AI vs. Traditional In-Store Music

The in-store music industry has operated on the same model since Muzak started in 1934. License existing recordings, organize them into playlists, distribute them to stores. The delivery technology has changed (satellite, streaming, cloud), but the underlying structure has not. The music was made for consumers. It is repurposed for retail.

Generative music changes that structure at every level. Where the music comes from. Who it is targeted at. Whether anyone is measuring what it does. Who owns it.

Traditional providers license recordings from labels and publishers. The catalog is finite. Even a library of two million tracks is a fixed set, and playlists built from it rotate and repeat. Employees hear the same songs dozens of times a month. Frequent customers notice.

Every track also carries licensing obligations. ASCAP, BMI, SESAC, and direct label deals all apply. For a multi-location retailer the compliance burden and cost scale with every store.

Original generative music produces new compositions as it goes. No catalog, no repetition across the day, no third-party licensing exposure because there is nothing to license.

Traditional services target by genre, era, or mood label. A retailer picks “Upbeat Pop” or “Chill Acoustic” and the platform delivers a playlist matching that description.

Genre is a blunt instrument. The variation within any genre is enormous. Two tracks both labeled “jazz” can produce completely different behavioral effects in a store. A Miles Davis ballad and a Buddy Rich big band number have almost nothing in common from the perspective of what they do to a customer’s pace and mood. Genre tells an operator very little at the resolution that matters.

Generative music can be built around the customer a specific retailer is trying to serve rather than around a genre bucket. The output might resemble jazz, ambient electronic, indie folk, or something less categorizable. The genre label is a byproduct.

This is the consequential difference. Traditional music providers do not measure the behavioral impact of their playlists. They cannot tell a retailer whether the music playing at 2 PM on a Tuesday helped or hurt sales. The audio is an input with no measured output. An expense line with no return calculation.

Generative music, when the provider has bothered to instrument it, can produce data about what played and when. A retailer can put that alongside the analytics they already pay for and start asking correlation questions that no catalog provider has ever been set up to answer.

For anyone who has to justify line items, that is the difference between an overhead cost and an investment that can actually be reviewed.

Traditional providers charge fifteen to fifty dollars per location per month for the software, plus thirty to one hundred dollars or more per month in performing rights organization fees depending on store size and territory. For a retailer with fifty locations, the annual cost of licensed music can reach fifty to ninety thousand dollars. That buys access to a catalog of songs that were not designed to move a business metric.

The comparison to a generative provider is not apples-to-apples. The relevant comparison is cost-per-outcome, not cost-per-location.

If a retailer does not intend to measure the impact of in-store audio on business outcomes, a thirty-dollar-per-month playlist service is fine. The music fills silence. It probably does not actively harm the business. It costs very little.

If a retailer intends to treat audio as a performance variable and hold it to the same standard as visual merchandising or digital spend, the catalog business model was never set up to deliver that. The operators who take the measurement question seriously are the ones this comparison matters to.

For what a pilot actually looks like end to end, see how the pilot works.