Radio InSights

Is Television Smarter Than Radio?

Are television people smarter than radio people?

Both television and radio face aggressive challenges from digital alternatives. Both media have seen a loss of audience as well as a decline in revenue.

The arc of the decline, however, differs between radio and television.

Over the past few years television has seen a greater decline in audience, but a smaller decline in revenue, which begs the question: Do television people understand the digital threat better than radio people?

The graph below encapsulates what’s been happening over the past few years. (Click to enlarge.)

We’ve indexed the latest numbers to comparable numbers in 2011. The current index for daily time spent for radio and television is shown on the left in red.

Radio stands at 95, which means that listeners are spending 5% less time listening to radio compared to 2011. Television stands at 90, which means television has lost 10% of viewing time over the same period.

In other words, television has lost twice as much “time spent” (think of it as consumption) as radio has.

Relatively speaking, good news for radio.

The right hand blue bars show the change in revenue over the same period of time. Radio indexes at 93 compared to television’s 96.

It means that while television has seen its audience decline twice as much as radio, radio has seen its revenue drop nearly twice as much as television.

We're only looking at the past four years, but television’s resiliency is nothing new.

It is a phenomenon that we've been discussing as far back as 2009 on Radio Insights. Here are two posts from 2011 here and here that illustrate that television’s defense against digital has been more effective than radio’s for quite some time.

What is television doing better than radio?

Both legacy media face growing competition from pure-play digital alternatives. Both are witnessing declines in audience.

And one can make the argument that local radio ought to be more nimble able to more quickly respond to digital threats.

Yet television, already a medium three times as large as radio has seemed to manage digital assaults better.

It’s an important issue for both media, so Radio Insights is going to explore this issue in greater detail in future posts.

We’ll identify differences between the two media’s responses to digital’s threats, and what it says about how best to minimize the negative impact of digital.

Stay tuned.

 

Nielsen: Let's Kill the Quarter-Hour

Little attention has been given to one important comment made by Arun Ramaswamy during the July 21st Nielsen PPM presentation. Here’s what he said:

The final point I’ll make is that the Voltair was more pronounced at the minute level than just the quarter-hour level....(Because of complimentary crediting and the smoothing system) we found that even though the Voltair product was more visible at the minute level, it wasn’t always translated at the quarter-hour level.

So even with Voltair, radio stations fail to get credit for fragmentary listening because of a quarter-hour tradition dating back to the 1940s.

In the 1940s and 1950s, radio programs were 15 minutes or longer in length. It made sense for radio ratings to be measured in 15 minute units, so the quarter-hour metric was created.

Most everything in radio has since changed, but not the quarter-hour.

The quarter-hour metric is an anachronistic throw-back to a era that radio has long passed.

We need to kill the quarter-hour. Here’s why.

Voltair enhances PPM codes so that decoders are more successful at identifying the station a panelist is listening to.

It takes at least five successfully identified minutes of listening for a station to get credit for a quarter-hour. If Voltair helps you get only four minutes credit in a quarter-hour, you don’t get a quarter-hour.

You get zero credit, despite the fact that the panelist was listening to you.

Nielsen argues that the editing rules help radio because you can get 15 minutes of credit even if the meter only successfully identifies five.

But they ignore the flip side.

A station gets no credit if the meter identifies even longer listening spans when it bridges two quarter-hours.

Say a panelist listens for five minutes from 5:00 to 5:05 and the meter successfully credits the station for all five minutes.

The station gets credit for one quarter-hour.

What if the panelist listens for eight minutes from 4:56 to 5:03?

The station gets no credit for any listening–-even if the meter credits the station for all eight minutes.

In diary markets this is hardly an issue. People are not that precise when they write down their listening, generally rounding to the quarter-hour.

PPM is different.

Because PPM measures exposure and the meters take time to identify a radio station (if at all), exposure spans are brief with gaps as the meter struggles to identify the station.

While a diary keeper is unlikely to write down 4:56-5:03, it is quite likely that meters will detect listening over odd periods of time like this.

That’s why Voltair has a bigger impact for Canadian stations than US stations. Canadian radio is measured in minutes rather than quarter-hours.

So Canadian stations get credited with listening that US stations do not.

Nielsen claims that PPM propelled radio audience measurement into the 21st century. Isn’t time to abandon the quarter-hour, an anachronistic vestige of the first half of the 20th?

 

Now We Know: Nielsen PPM Short-Changing Radio

Rumors of double digit rating gains using Voltair, a new processor, are fueling growing confusion over its implications.

Is it really true that a new processor can boost ratings, and by double digits?

And if true, what does it say about the accuracy of Nielsen PPM rating estimates?

Much of the confusion is due to a lack a data.

Stations using the processor are reluctant to even admit they’re using one let alone share ratings information.

But it’s pretty hard to hide.

Every month now we see growing numbers of stations with pretty suspicious rating spikes.  Are we to believe that all these stations just got smarter...and all at once?

And how many "best book ever" stories are we going to read in Nielsen’s monthly rating reports before somebody admits that maybe the box is behind many of these records?

Are we to believe that it's just coincidence that the success stories started appearing soon after Voltair's launch?

Given the lack of hard data, Harker Research decided to do something about it.

We set out to find stations about to install Voltair that would be willing to share ratings data with us so we could see for ourselves what impact Voltair could have.

What we are about to share is the first verified pre/post Voltair data, and what we found was just as amazing as all the rumors.

Two stations agreed to share their Voltair experiences with us.

We were able to track ratings as Voltair was installed. We were able to establish a pre-Voltair baseline of ratings, and then we were able to see what happened with the ratings after the launch.

We were also able to confirm that these markets were pretty quiet over this time, and that the two stations we followed changed nothing other than installing Voltair. These stations were also the first in their markets to get Voltair.

Like all other Voltair clients, the two stations insisted on anonymity, so we won’t be identifying the stations, markets, or formats. Both formats are highly competitive mainstream formats.

Station A was our first look at Voltair, so we looked at full week numbers.

For Station B we did something different. We looked at Voltair’s impact on morning drive. 

Station B plays very little music in morning drive, and it has been reported that PPM is particularly cruel to talk shows.

