Next time someone makes the case that Pandora is killing traditional radio, point them to this chart.

Yes, it’s incredible that Pandora has grabbed more than 4 percent (by Pandora’s metrics) of terrestrial radio listening. But that took 12 years, and even with listening hours growing each quarter, the company doesn’t have enough advertisers on board to turn a profit. Because of that, it is in a desperate fight to draw ad dollars from traditional radio. As of now, the score is approximately $17 billion (broadcast radio) to $800 million (Internet radio).

The other issue Pandora has is bandwidth. The majority of the company’s listening hours now come from mobile devices, but if even half of the traditional radio listeners switched over to mobile streaming for a few minutes, the entire network would crash. Put simply, terrestrial radio has become an incredibly outdated way to advertise — it’s fragmented and offers zero targeting (beyond region) and zero ways to measure effectiveness. But it’s still the simplest, cheapest most efficient way to broadcast.

Over the summer, I came across a company taking a unique hybrid approach to this problem. It’s a radio startup called Jelli. The company installs a server in a local radio station’s broadcast tower. It then begins broadcasting music on new radio stations that require no DJ, infrastructure, or ad sales team. Programming is completely determined by voting from Jelli users. The site and app are a bit like with its elements of group listening, voting, and chat. But they’re played via broadcast, and the stations are accessible from a radio, Web, or a smartphone, founder Michael Dougherty explains.

Today the company announced an important element to its business model: real time ads. They’re unique in that the feedback loop on Jelli is closed. Meaning, if there’s a call to action on a traditional spot radio ad, who knows if it is effective? Who knows who even heard it? But if a Jelli ad says, “download our app now,” with a display ad and audio messaging, Jelli can actually tell advertisers that 5 percnt of viewers downloaded that app. Jelli’s new ads feature anything from store finders for large retailers to local daily deals.

Currently the dominant player offering to close the feedback loop in streaming audio ads is Targetspot, which is like an ad network for online radio. Anyone from to Slacker Radio uses it.

Jelli is different because it’s delivered via terrestrial radio. More than 80 percent of its users listen via FM, so they’re not using wireless bandwidth. Which means only 20 percent actually use the feedback loop, but Dougherty says that is still attractive to advertisers because right now they have zero visibility. “Any feedback they can receive is very helpful for them to test ad copy, or simply provide some data to discuss with their client how the radio campaign went,” he said. Right now 650,000 users listen to 14 million hours of Jelli radio a month, Dougherty said.

Ironically, it has the exact opposite problem that Pandora and Spotify have. Those services have too many listeners and not enough ads. For Jelli, it’s backward: The demand for ads far outstrips inventory, with more advertisers hungry to buy ads than there are listening hours to run them. “We’re significantly oversold,” Dougherty said. “We have millions of dollars of demand, and we’re in the process of talking to big radio groups around partnering to expand rapidly to satisfy interest.” Nice problem to have.

San Mateo-based Jelli has backing from Intel Capital, Battery Ventures, First Round Capital, Triple Point Capital and a long list of angels.