Podcasting is about to become more like radio. Nothing will change with the actual mechanics of podcasts — how they’re produced, how they’re distributed, or who listens. The change will come in what producers know about who listens, and when they stop listening.
This fall, Apple will release new analytics for podcasts
that will show producers how many people listen to episodes, how long they listen, and whether they skip ads. Previously, producers relied largely on the number of downloads to try to determine their show’s popularity, not knowing whether anyone listened to the files they downloaded. Some advanced data is available through other listening apps already, but since the majority of listening passes through Apple, the detail and scope of the new metrics will be unprecedented for podcasting.
But it won’t be unprecedented for audio.
This isn’t the first time audience measurements for audio have become more sophisticated. And
last time data significantly improved, it changed the medium forever — and not necessarily for the better.
Compared to tools like Chartbeat or Google Analytics, which track users’ every click, scroll, and ctrl+w, the way radio ratings were determined 15 years ago — and are to this day outside of the largest markets — seems quaint. The ratings company Arbitron (now Nielsen) mailed paper diaries to a sample of listeners and asked them to write down what stations they listened to and when, then mail the diaries back.
“We knew many people were not writing down their listening in real time,” says radio consultant Fred Jacobs
. “Many people waited until the end of the day or even the end of the week to write it down.”
Though they may not have entirely reflected reality, the diary ratings were the only measurement available beyond the anecdotal (calls from listeners, response to advertisers). And they were a good indication of what people liked.
“People overstated their favorite station,” says Corey Lewis
, the station manager of WBUR in Boston. “They might draw a vertical line down the hourly diary chart and say they were listening for a longer contiguous time.”
Then, starting ten years ago, Arbitron changed its methods. Instead of diaries, sample listeners in the largest radio markets were sent Portable People Meters (PPMs) — pager-sized devices that picked up inaudible frequencies in radio broadcasts and kept a log of everything a person listened to throughout the day.
When the first PPM ratings came in, it was clear that some people hadn’t been filling out their diaries correctly.
“There was a big difference between what people were saying [with diaries] and how they felt about it and what their behavior was,” says Tamar Charney
, the managing editor of NPR One and the former program director of Michigan Radio.
“Some stations that were highly popular turned out not to be,” Jacobs says. “In some cases, stations that weren’t successful during the diary days looked better at PPM.”
Results varied by market, but generally, PPM ratings showed fragmented listening: People listened to more stations for less time. A person who wrote in their diary that they were tuned to public radio nonstop for their one-hour commute might instead be shown as listening for 12 minutes, then switching between Howard Stern, classic rock, and top 40 before switching back to NPR a half hour later. Radio programmers adapted to this, and they learned how to play to the PPMs.
“Talk got dialed back. To some degree, playing new music was seen as something of a liability,” Jacobs says. “It even had an effect on something as mundane as where stations played their commercials — the more commercial breaks you take, the lower your ratings are going to be.” That led to stations playing more commercials in fewer blocks during each hour.
“Certain formats disappeared,” says media consultant Mark Ramsay
. “There was a format called smooth jazz
. PPM killed it
. It was doing quite nicely prior to PPM and PPM killed it. There are formats PPM likes and some formats PPM doesn’t like.”
There were disputes about the accuracy of this data. Programmers questioned whether the PPMs registered signals on some formats accurately
. The meters detected when people visited stores or restaurants that were playing certain stations and considered that listening. And Ramsay points out that PPM sample sizes tend to be smaller and PPM listeners hold on to the devices for longer than diary listeners.
Despite questions over accuracy, the numbers were the numbers. And ratings drive business. Unfortunately, this was happening in 2007 and 2008, when business wasn’t looking good.
“The timing couldn’t have been worse,” Jacobs says. “Internet radio is coming on, satellite radio is expanding. There’s all this choice, all these interesting things happening outside of the AM/FM band, and at precisely that point, broadcast radio in the top 48 markets [where PPMs were introduced] is becoming a very gray-flannel-suit type of industry.” And, with a recession looming, “the economy was part of the perfect storm.”
Now podcasts are facing a similar giant leap in data, with a less-than-robust national economy and an easily disruptable media landscape. But that could be where the comparisons end.
For one, podcasting isn’t the financial behemoth radio was, so producers won’t live and die by ratings the same way. “I think we overreacted to it in the beginning,” Jacobs says. Not everyone has to just try the same tricks to maximize audience. Some programmers used the data to add more variety to their programming, or to put their most vital journalism in the places when they knew more people were listening.
“The reality is that great programmers have found a way to still make their stations sound really interesting and really great and really entertaining even though they still below the surface are playing the PPM game. Maybe in some ways the PPM rules end up being a crutch or an excuse to not have a particularly exciting station. The really great stations, the really great programmers, have found a way to make sure their stations sound vital.”
And some smart programmers learned to use the fact that PPM ratings reports come faster than diary reports to their advantage. “Diary to PPM may have actually increased experimentation,” Lewis says. “You could do it, then pivot quickly. In the old world, you didn’t get the data until two or three months later.”
Second, podcasters already have some of this data provided by other apps, like NPR One or Stitcher. And Midroll Media CEO Erik Diehn
isn’t too worried about Apple’s numbers. “I’m fairly confident that the number we see with regard to downloads does accurately reflect listener behavior,” he says.
Diehn also says that, while some listeners may be shown to be skipping ads, he knows not all listeners do. Ever hear a podcast ad that says something along the lines of “Go to this website and enter our offer code at checkout?” Well, those work. “The absolute best test that we have had of our listener numbers is the ongoing renewal and heavy interest from direct-response advertisers,” Diehn says.
The level of effort to even listen to a podcast shows a certain degree of intent, making download numbers similar to diary ratings in that they show what people like (or what they think they like).
Ultimately, there are a lot of ways people can use this new data. Charney says the metrics she sees with NPR One shows that, as with diary and PPM measurements, there is a discrepancy between what people say they like and what they actually listen to. But knowing that is valuable. If people remember they like a show, they can still be lured in.
“I think data and information is really good, but you also have to stay true to your editorial goals and mission,” Charney says. “We’ve been able to help producers develop podcasts that hold listeners. That just makes the product better for everybody.”
Not everyone will be responsible with the data, though. The more apt comparison to what’s about to happen with podcasting could end up being like what happened with the browser-based web when tools like Google Analytics and Chartbeat came along, or when Facebook became the primary source of traffic. Some people used the knowledge of visitors’ habits to make their writing more compelling and their sites more user-friendly. Others gamed their headlines, led their stories with slideshows and autoplay videos, and moved more ads above the fold.
Maybe some podcasters will find out listeners drop off after 10 minutes and cram all their ads into the beginning of the show. Or maybe new producers will come in and game the system some other way. It’s not clear what a clickbait podcast might sound like, but if there’s money to be had in making it, we’ll probably find out soon.
As for everyone else, one fact about the diary-to-PPM switch will stay true for podcasting once Apple releases its data: The listeners don’t change, they’ll just be measured differently.
Gabe Bullard is a senior producer of the public radio program 1A and a former Nieman Fellow.
Photo of an old Studebaker radio by Liz West
used under a Creative Commons license.