Since the encoding process does a poor job with spoken word, we wanted to see what impact Voltair would have on morning numbers. We’ll report what happened with morning drive, but we’ll also touch on full week numbers for Station B.

Ironically, Voltair increased both stations’ market share by more than twenty percent, but how that growth came about differed between the two stations.

First, a little background.

Voltair is designed to process station programming to create more consistent encoding. More consistent encoding increases the likelihood that decoders will be able to identify the stations to which panelists are  exposed.

So Voltair should have minimal impact on cume. All it takes is one successful code identification for a station to get credit for a cume listener.

The ratings benefit of Voltair should come primarily by increasing time-spent-listening (TSL).

If Voltair improves the encoding/decoding process, then listening normally missed by the meters will be picked up.

This might manifest itself as longer listening spans, but it could also increase TSL by increasing the number of episodes of listening, what Nielsen calls Occasions.

A 2011 analysis by Arbitron (since taken down) concluded that top performing stations in PPM didn’t have longer listening spans than weaker stations. Instead, leading stations had more Occasions of listening.

So to see the impact of Voltair, we began by looking at its impact on Occasions of listening.

As the graphs illustrate, both stations dramatically increased Occasions. Station A increased full week Occasions by 26%, and grew market share by a similar percentage.

Station B’s morning drive Occasions increased by a whopping 61%!

Full week TSL for Station A rose by 33%, but Station B’s full week TSL increased very little. Almost all the growth was in morning drive.

Station B saw its full week cume increase slightly more than Station A.

Station B’s talky morning show attracts a different audience from the rest of the day. We suspect that PPM was not catching the morning listeners, so the station was losing much more than just a little TSL.

Consequently, when the station started getting more credit in morning drive it gained Occasions, quarter-hours, and cume.

While this is the most definitive published analysis of Voltair to date, it comes with several qualifications.

We’ve only looked at two stations and two formats. We can say with confidence that the two stations’ gains were a direct result of Voltair.

However, as we have repeatedly emphasized, results will vary across formats and content.

The quality of PPM encoding depends on content. Some content encodes well, other content not so well. Voltair’s potential for improving a station's ratings ultimately depends on how poorly PPM is encoding the station's content.

Some stations will benefit more than others. Based on Station B’s morning gains, stations with talk formats should be particularly concerned about PPM’s "blind spots."

Some formats may not benefit at all, so don’t assume that your experience will be similar to either station.

The most important lesson here is that the rumors about Voltair helping improve ratings are true. This is the clearest evidence that PPM as implemented is flawed.

PPM does not capture all listening, so it under-estimates radio listening–potentially by a lot for some formats.

The solution is for Nielsen to admit that the problem exists and agree to a time-table to fix it.

It is likely that PPM will never be perfect. It will never capture all listening, but missing 20%, 50%, or more should be unacceptable.

 We invite additional Voltair users to confidentially share their experiences so that we can have a clearer picture of PPM and its weaknesses. Collectively we can finally get answers to PPM questions never asked as PPM rolled out.

Did Arbitron Mislead RAB About PPM's Missing Listeners?

Did Arbitron mislead the Radio Advertising Bureau about how successful PPM was in capturing listening?

You be the judge.

In 2004 the Radio Advertising Bureau investigating the technical aspects of PPM posed 188 questions (PDF) to Arbitron regarding PPM. Here’s one of them:

Question: What percentage of the transmitted codes is captured? What percentage with low volume?

Answer: In November of 2004 the PPM participated in a test where 33 radio stations were played in various environments and levels of “noise.” The PPM correctly identified 59% of the total radio sessions, in an environment where the goal was 50-70%.

The RAB seemed satisfied with the response. However, there’s no evidence that the RAB actually looked at the study Arbitron referenced.

And strictly speaking, Arbitron didn’t really answer the question. More on that after we look at the study.

The study was not conducted by Arbitron. The company played no role in it other than supplying the equipment.

Makes one wonder why the company didn’t cite its own research.

The study was funded by the U.K. radio community. We first reported on the study in 2010 because it was (and remains) the only publicly released study on PPM.

We personally spoke to the lead researcher and concluded the study was thorough, objective, and we believe well designed.

The investigators created over thirteen hundred different scenarios, testing various combinations of formats, radio listening levels, and interfering ambient noise levels.

They then divided the scenarios into categories based on how easily the meters should be able to identify the stations.

As the summary at the left indicates, PPM’s performance was mixed.

In the 28 cases where conditions were nearly perfect with the radio playing loudly in a quiet room where the meter was exposed and close to the radio, PPM got it right every time.

However, as soon as conditions dropped from perfect, success rates fell rapidly.

For example, the meter was able to identify less than three-quarters (73.4%) of the stations in the next two easiest scenarios.

At level 4, a level at which the researchers thought the meter should be able to identify the station, PPM only got 151 stations right out of 266 tries, a little better than half (56.8%) of the trials.

PPM identified only one quarter (24.0%) of the stations played in the most difficult three scenarios. These were where the stations were played at a low level in a noisy environment where the meter was tucked away in a bag or jacket.

So the response to RAB that PPM correctly identified 59% of the radio stations is somewhat misleading. In the most natural scenarios where real panelists are listening to the radio, the success rate could be considerably lower.

(It should be noted that in a comparison PPM alternatives proved no better, so PPM is not alone in failing to ID stations.)

But there’s a bigger issue here.

The only criterion used in the study was identifying the station. The sessions were ten minutes long, so the meter had ample time is identify the stations.

The Question “What percentage of transmitted codes is captured?” goes well beyond identifying a station.

The RAB was not asking what percentage of stations are identified. The question is about how often the decoding process works. And the longer it works, the more quarter-hours.

Arbitron cleverly dodged the question.

So based on Arbitron’s admission to RAB that PPM captures 59% of radio station IDs over a ten minute period, that means 41% of stations are unlikely to be identified in a ten minute period.

Let’s just round that to a quarter-hour.

Isn’t it curious that Arbitron/Nielsen report that average listening spans for all formats are ten minutes?

Is it because nearly half of stations remain unidentified after ten minutes? And is this why Nielsen cross-tuning reports show that most listeners don’t switch stations, they just disappear?

