The Night a Song Didn’t Finish

In the summer of 2009, the song “The Final Countdown” began playing on Live 105 in San Francisco, sliding into the station’s rotation the way thousands of songs had before it, carefully programmed, thoroughly tested, and timed to keep listeners from drifting away.

But the song didn’t make it to the end.

Somewhere between the first verse and the chorus, it simply stopped, and the decision came from the audience who had voted on whether or not the song should keep playing.

When the song was playing on the air, the audience was listening to it online through a platform called Jelli, where every track came with two buttons to vote: Rocks or Sucks. Enough votes in either direction determined whether the current song deserved to keep going.

For the first time in music history, radio and listening habits were being impacted by an audience in real time. The popularity of new songs was democratized to listeners and the fate of music wasn’t in the hands of some old school DJ or Studio Exec.

A Medium That Never Changed

Radio had always presented itself as interactive. Stations took requests, counted down listener favorites, and encouraged participation, but the structure behind it was anything but democratic.

Playlists were built by program directors using research, ratings data, and an understanding of what kept audiences from changing the station. By the late 2000s, the system worked reliably, but it had also become predictable. The same songs reappeared on tight cycles, and the sense of discovery that once defined radio had largely been engineered out of it.

At the same time, the internet was moving in the opposite direction. Content was no longer just delivered, it was surfaced, voted on, and reshaped by the people consuming it.

Jelli emerged from that contrast as a way to bend music and listening habits slightly closer to how the rest of media was starting to behave based on audience preferences.

Turning Radio Into Something You Could Play

Jelli’s founders, Michael Dougherty and Jateen Parekh, built the system as a layer on top of existing stations rather than a standalone product. Listeners tuned in as usual, but opened Jelli on a second screen, where they could see what was playing and influence what happened next.

The interface was simple, but the system behind it created something truly unique. Songs weren’t locked into place once they started. They moved, stalled, or disappeared depending on how listeners responded. Some users gained additional tools that let them push tracks up more aggressively or knock them down faster, adding a subtle competitive dynamic to what had traditionally been a passive experience.

The founders described it as a multiplayer game played through a radio station, which sounded like a stretch until you watched a song rise or collapse in real time based on collective input.

For a brief period, radio felt less like a broadcast and more like a shared system that could turn unpredictably depending on who was paying attention.

From Segment to Full-Time Format

What began as a contained experiment on Live 105 quickly expanded. The station turned it into a recurring show, giving listeners a regular window where the playlist responded to their input instead of following a fixed rotation.

As interest grew, Jelli spread to other stations, and the experiment started to shift from a novelty into something closer to a format.

The most ambitious version appeared in Las Vegas, where two stations, KXLI and KYLI, adopted 24-hour listener-controlled programming. Instead of limiting the concept to a few hours each night, the entire station became responsive to audience voting, with every song subject to the same real-time feedback loop.

For a moment, it suggested a different future for radio, one where programming wasn’t decided in advance but continuously shaped by the people listening.

Where It Started to Break

The system worked exactly as intended. Votes came in, songs moved, and the playlist reflected the preferences of the people actively participating.

Radio had always succeeded because it required almost nothing from its audience. You turned it on and let it run in the background. Jelli introduced a layer of participation that, while simple, changed that dynamic. It asked listeners to open another device, pay attention to what was playing, and keep making small decisions over time.

A core group of early adopters embraced the role of “social listening,” but most listeners didn’t migrate their behavior and listening habits.

At the same time, control over music was becoming easier in a different way. Streaming services were beginning to offer something more direct, allowing people to choose exactly what they wanted to hear without relying on a shared system or the preferences of others.

Jelli depended on collective input, but the market was moving toward personalized, individual control.

The End of the Experiment

As participation leveled off, stations began stepping away from the format. Some changed ownership, others returned to traditional programming, and the broader sense that this might become a new standard for radio gradually faded.

In June 2014, the remaining stations using Jelli shut it down.

The idea of a listener-controlled radio station, where songs could be voted off the air in real time, disappeared as quickly as it had arrived.

The Part That Stayed

What remained was the platform that made it possible.

Running Jelli had required building software that could react instantly, coordinate with station infrastructure, and manage constantly changing inputs. Those same capabilities turned out to be far more valuable in another part of the business.

Radio advertising.

Buying and scheduling ads was still largely manual, handled through direct negotiations, spreadsheets, and delayed reporting. Jelli’s infrastructure made it possible to automate much of that process, allowing campaigns to be planned, adjusted, and monitored with a level of speed the industry hadn’t previously had.

The company shifted its focus, and this version of the product spread. Within a few years, hundreds of stations were using Jelli’s platform to manage advertising, even as the listener-controlled format that introduced the company faded away.

What the Experiment Actually Proved

Jelli set out to make radio behave more like the internet, interactive, participatory, and shaped by the crowd in real time.

For a brief period, it succeeded in showing what that might look like. Songs could rise, fall, or disappear entirely based on collective input, turning a traditionally one-way medium into something more dynamic.

But it also revealed that most people didn’t want to program the radio in a social way. They wanted either to listen without thinking about it or to choose exactly what they wanted themselves.

The shared layer in between, where control was distributed across a crowd, was interesting, but it never became essential.

What lasted instead was the infrastructure built to support it, software that fit more naturally into how the industry actually worked.

In 2018, iHeartMedia acquired Jelli, since they had pivoted successfully and built a strong ad-tech platform for media buying in the world of audio streaming.

Jelli.com didn’t become the “future of radio” like they initially thought they would, but like a lot of internet bets, the initial idea for the startup faded and the infrastructure they built within audio streaming ended up being the real asset of their business.

‍