Why Playlist-Based Music Systems Consistently Underperform

Professionally curated playlists showed no significant sales effect across a 140-store field study — the largest retail music experiment ever conducted. The playlist model has four structural flaws that no amount of curation can fix: static tempo, repetition, preference mismatch, and congruence drift.

In this video I break down why the entire playlist-based approach to retail music is fundamentally broken, regardless of how good the curation is, and what the research says about the alternative.

Every major retail music provider in the world is built on the same model: curate a playlist, push it to stores, repeat. And the research suggests that model is fundamentally broken. Not slightly underperforming. Structurally incapable of doing what the science says actually works.

Here’s the core issue. Playlists treat music as a single variable — pick the right songs, in the right order, and you’re done. But forty years of peer-reviewed research says music doesn’t work as a single variable. It works through interactions. A 2012 study in Marketing Letters found that tempo and musical mode interact — slow tempo in a minor key lifted sales by about twelve percent, but slow tempo in a major key did nothing. The tempo effect completely depended on the key. A playlist builder might nail the tempo and completely miss the mode, and the research says that combination is no better than random. It gets more complex. A 2017 study in the Journal of Retailing — 43,676 shopping baskets across real stores — found that tempo interacts with crowd density. Fast music helped when the store was packed, lifting spending by around eight percent. But that same fast music did nothing when the store was empty. A static playlist can’t respond to how many people are in your store at 2 PM versus 6 PM. It just plays the next song.

Playlists also can’t solve the congruence problem. Research published in the Journal of Retailing in 2001 showed that music and ambient scent need to match on arousal level — high-energy music with stimulating scent, or calm music with relaxing scent. Mismatches performed worse than either element alone. A playlist doesn’t know what your store smells like. It doesn’t know you just lit a lavender candle and it’s about to serve up a high-BPM track that creates sensory dissonance. A 2006 study in the Journal of Business Research made it even clearer: music that doesn’t fit the brand actively damages brand perception. Not fails to help — damages. So a playlist that’s generically “good” but not specifically yours is a liability dressed up as a solution.

There’s another structural problem with playlists: they repeat. They have to — they’re finite. Even a playlist with 500 songs cycles through in a week of eight-hour days. And repetition triggers a familiarity penalty. Research from the Journal of Business Research in 2000 demonstrated that familiar music makes customers leave faster while perceiving time as slower — the exact opposite of what you want. Unfamiliar music kept them longer while making the visit feel shorter. A playlist that your regular customers hear three times a week is training their brains to build internal clocks that accelerate their exit.

And the biggest indictment? When researchers tried to validate the classic music-drives-sales findings at real scale, the results collapsed. A 2025 study across 140 stores, presented at the European Marketing Academy conference, found no overall tempo effect. Zero. The only customers who responded to tempo changes were loyalty members — people who were already committed to the brand. Casual shoppers didn’t respond at all. What happened? The researchers noted that playlist-based interventions in real stores couldn’t control for all the interacting variables — mode, density, brand fit, time of day. The effect is real in controlled settings but vanishes when you try to execute it with a static song list across diverse conditions.

This is the gap we built Entuned into. Generative music isn’t a better playlist — it’s a different architecture entirely. Every track is composed in real time, which means we can control tempo and mode and energy level simultaneously. No repetition penalty because nothing repeats. No brand-fit gambling because the music is generated to match your specific parameters. The research doesn’t say music doesn’t work. It says music works through complex interactions that playlists can’t manage. Forty-three thousand baskets of data say tempo matters — but only at the right density, in the right key, for the right customer. A playlist can’t solve that equation. A generative system can.

A 2025 field study across 140 stores — the largest ever conducted — found that professionally curated, brand-matched playlists produced no significant overall effect on sales. The only customers who responded were existing loyalty members. The playlist format has structural problems no amount of curation can fix.

Four things. They’re static in a dynamic environment (can’t adjust tempo for crowd density). They repeat (torturing staff and degrading the experience). They can’t match individual preferences across a diverse customer base. And they drift out of brand congruence over time — and even slight incongruence reverses the positive effects.

Generative music — music created in real time for your specific environment. No repetition, no congruence drift, no static tempo locked to the wrong conditions. That’s what Entuned does. It’s not a better playlist — it’s the end of playlists. Free tier at entuned.co. Full citations in the description. This is video 26 of 50 in this series.