The bottom line is that we still don’t know how much listening PPM is missing. Looking more closely at the study Arbitron quoted to the RAB, it could be a lot.

We don’t think the Voltair controversy is going away any time soon.

 

Now Media Rating Council to Dictate Audio Processing?

The Voltair saga has now entered the Twilight Zone.

If you’ve followed developments, you know that 25-Seven has developed an audio processor that strengthens weak embedded PPM watermarks, the things that tell PPM that a panelist is listening to your radio station.

You’d think that is great news for radio stations and great news for radio in general.

We know from independent studies that PPM is not perfect, that PPM misses some listening. Any thing that moves PPM from (say) 70-80% accuracy to closer to 100% accuracy ought to get an honor from the NAB or maybe even Nielsen.

You’d think.

Nielsen has had one of the devices since last fall for evaluation. That’s about the same time that stations starting putting the units on the air.

Nielsen is still “thoroughly” examining the unit.

Radio stations didn't need a long look. Despite a quiet roll-out, word got around. It worked.

There are now something like 300 stations using the device, although few publicly admit it.

We’ve seen anywhere from single digit growth to over 50% gains in TSL. To our knowledge, not a single station has returned the unit.

Good news, right? Not to Nielsen.

Not only has Nielsen reserved judgement, they’ve decided that they want the Media Rating Council (MRC) to bless Voltair.

MRC deliberations make a mafia get-together look like a city council meeting. Deliberations are secret, criteria for granting accreditation are kept secret, and their decisions are absolutely final.

We suspect the MRC is modeled after the 13th century Vatican without the Inquisition.

In an Inside Radio interview MRC CEO George Ivie left his Cone of Silence long enough to comment on Voltair noting:

In general, changes need to be improvements in encoding or code-capture efficacy. We are very unlikely to approve degradations in these processes.

To which we can only add WTF???

The PPM codes are simply audio tones embedded in a radio station’s audio. Every station has an audio chain that includes processing. Every station.

And those audio chains vary in complexity and sophistication.

Radio stations have been trying to process their audio to “game” PPM since day one, mostly unsuccessfully. If Mr. Ivie is concerned about degradation, the MRC should grant accreditation to audio chains and Nielsen should authorize every encoder installation.

25-Seven isn’t the first to try to fix an imperfect encoding/decoding process. They just appear to have figured it out better than any before them.

For Nielsen and now the MRC to make thinly veiled threats suggesting that Voltair may potentially be somehow degrading the process is hypocritical silliness.

If anything, the processor appears to be making PPM encoding more robust. It does nothing more than help stations get credit for listening that they might have otherwise lost.

We suspect that the PPM accreditation process did not test encoding/decoding robustness with each of the dozens of formats, and the multitude of environments in which people listen to the radio, let alone with the hundreds of processors and thousands of processing chains in use today.

To single out Voltair for attention is the wrong thing on which Mr. Ivie should be focused. If he really is concerned about PPM code-capture efficacy, he should reopen the accreditation process of PPM.

He should ask Nielsen how it is that hundreds of radio stations that install Voltair get better numbers. What does that say about the intrinsic robustness of PPM?

 

Remember the Promise of Engagement?

 
Remember Arbitron’s promise of developing an Engagement metric for radio?

As Radio Insights reported in 2009, Arbitron intended to:

Develop an ongoing measure of "affinity" that is designed to reflect the advertising value of the unique relationship listeners have with their stations. The mission of this (effort) is to design a relevant, timely and accessible metric that captures audience involvement.

At the time we noted that the company’s new found interest in measuring affinity was ironic because Arbitron had essentially stopped measuring affinity when it switched to PPM.

A year after the announcement without much explanation the idea was dead. There would be no engagement metric–at least from Arbitron.

We speculated that Arbitron may have killed the idea because engagement measures an active listener wilfully choosing one station over another.

PPM does not measures listening. It measures only “drive-by” exposure. Arbitron could not repurpose PPM ratings to create a measure of engagement.

Ironically, while PPM cannot tell us anything about engagement, the diary does.

It is a good instrument to measure engagement because the listener herself writes down the stations she listens to, and how often. If she is really engaged with a station, she is going to write it down a lot.

Now five years later, another company is rolling out an engagement metric, but not for radio. For web advertising.

And the irony is particularly telling given that the Media Rating Council (MRC) has already accredited the engagement metric. It’s from a company you’ve probably never heard of, Chartbeat.

As reported by Gigaom:

Web analytics firm Chartbeat says it is the first to be certified by the Media Ratings (sic) Council for a new way of measuring the actual attention of readers, as part of a move to get publishers and advertisers to stop focusing only on clicks and pageviews.

Chartbeat measures the amount of time a viewer spends reading online content arguing that the time spent with content is more important than things like unique viewers.

As we pointed out in the 2009 post:

We at Harker Research have found that there is a strong correlation between high Time Spent Listening (TSL) and affinity. As one goes up, the other one does too. If a listener has a natural attraction towards a radio station, she probably listens to it a great deal. She listens less to a station that she feels less strongly about.

Read the prediction of one digital ad analytics firm as reported by AdAge:

With the viewability standard in place, creative agencies should expect the ground to continue shifting toward attention metrics -- whether cost per hour, cost per 30 seconds or something else. "I don't think it will be one single metric," said Jonah Goodhart, CEO of Moat, a digital ad analytics firm. "We will see a shift on transacting on viewability in late 2014 or early 2015," he added. "This idea of buying media on the metric that makes the most sense for you is where we'll see the world going."

Yet at a time when advertisers are embracing engagement and turning away from metrics of reach and unique viewers, radio is heading in the opposite direction, using a measurement that emphasizes reach over engagement.

PPM “atomizes” radio reducing radio listening into homogenous ten minute listening spans preventing radio from capitalizing on its unique ability to engage.

Maybe Nielsen should take another look at engagement. Maybe Arbitron was on to something in 2009, but didn’t have the right tools to develop the metric.

And radio could sure use some engagement.

Fitbits, PPM, and Arbitron's Missing Quarter-Hours

Want to better understand why PPM listening levels are lower than diary levels?

