. It is interested in where the recording came from and how it got there.

There are legitimate reasons for this. Generative AI makes it possible to manufacture huge quantities of audio extremely quickly. Voice cloning creates impersonation problems. Automated systems can generate fake catalogs, and streaming fraud can operate at a scale that would have been difficult to imagine only a few years ago.

Somebody has to deal with that mess, and increasingly the distributor is being asked to do it.

The question I care about is what happens when the system gets something wrong. Fraud detection makes mistakes. Copyright identification makes mistakes. AI detection makes mistakes. The platforms themselves are openly acknowledging that.

That means I would now include something on my distributor shopping list that I never would have thought about ten years ago: appeals infrastructure.

If your music gets flagged, can you reach somebody who understands the issue? Can a decision actually be reviewed? Is there a human being somewhere in the process with enough authority to say, “No, the computer got this one wrong”?

Those questions are starting to matter more to me than the dashboard.

LANDR: Out AI Will Check Your AI

LANDR is a fascinating case because AI isn’t simply something it has to police. AI is part of what the company sells.

LANDR began with automated mastering and has grown into a larger ecosystem involving production tools, plugins, samples, collaboration, distribution and AI-assisted services. At the same time, it monitors AI-generated music submitted for distribution and can ask artists for information about how a recording was created.

There is something wonderfully 2026 about that arrangement. A technology company can help you create music with advanced technology and then use other advanced technology to determine how much advanced technology you used.

Perfect.

LANDR also represents one of the more constructive approaches to the larger AI argument because it has explored licensed AI training in which artists can opt in and potentially participate economically.

That interests me considerably more than another round of AI GOOD / AI BAD.

AI exists. Musicians are using it. Technology companies will continue developing it. The meaningful questions are consent, ownership, attribution and compensation. If the industry can develop systems where music enters training datasets legitimately and rights holders share in the value, that seems considerably more useful than pretending AI can somehow be shoved back into the laboratory.

LANDR also makes the catalog question interesting because its model can allow music to remain available after a subscription ends while the economics of the relationship change. Conceptually, I like the recognition that recordings shouldn’t simply evaporate because somebody stopped paying a yearly bill.

CD Baby: No Robots, Please

CD Baby has taken one of the simplest approaches to generative AI: it currently doesn’t accept AI-generated music for distribution.

You can argue that the rule is too broad, and I probably would. There is a huge difference between using AI as one tool inside a human creative process and typing “write me twelve progressive-rock songs about heartbreak” and going out for lunch.

Still, there is something refreshing about understanding the rule.

CD Baby is also structurally different from subscription distributors because you pay per release instead of maintaining an unlimited annual pipeline. It takes a percentage of streaming and download revenue, but the catalog isn’t dependent on an annual distribution subscription.

That may be unattractive for somebody releasing twenty singles a year. For an artist with a handful of important records that should still be available decades from now, the calculation is different.

This gets to one of my problems with most distributor rankings. They assume every musician is optimizing for the same thing.

We aren’t.

Some artists release constantly. Some are protecting a mature catalog. Some have dozens of collaborators. Some need accounting tools. Some want access to label services. Some want everything as cheap as possible, while others would gladly pay more if an intelligent human answers an email when the music suddenly disappears.

There isn’t one “best” distributor because there isn’t one version of an independent artist.

Why I Ended Up at AWAL

After all this research, I didn’t choose any of these five.

I chose AWAL.

AWAL operates somewhat differently because it is selective rather than simply open to anybody with a credit card, and it now operates within Sony Music. But neither of those things was the deciding factor for me.

I chose AWAL partly because I knew them from my years running Real Music.

That history changed the decision.

Every competent distributor can put my music on Spotify. That part is a commodity now. What I was really deciding was where I wanted my catalog, royalties, information and inevitable future problems to live.

When you think about it that way, familiarity starts to matter.

We have built a music business filled with anonymous dashboards, automated emails, support tickets, fraud detectors and policy systems making decisions somewhere behind the curtain. Having some history with a company handling your work starts to feel almost luxurious.

It also made me realize how strange our relationship with distributors has become. Musicians complain about them constantly. Support is slow. Payments can be confusing. Metadata gets screwed up. Releases appear on the wrong artist page. Something gets flagged. Nobody responds. The distributor points at Spotify, Spotify points back toward the distributor, and eventually somebody sends you an FAQ that has absolutely nothing to do with your actual problem.

Then we upload another album.

That’s the strange part. These services are widely taken for granted and often not particularly trusted, yet we depend on them completely.

Trust Is the Feature We Never Compare

The more distributors become involved in copyright enforcement, fraud detection, AI disclosure and rights verification, the less comfortable I am treating them as interchangeable utilities.

I want to understand the rules and know what happens when one of those rules is triggered. I want to know whether an automated decision can be appealed and whether somebody with actual authority can correct a mistake. I want to know what happens to the catalog when the artist leaves the company, stops paying or dies.

Nobody is launching an exciting marketing campaign around GREAT APPEALS PROCESS!

It may still be more important than unlimited uploads.

Independent musicians spent decades trying to escape gatekeepers, and digital distribution finally seemed to eliminate them. Nobody at a record label had to approve your record. No physical distributor had to take it. No retail buyer had to give you shelf space. For a small amount of money, almost anybody could enter the marketplace.

Now we have gone to the other extreme.

At Deezer’s recent peak, more than half of all newly delivered recordings were fully AI-generated. Spotify is judging whether some artists themselves are synthetic personas. YouTube is asking distributors to distinguish fully AI music from partially AI music. Apple is building AI disclosure into the delivery chain. TIDAL is deciding that certain AI recordings shouldn’t earn royalties at all. Fraudulent streaming is mixed into the same ecosystem, along with impersonation, cloned voices, copyright claims and an avalanche of material that no human being could possibly review one song at a time.

Of course the gatekeepers are coming back.

They just have better software.

Some of this policing is necessary. I don’t want someone uploading my music under another name, cloning my identity or siphoning money out of the royalty pool with bots. I also don’t want technology companies treating the entire history of recorded music as free raw material.

But if distributors are going to become the enforcement layer between independent musicians and the streaming platforms, then choosing one because it saves you ten bucks a year suddenly seems ridiculous.

That is why I ended up choosing a company partly because I already knew them.

The distributor barely matters when everything works. The file uploads, the metadata is correct, Spotify accepts the release and the money arrives. Almost every competent distributor looks good under those conditions.

The relationship matters when something goes wrong. It matters when a copyright is challenged, when Spotify says your streams are suspicious, when an AI detector decides your recording is AI-generated when it isn’t, when royalties disappear or when six different platforms want six different versions of the truth about how your record was made.

That is when you discover what your distribution relationship is actually worth.

We call these companies aggregators. They increasingly look like distributors, accountants, compliance departments, fraud investigators, copyright cops and digital customs agents rolled into one.

Maybe that is unavoidable.

But if I am going to hand somebody that much control over my music, I want something considerably more valuable than a cheap annual subscription.

I want to trust them…