Monday, May 21, 2018


A new report from Edison Research concludes that radio station streaming is overstated [link]. 

Commercial and noncommercial stations have made significant investments to make their program streams available. But are audio streams worth it when few people hear it? 

According to Edison’s report, streaming radio audio reaches only around 8% of radio listeners – the remaining 92% hear radio the old-fashioned way, via over-the-air signals. Last week, Edison VP Randy Brown presented the data at the American Association of Public Opinion Researchers (AAPOR) annual conference in Denver [link]. AAPOR is a nonprofit trade organization that sets research standards for public polling and perceptual surveys.

To be clear, we are not talking about Nielsen Audio PPM ratings – we will get to that further down in this post. We are talking about private research, particularly Edison’s Share of Ear studies.

Edison found anomalies in their own methodology concerning online surveys that attempt to measure Internet behaviors. The problem stems from the difficulty of reaching a representative sample of light internet users and the 10% of Americans still have no online access. You can download Edison’s AAPOR report here. It provides a fascinating behind-the scenes look at the process Edison uses to compile their Share of Ear studies.

Edison says that listening to radio streams varies by the type of programming on the station. News and sports audio streams comprise 12% of the total listening, but only 6% for music stations.


Companies such as Edison track self-reported behavior and perceptual trends. Nielsen Audio tracks actual listening and hearing. We reviewed data from all of the markets surveyed by Nielsen PPM methodology in April 2018 and found very little listening to noncommercial station streams.

The chart on the left shows estimated weekly listeners to streaming station audio on all eight (8) listed noncom stations in April 2018. 

The pattern for commercial stations was very similar. 

Fact: There are very few people listening to station's audio streams.

Three of the eight noncom stations listed in April book fell below Nielsen’s minimum listening threshold. This standard is hardly arduous. To be included, a station must have received at least one (1) average-quarter-hour of listening and must have a Metro Cume rating of 0.495 or greater.


For additional perspective we asked Aaron Reed, Director of Information Technology & Engineering, for his thoughts regarding the Edison study. Reed writes frequently for publications and lists about changes in media technology. Reed feels Edison is on the right track but has serious flaws in its own methodology:

"Very broadly speaking, I think the overall point that there is a lot more listening happening via AM/FM than internet is probably correct.  However, I also think these studies' methodology is so flawed that you can't take them terribly seriously."

Aaron Reed
"If you read the full presentation Edison is pretty up-front about exactly how this data was gathered.  Personally, while cautioning that statistics is NOT my strong suit, the data gathering methods seem so inexact as to be a questionable usefulness at all.  Virtually everything relies on self-reporting in a diary by participants.  That's a recipe for terrible data as the human mind is very bad at accurate self-reporting."

"What immediately comes to my mind is how many people hopelessly confuse the difference between different content delivery systems.  "HD Radio" vs satellite radio (SiriusXM) is a good specific example, but more broadly: people have a lot of trouble distinguishing between which method within their smartphone they consume media through.  For example, they might remember listening to their iPhone vs listening to an FM broadcast...but they won't remember which app they used on the iPhone: a custom app by the radio station?  The iHeartRadio app?  TuneIn?  Youtube? Soundcloud?  iTunes?  These details matter to content distributors & creators but it's often too much to ask consumers to really remember them."

"To break this down a little: Nielsen says Rhode Island Public Radio's most recent AQH was around 2,300 if I recall correctly.  I just checked our webcast stats and a typical weekday swings between about 150 and 500 listeners, with the peak coming around 8am.  Call it a "weighted" average of 300 listeners since the peak time is relatively short.   300 is about 13% of 2300 so while that doesn't quite line up with Edison's assertion, it's pretty close.  Does that mean Edison is right?  No, not really.  Statistically we have an interesting coincidence but I don't think we have anywhere near enough confidence in the methodology to start making business decisions on this data."