Want to see why Arbitron only asks that panelists carry their meter eight hours a day while most people are awake twice as many hours?

Want to see why PPM may be costing radio million dollars in lost revenue?

If you really want to better understand PPM’s limitations and challenges, there's a new tool you must try.

Arbitron won’t let you experience what it’s like to be a PPM panelist. The company won't even let you carry a meter for a day or two.

The good news is that the new Quantified Self movement has created an opportunity for us all to feel what it is like to be a PPM panelist.

And you won’t like what you find out.

The Quantified Self is a movement that uses new technology in portable data acquisition to document a person’s daily activity.

One of the technologies is a new category of small fitness tracker toys devices that use an accelerometer to measure our movement throughout the day. The PPM uses a similar accelerometer to determine whether a PPM is in motion.

Devices like the Jawbone fitness tracker (right) are worn on the wrist like a bracelet, while others are small clip-on devices like the Fitbit Zip.

The Fitbit Zip is particularly interesting because in many ways it can serve as a valuable surrogate for the PPM meter.

Wearing a Fitbit Zip is just like a panelist carrying a PPM meter. And the Zip will tell you how many hours in a day a PPM meter would have been able to capture listening.

A few months ago I joined the Quantified Self moment and began wearing both a Jawbone Up and Fitbit Zip. Little did I know that I would soon develop great empathy for Arbitron’s panelists.

In the beginning I made a conscious effort to wear both the Jawbone Up and the Fitbit Zip continually.

The Jawbone slips easily onto the wrist and I quickly forgot about it. Were Arbitron’s PPM meter a bracelet, it would have captured all of my radio exposure.

The Fitbit was another matter.

First, it required some effort to securely clip it to my clothing. While it is considerably smaller than Arbitron’s PPM meter, I found it inconvenient.

When I changed clothes I had to make sure I also transferred the Fitbit Zip.

As time passed, I found myself increasingly forgetting the Fitbit, leaving it in a pocket, or forgetting it was clipped to yesterday’s pair of pants.

The Fitbit Zip was capturing less and less of my activity.

Were the Fitbit a PPM meter, it would have captured less and less of my exposure.

Mornings were particularly troublesome. I kept the Fitbit Zip on the night-stand, where a PPM docking station would possibly sit.

Clipping the Fitbit Zip to my clothes was generally the last thing I did as I got dressed, which means it rarely logged my morning activity.

The graph shown above illustrates this. The graph shows my activity in quarter-hours. While I get up around 7:00 A.M., I didn't get around to attaching the Fitbit to my clothes until nearly 8:00 A.M.

Now think about a PPM panelist.

A panelist is just like a person carrying a Fitbit Zip.

It takes a conscientious effort to make sure the meter is with him/her. It is likely that from time to time the panelist is going to forget to carry it.

Arbitron must know this. How else can we explain Arbitron’s rather low bar for counting a panelist?

A panelist is required to carry his/her meter no more than eight hours a day. Just eight hours (and even less for children).

The average adult is awake 16-17 hours a day, which means that a panelist is only required to carry the meter half the day to stay in Arbitron’s good graces.

How can a meter capture all radio exposure if it is sitting at home up to half the day?

One big difference between the diary and PPM is that listening begins later in the morning for PPM. It's like PPM panelists get up later than diary keepers.

Could it be that panelists don't bother to clip-on their pager meter until they are well into their morning?

In my case, the Fitbit missed about four quarter-hours of activity in the morning before I got fully dressed.

Coincidence?

A Fitbit Zip is $60. Even if you don’t want to join the Quantified Self movement, buy one to see what a PPM panelist goes through.

Then you’ll know where all those quarter-hours went, and why PPM may be costing radio millions of dollars.

Smart-Phone: Radio Salvation or Another Nail in Coffin?

Over half of cell phones sold in the US are now smart-phones like iPhones, Androids, and Blackberry.

If the trend continues, one day virtually everyone will be able to listen to radio where ever they go.

While some have heralded this development as a huge opportunity for radio, the growing use of smart mobile devices may be at best a mixed blessing.

The reason is that compared to listening on a radio or computer, mobile listening-spans are dramatically shorter. And shorter listening-spans makes radio time harder to sell.

Even Pandora is not immune to the impact.

While Pandora regularly crows about all sorts of new records like number of subscribers, or session starts, the service never mentions TSL, the amount of time users spend with the service.

Maybe that’s because while Triton Digital ratings show average sessions and session starts growing, TSL, the average length of time a person listens to Pandora, is headed in the opposite direction.

The graph above traces Pandora’s TSL for the past three years.

In 2009 Pandora TSL was about an hour. At the time only computer listening was included in the ratings.

Then in December 2009 mobile listening was added to the service’s ratings for the first time.

Pandora TSL immediately started dropping. In just four months TSL plummeted to just 46 minutes.

Aside from a few positive months in late 2010, the service’s TSL has continued its steady erosion throughout 2011, even accelerating towards the end of the year.

Today Pandora TSL is down to 38 minutes, a loss of 34%.

The precipitous decline in TSL coincides with a dramatic increase in the number of users listening on smart-phones and iPads.

 As Business Insider points out:

While 88 percent of Pandora's listening hours came from desktop use in Q1 of 2010, by 2012, users spend (sic) 70 percent of their time on Pandora listening on their mobile devices.

On a computer one can listen to their favorite station as they continue to do other work on the computer.

Multitasking on a mobile device is much more problematic than on a computer. And if one is using the phone to make and answer phone calls, listening to radio for any length of time isn’t going to be practical.

So it stands to reason that TSL will decline as more people listen on mobile devices.

Is TSL for broadcast streams following the same pattern? Sure is.

Harker Research tracks broadcast streams measured by Triton Digital, and broadcast streams are suffering a similar decline.

The second graph above reflects the combined TSL of 12 broadcast groups for which we have multi-year trends.¹ In the beginning of 2010 the combined TSL was over 35 hours, nearly three hours per radio group.

By 2011, the combined TSL was down to 24 hours, and today it is down to 16 hours, a decline of over 50%, a decline even greater than Pandora’s.

And what about Bob Pittman’s Clear Channel Communications, leading the digital charge with its iHeartRadio app?

As measured by Triton Digital, Clear Channel is the worst performing streaming service in terms of TSL, and one of the worse performing in terms of its decline.

The third graph tells the story.

Two years ago Clear Channel average sessions consistently lasted over an hour, some months exceeding 82 minutes.

As of August, TSL had fallen to 32 minutes, a decline of 61%.

If the time spent with mobile devices continues to grow at the expense of traditional computer use, we can expect TSL to continue to decline.

And that could cost radio a lot.

For Pandora who started with zero listeners and zero dollars in sales, the growth in audience more than makes up for the loss of TSL per user.

A loss of TSL for broadcast radio can’t help but hurt revenue.

Radio already reaches more than 90% of Americans. There is no more room to grow cume audience.

To grow AQH and revenue, broadcast radio has to increase TSL. And TSL is headed in the wrong direction.

Radio took a financial hit when AQH ratings dropped in the markets that switched to PPM.

Streaming is likely to be another financial hit as AQH ratings decline even further as people replace their radios with mobile devices.

Think about that next time you hear a promo for iHeartRadio.

¹The 12 groups are Cox, Cumulus, EMF, Emmis, Entercom, ESPN, Greater Media, Hubbard, Radio One, Salem, Townsquare, Univision. Neither CBS nor Clear Channel are included because of their non-broadcast streams.

Beating PPM Easier Than You Think

Which strategy is better: One that has an 87% chance of growing your share, or one that might have an 8% chance?

That’s an easy call. You’d pick the 87%, right?

Then why are programmers focusing on the 8%?

Arbitron looked at PPM ratings across all 48 markets and showed that Cume is far more important than TSL, roughly ten times more important in gaining share.

Yet all the focus has been on TSL, trying to get people to listen more often.

It doesn’t work. Arbitron proved it. Harker Research confirmed it.

Our recent post The Myth of PPM Occasions showed that all stations have pretty much the same Time per Occasion, and that numbers of Occasions is a lousy predictor of share.

Some poorly ranked stations have plenty of Occasions, and some successful stations have relatively few Occasions.

With PPM, the key to success is Cume.

But don’t take our word for it. Arbitron found the same thing.

Take a look at this table from Arbitron’s Key Indicators of Hightly Rated PPM Stations (PDF).

It compares 25-54 metrics across all 48 PPM markets for average stations, the top three ranked stations and number 1 stations.

Notice that top stations have over five times the share as average stations, but TSL is only 30% higher. The real difference is Cume.

Top stations have over five times as many Cume listeners as average stations.

Five times the Cume listeners. Five times the share. Coincidence?

Comparing average stations to #1 stations isn’t very useful, however. The average includes a lot of stations that have no chance of challenging market leaders.

We think the comparison between top 3 stations and #1 stations is more useful, because to challenge the top dog, you have to be within striking distance.

Notice that there is no difference in TSL between challengers and #1 stations. None. Both Time Spent per Occasion and Daily Occasions are virtually identical.

The only difference is Cume. Top Dogs have 16% more listeners than the challengers.

The report is from 2011, but Arbitron continues to tout Occasions despite little evidence that Occasions matter.

At the latest Conclave Arbitron updated its analysis, once again comparing average, top three, and #1 stations.

And once again Arbitron’s data confirmed the relative unimportance of TSL in determining PPM ratings. Number one stations had 9% more Occasions (a whopping half Occasion per day) over average stations.

The far greater difference was in Daily Cume, where #1 stations had 70% more listeners.

But the notion that the key to winning is increased Occasions still won’t die. Reflecting on the latest data, one pundit noted:

The updated study reinforced that "how often" has replaced "how long" in driving TSL.

As if either matters when it comes to PPM.

Unnoticed was the fact that Occasions for #1 stations declined in this year’s analysis. In 2011 top stations had 5.3 Daily Occasions. Today #1 stations have only 4.7 Occasions, an 11% decline.

So a year after Arbitron declared that:

Getting people to listen longer was Old School. New School is to get people to listen more often.

And radio excitedly responded with a flurry of efforts to grow appointment listening, Occasions for top ranked stations in PPM markets actually declined.

So much for that idea.

Do you want to gain share? Do you want to beat your competitor? The only way to do it is to grow your audience. Focus on Cume.

The Myth of PPM Occasions

One of the most entrenched ideas on how to win at PPM is the belief that increasing Occasions, the number of times a person listens to your station, makes a difference.

But is it really true? Is the key to growing share increased appointment listening?

The notion that Occasions is the key to winning at PPM is based on a well-intended, but ultimately misguided, Arbitron analysis of PPM called Key Indicators of Highly Rated PPM Stations (PDF).

Asserting that No One Wants to Be Average, Arbitron compared average station metrics to highly rated stations and declared:

Getting people to listen longer was Old School. New School is to get people to listen more often.

Thus was born the oft repeated (but never tested) mystic power of Occasions and appointment listening.

We recently collaborated with a skeptical client who wanted to better understand the relationship between occasions, listening spans, and share.

The Arbitron study looked at formats across markets, but our client just wanted to look at their own market.

After all, you compete against the other stations in your market, not similarly formatted stations in the other 47 PPM markets.

We looked at the relationship between share and three measures of listening: Occasions, Time per Occasion, and daily cume.

Three graphs shown here tell the story.

The first graph shows the relationship between share and listening spans, Time per Occasion (TpO). Each square represents a station.

Share is shown on the vertical axis. The larger the station’s share, the higher the square appears on the graph.

Time per Occasion is shown along the horizontal axis. Stations shown towards the right have higher TpO, while stations towards the left have lower TpO. The actual numbers are irrelevant, it is the direction of the numbers that matter.

Notice that nearly all the stations are stacked on two piles. The two piles represent 9 and 10 minute listening spans.

This means that nearly all of the stations in this market have virtually identical listening spans.

Arbitron found the same thing in their national study.

Arbitron found that the average listening spans for pretty much every station in every format is the same, either 9 or 10 minutes each occasion.

This curious identical listening span pattern has been repeatedly reported by Arbitron, yet it has generated virtually no curiosity, no discussion.

How can it be that the listening spans of every format from Classic Rock and AC to CHR, Country, and All News are all within a minute of each other?