"Heck, to start with we had to drink Nielsen's koolaid and trust their reports that our AQH was 2,300.  I don't really trust Nielsen's large part because they won't reveal any part of it and historically any service, product, or procedure that relies on a "trade secret" to be useful has almost always been proven to be false, if not fraudulent, once the "trade secret" is exposed.  And while Streamguys can provide firm statistics that X number of playback clients were connected to the server at a given time, that doesn't mean anyone was actually listening to the webcast at that point.  I don't mean it "on in the background", I mean someone could've fired up our app in the bedroom, listened for two minutes, then walked out to the kitchen to make breakfast but left the phone...still playing but nobody to the bedroom."

- Aaron


Tina Pamintuan
For over a year, KALW in San Francisco has been searching for a new General Manager to replace departing GM Matt Martin.  Last week the station announced they had hired Tina Pamintuan for the job.

Pamintuan worked at NPR in the late 1990s and early 2000s, first as an intern and then as an editorial and production staffer. While at NPR, she worked with Morning Edition’s Radio Expeditions and various assignments for NPR’s Cultural Desk.

In 2006 she became Director of Radio at City University of New York’s Graduate School of Journalism. Pamintuan has been a mentor in NPR’s Next Generation project since 2002. She has no previous experience in management at a radio station.

Pamintuan will be facing considerable challenges as GM of KALW. As we reported in March [link], KALW lags far behind KQED in audience size, the number of members and revenue raised from the community. In that story, we said KALW’s new GM while need to turn around the station’s sleepy corporate culture. The status quo is deeply ingrained at KALW and there may be resistance to change.

We wish Pamintuan well in her new gig and we hope she makes KALW shine.

1 comment:

  1. Thanks to Ken for publishing my commentary. I'd like to add something I thought of today well after I sent him my thoughts...

    While I have several well-known concerns about the Nielsen PPM system, I do want to give it credit. PPM is a strong attempt to solve the very problem I raised with Edison's methodology. It gets away from relying on self-reporting...which as any psychological researcher, pollster, or even readers of "Freakonomics", knows is highly problematic at yielding useful, accurate data...and tries to get more at the reality of what people are hearing.

    Sure, there's gonna be tradeoffs. People often are quick to point out that PPM only measures *exposure* whereas diaries (self-reporting) measures *retention*. That's not an idle difference, but I can live with that difference just fine. Exposure is valuable to know, too, and I think it's more important to get objective listening rather than the memory of listening.

    My real concern with Nielsen is that we're given no reason to have faith in their methodology. Okay, the PPM system itself is a little iffy, but we can gleam enough from the hardware to make educated assessments about its efficacy. And yeah, the sample size is arguably too small but while (as I said) statistics isn't my strong suit, I know that often you can accurately gleam trend information from a surprisingly small sample set. And Nielsen deserves a tremendous amount of credit for how much information they collect about the people in the sample set. The granularity of the data they have about those people holding the PPM's is astounding. Everything down to their preferred wines to drink.

    But what we DON'T know is how Nielsen "processes" the raw data collected by the PPM system and turns that into ratings books. And thanks to Voltair, we now know that there was clearly a PROBLEM with the system. Stations that installed Voltairs suddenly saw 40 to 60% increases in listenership practically overnight. (!!!!)

    Logically, that means either the ratings information was incorrect before Voltair, or it's incorrect after Voltair, or it was incorrect both before AND after Voltair, just in different ways. It has to be one of those three. But Nielsen has steadfastly asserted that the ratings information was correct BOTH before AND after Voltair. That logically cannot be true, and that's the source of much of my skepticism of Nielsen's assertions.

    And one more thing about the Edison research: it's important to remember that the broad trend might be that "radio" listening is still primarily through actual AM/FM broadcasts and not the web...but it can vary quite a bit depending on your format, your market, and your audience.

    For example, I think of KQED's "pledge-free webcast" which is a system that probably wouldn't have worked if weren't in the Bay Area. A place with very, very devoted public radio listeners AND listeners with access (and inclination to use) a lot of high-level technology well before the rest of the country. I would bet that public radio listeners in general are a little more willing and able (operationally and financially) to listen via non-AM/FM means, like smart speakers. I'm not sure of the details here; these are just guesses, but they're good things to keep in mind when digesting these studies and applying them to your own station's plans.