Is this reasonable? Is it possible that virtually every format has the same listening span?

We’ll save this topic for another time, but give that some thought.

In any case, if it is true that listeners to every station in every format across all PPM markets listen the same length of time, the only way to increase TSL is by increasing Occasions.

Using that logic, we should find that higher ranked stations have more Occasions.

Take a look at the second graph. Based on Arbitron’s assertion, we should see stations arranged in a sort of diagonal cloud moving from lower left to upper right.

The stations with smaller shares would have fewer occasions and would appear in the lower left, while stations with larger shares would have more occasions and appear in the upper right.
 
We don’t see that. Stations are all over the place.

Some poorly rated stations have many more Occasions than much higher rated station. And some highly rated stations have very few Occasions.

Arbitron’s assertion that winning stations have more Occasions turns out to be untrue when we look at stations across formats in a specific market.

So much for appointment radio.

Then how do you grow share? What’s the secret? The answer can be found in the third graph.

The graph shows the relationship between share and daily cume. As with the other two graphs, share is shown vertically, while cume is shown along the bottom.

Notice that unlike the other two graphs, the cloud of stations cluster around a diagonal line. Small cume stations also have small shares, and stations with larger cumes have larger shares.

Things get a little strange towards the top, but in this market ethnic FMs and heritage AM stations distort the relationship at the upper end.

So here’s the real deal: With PPM, the only reliable sure-fire way to grow share is to grow cume.

To Arbitron’s credit, their analysis does mention daily cume as a key driver, but somehow that got buried in the excitement over Occasions.

Rather than a New School revelation about how to win at PPM, the answer to growing share turns out to be decidedly Old School.

To grow share in PPM, you have to grow your audience. You need more listeners!

Keep that in mind when you start budgeting for 2013 and corporate red-lines your marketing budget...again.

Pandora & Arbitron: Co-Conspirators in a Lie

Joe Kennedy, CEO of Pandora, recently bragged that virtually all of Pandora’s growth came at the expense of AM/FM listening.

He reasoned that since listening to CDs, iPods, and MP3 players had increased, Pandora listening had to come from broadcast radio.

Pandora has a propensity to twist facts in the company’s favor that verges on the pathological. However in this case, the key twist should be laid at Arbitron’s feet.

Kennedy pointed to an apparent 5% decline in TSL between Arbitron’s Radio Today 2010 and 2011 reports as proof of his claim.

The problem is that it’s a lie.

Arbitron started changing the way radio is measured in 2008, and the roll-out continued through 2010.

During this period, Arbitron was forced to mingle audience data from two very different methodologies that produce very different estimates.

National estimates during this period are essentially statistical gibberish.

Arbitron pretty much admits it, declaring: Different measurement methodologies can and do produce different results.

Arbitron analyses at the time showed that TSL drops as much as 30% in markets that switch to PPM.

The first TSL number Kennedy cites is based on mostly diary estimates while the second number is based mostly on PPM estimates. He is comparing apples and oranges.

The fact that TSL declined only 5% during radio's transition to PPM suggests just the opposite of what Kennedy claims. 

His assertion that Pandora's gains came at broadcast radio's expense is just wishful thinking. 

If anything, research shows just the opposite. Pandora's growth seems to have little impact on local radio listening. Online listening appears to be additive radio consumption, not a replacement for broadcast radio listening.

Kennedy's claim that time spent with other audio entertainment is increasing is also demonstrably false, but we’ll save that discussion for another time.

We noted in 2009 that the transition to PPM couldn’t come at a worse time because new-media stakeholders would use the apparent losses as “proof” that broadcast radio is in decline.

Since then, our prediction has become all too true.

Apple iPad: Toy or Listening Tool?

Is the iPad a major advancement in computing, or is it just an upscale teckie trophy, something to be seen using?

The pace of sales would suggest the former. Apple is on pace to sell 30 million iPads this year. The rate of growth exceeds anything else Apple has launched.

But new app research raises some interesting questions.

Localytics looked at the length of time users spend with different apps, and the number of times they use them. Graphs of the results are shown at the left. Click to enlarge.

Users spend the most time with news and music apps.

 

Unfortunately, the results are expressed as a percentage of the average time spent with apps. Users spend 228% of the average with music apps.

 

We suspect that Localytics doesn’t share the actual time because it is quite short.

What’s more interesting is the second graph showing the number of times people use an app. Games lead at 13 times a month followed by news and music tied at 5 times.

So an iPad user accesses her music app about once a week, and spends about two and a quarter times more time than the average app–-whatever that number is.

 

 

What does accessing a music app just once a week say about music apps?

 

Perhaps the iPad is more a toy than a really useful tool. Maybe users access their music with their iPhone or computer mostly and the iPad only occasionally.

Or maybe the app is the toy, just something to play with occasionally.

Either way, the study raises more questions than answers. But they are interesting questions.

Want to Grow? You Need a Bigger Tent.

In a recent analysis of PPM ratings, we determined that cume has a bigger impact on share than either TSL or occasions. The natural follow-up question is the relationship between cume and TSL/occasions.

Is it true that cume and TSL are inversely related, that stations can't have a large cume and high TSL? More importantly, can a station grow both cume and TSL at the same time, or does one go down as the other goes up?

It’s a much more complex issue than you might think.

Returning to the rating data-set we used in the previous posts, we  plotted cume against TSL. (A graph of cume versus occasions would look very similar.)

 
There’s a clear inverse relationship between cume and TSL. Stations with the largest cumes tend to have the lowest TSL, while stations with lots of listening tend to have smaller cumes.

The key word, however, is tend. While the relationship tends to be inverse, it isn’t absolute. There are stations that have both above-average cume and above-average TSL.

And the chart doesn’t tell the entire story.

In long-term trend analyses, Harker Research has also found that growing stations tend to increase cume and TSL at the same time.

In fact, we have found that a good way to differentiate real growth from the more common month to month random fluctuations is to see whether both cume and TSL are growing.

If you see a spike in TSL without a spike in cume (or vice versa), it is probably a wobble. If you see both TSL and cume moving up, it is more likely to be real.

This is also true with downward trends, but more often with a lag between the two measures.

It stands to reason that TSL and cume would both grow for stations on the rise. The same elements that attract new listeners to a station also encourage existing listeners to tune in more often.

More occasions equal higher TSL. Consequently, cume and TSL grow concurrently.

One Radio Insights reader questioned whether external marketing was the only way to grow cume. It isn’t the only way, but it is an efficient way.

In diary markets, it is particularly difficult to grow an audience without external marketing.

A person typically listens to about five stations, but only writes down two or three stations in the diary. The missing stations make up a station’s phantom cume.

The goal of external marketing is to raise a station from phantom status to reported status by raising the station's top-of-mind awareness. You really haven’t gained a listener. You’ve gained a reported listener.

With PPM, listeners are replaced by exposed panelists. The meters pick up “drive-by” exposure, a few minutes at a time. That’s why cumes are higher with PPM.

External marketing is just as important in a PPM market, but it’s goals are different.

With PPM, every listener is essentially a phantom listener. You never know whether a panelist really listened to the station, or simply came in aural contact with it.

External marketing still works by raising top-of-mind awareness, but less directly.

While the goal in a diary market is to directly influence the diary keeper’s actions, with PPM you don’t care whether the external advertising works on the panelist or those around her.

The goal of external advertising in a PPM market is to grow the tent. The bigger the tent, the more people it can hold.

The more people in the tent, the more likely somebody in the tent will be listening to the station and the greater chance that a panelist will be exposed.

This difference between goals means that while external marketing is important in both diary and PPM markets, the messages need to be different.

That will be the subject of a future post.

So is PPM All About Occasions, TSL, Cume, or What?

Arbitron claims that highly successful radio stations share similar characteristics that distinguish them from average stations.

Analyzing top stations in 33 PPM markets, Arbitron concluded that TSL didn’t differ between top performers and average stations, but that top performers had greater numbers of sessions (occasions) than average stations.

The analysis and conclusions were widely reported, and the notion that the key to impacting PPM is increased occasions entered PPM folklore.

Unfortunately, the analysis suffered from several problems raising doubts about the conclusions. A closer look at the relationships between share and the other rating variables suggests that occasions are not as important as Arbitron claims.

We believe that several technical mistakes led to the misguided conclusions. For example, Arbitron simply averaged share, cume, TSL, and occasions. Averaging can be misleading because averages are distorted by outliers, the few stations at the extreme top and bottom of any ranker.

What is the average share for stations in this hypothetical four station market?

Station A     9.7
Station B     2.9
Station C     2.5
Station D     2.0

The arithmetic answer is 4.3, but does a four share average really capture the essence of the market?

The “average” station is in the twos. Yes, there’s a ten share station, but this one “top performer” station has pulled up the market average by nearly two shares.

Understanding how the two share station can beat the 2.9 station is far more valuable than understanding the ten share station.
 
Because many markets have the equivalent of our hypothetical ten share station, rating analyses across markets should use what’s called the median, the station in the middle, rather than the average. In our example, the median would be 2.7, a more realistic view of the “average” station.
 
A second problem with the analysis is the number of markets and stations involved. If one is going to aggregate ratings data for disparate sized markets, the data needs to be “pretreated” to equalize the impact of each market.

To better understand the role of occasions in impacting PPM, and illustrate the flaws in the Arbitron analysis, Harker Research examined a large Midwest market. We began by graphing the relationships between share and cume, TSL, and occasions for each of 30 stations with a 1+ share.

Share is the vertical axis, and the horizontal axis is one of the other three measures. Each station is shown as a marker.

 
 The first graph is cume. Note that there is a strong trend within the cloud of stations. As cume increases, so does share. It isn't a perfect relationship, but a very strong one.

 
The second graph is share charted against weekly TSL. Note that except for a couple of outlying stations, the station cloud is more or less round. This indicates a very weak relationshup between share and TSL.

The circular pattern indicates that TSL is far less predictive than cume of ratings success.There are stations with four shares that have the same TSL as stations with two shares.

 
The third graph is share charted against weekly occasions. Note that the pattern is very similar to the TSL graph. There is no directional pattern to the cloud. Stations with a wide range of share have the same number of occasions.

So neither TSL nor occasions is very predictive of PPM success.

These graphs show what’s called correlation, how share changes as the other measures change. Correlation can be expressed as a percentage, an estimation how much a change in cume, TSL, or occasions impacts share.

Cume explains about 68% of the change of share compared to 25% for TSL, and 21% for occasions.

In other words, occasion is the least predictive measure. It has an impact on share, of course, but increasing occasions has the smallest relative impact on share.

On top of that, there’s no practical difference between TSL and occasions when it comes to impacting share.

 
Take a look at the fourth graph. This shows the relationship between weekly TSL shown on the vertical axis and weekly occasions on the horizontal axis. It is almost a straight line.

What this says is that both total TSL and occasions move in the same direction at the same rate. With PPM they are inseparably intertwined.

Arbitron’s notion that TSL is not the key, but occasions is, is demonstrably false.

At the weekly level (which is what matters most in the ratings), there’s no distinction between TSL and occasions. Both impact ratings equally, and neither have nearly the impact of cume.

Arbitron used daily estimates to reach its conclusions and looked at a narrow demo. The graphs and conclusions presented here are based on weekly all person estimates. This is another reason we believe our analysis is more generally applicable than Arbitron's.

We believe that analyzing daily estimates of a narrow demo involves too few panelists to be conclusive. The difference between the average station and top performers was a one single occasion. 

The bottom line is that cume is by far the most predictive measure of ratings success. If you increase your cume, you have a very good chance of increasing your share. Not so much with TSL or occasions.

Isn’t it ironic that the single most important tool to grow an audience, external marketing to grow cume, was the first thing cut when radio got in financial trouble?

Robert De Niro, Deer Hunter, The Key to PPM Success.

George Orwell declared History is written by the winners. He might as well have been talking about radio.

When’s the last time somebody interviewed the program director of a station that just tanked in the ratings, asking about the keys to the station’s performance?

Radio isn’t interested in losers, only winners, so the stations that get all the press are the ones that are winning.

With over 2000 radio stations competing in PPM markets, each month hundreds of stations will have sizable gains and a handful will have spectacular gains.

And every month, the winners will get all the attention and speculation about why the stations did so well.

The problem is that nobody can really know why a station has a great month. Maybe it’s something the station did. Maybe it’s something the station didn’t do. Maybe it was something some other station did.

In reality, most of the gains from month to month are just by chance, but human nature and radio being what they are, the idea that most rating gains happen purely by chance is too unsettling.

Despite evidence that rating changes from month to month are purely random, radio pays close attention to the explanations of program directors, general managers and consultants in the belief that somebody must know the secret.

Let’s say a station does something like (say) commercial-free hours, has a good month and it makes the news. The following month other stations try commercial-free hours, and some of those stations also have a good month.

Seeing the number of stations doing commercial-free hours grow, the next month other stations try the same thing, and some of them have a good month.

Does that prove that commercial-free hours are the key to winning in PPM?

Not unless the number of commercial-free winners far out-number the commercial-free stations that were flat or went down.

It is called survivorship bias. We tend to only notice the winners. There may be just as many losers as there are winners doing the same thing, but we ignore the losers and pay attention to the winners.

Think about the Russian roulette scene in Deer Hunter. Robert De Niro’s character spins the gun cylinder, pulls the trigger, and lives. Does that mean De Niro has a special gift?

What if a hundred people decide to do the same thing?

One hundred people spin the cylinder, pull the trigger, and only ten people shoot themselves. That means 90 people are winners. They will be heralded and interviewed about the secret formula for winning.

The 90 survivors spin the cylinder again, 9 shoot themselves, but that still leaves 81 winners prepared to offer their key to staying alive.

You can see where this is going.

Most people win at Russian roulette, but who wins is determined purely by chance. There isn’t a secret formula for winning (except not playing).

Every month when the Arb numbers roll there will be plenty of winners. That means there will be plenty of stations that seem to have the winning formula, even if most won just by chance.

Last December even Arbitron got in on the act by presenting a piece called The Ratings Characteristics of PPM Top Performers (pdf).

The presentation was on pretty shaky methodological ground, but that didn’t dampen the excitement in learning PPM’s secrets.

One trade publication flatly declared:

Arbitron found hardly any difference in the length of listening durations among average stations and top performers, but top performers had significantly more daily listening occasions. It's all about the number of listening occasions.

This is how PPM myths are born.

Let’s look at what Arbitron did. They took the top 3 stations in 33 markets, averaged their share, cume, TSL, and occasions, and then compared them to the average station. You can read their conclusions above.

The table clearly shows that the top 3 stations had much higher cume audiences than the average station, slightly more TSL, but only one more occasion.

 

So the biggest difference between the top stations and the average station was cume. If you want to be a top three station you need nearly four times the audience as the average station.

Not an earth-shattering observation, but a useful reminder.

TSL differences were considerably smaller. Top three stations only have 22% more TSL. Not a huge difference, particularly when the cume difference is 247%!

Daily occasions differ by one single occasion. That’s it. One occasion.

On top of that, the top station in each market has no more occasions than the average of the top three. Despite Arbitron's conclusion, the evidence suggests a rather weak relationship between occasions and winning.

Yet we're told that it's all about occasions.

Arbitron released only 18-34 results for some reason. The presentation alluded to a 25-54 analysis, but those results were not shown.

In our next post we’ll continue discussing the issue of PPM winning secrets and show a Harker Research analysis that raises more questions about Arbitron’s Top Performer claims.

In the meantime, be very wary of people claiming to know the secrets of PPM....and stay away from guns.

Brits Know How to Measure Radio. US? Not So Much.

The US was once the world leader in broadcast radio, but no more.

As one of the first countries to welcome entrepreneurial commercial broadcasting, the US established a tremendous lead over other countries in radio broadcast innovation.

Decades before other countries had any independent broadcasters, we had nearly ten thousand radio stations operated by thousands of independent owners.

This giant collaborative network produced diverse programming, rapid technological innovation, and a well developed system for measuring and reporting radio listenership.

As other countries gradually opened the airwaves to private operators, they looked to the US to learn how competitive commercial radio was done. One country that took a keen interest in US radio was the United Kingdom.

Harker Research began consulting British radio stations in 1990, and we found an industry that had a lot to learn, but an industry eager to learn.

Now two decades later, UK radio has not only caught up. In many ways, it has quickly moved beyond US radio, leaving it in the dust.

One important area where the UK is well ahead of the US is how radio is measured and reported.

The US has a patchwork of different companies measuring the various ways people listen to the radio.

Arbitron measures over-the-air terrestrial radio using two different methodologies that produce contradictory estimates.

But Arbitron doesn’t measure the fastest growing segment of radio listening, Internet radio streaming. Triton Digital among others does.

Triton produces ratings using different listening metrics and different editing rules from the way Arbitron measures terrestrial listening, however.

Consequently, Arbitron and Triton numbers are not directly comparable. They cannot be combined as reported.

On top of that, Triton only measures the streams of those companies who pay Triton to be measured, a small percentage of US broadcast stations streaming.

This hodgepodge means that the US, once the world leader in commercial broadcasting, has no reliable measure of total broadcast radio listening.

The US radio industry does not know how many people are listening to commercial radio, how long they listen, the proportion using each platform, or even whether broadcast listenership is growing or declining.

Not a clue.

Compare the plight of US broadcasters to UK broadcasters.

In the UK, commercial broadcasters and the BBC jointly own RAJAR, Radio Joint Audience Research, a service who’s task is to measure all radio consumption.

Terrestrial, digital, and Internet listening are measured by one service owned by broadcasters using a consistent methodology that produces reliable and comparable estimates of radio listening.

UK broadcasters know how many people listen to radio and how long they listen across all the different ways people listen to radio today.

The rating service was able to recently announce that radio listening had reached an all time record high, something US broadcasters can only guess about.

While US broadcasters focused their hopes on PPM to advance US audience measurement, RAJAR was looking beyond PPM and recognizing the need for a system that could measure all platforms equally as well.

UK broadcasters have already succeeded in that goal while US broadcasters haven’t even begun discussing the need.

After years of leading the world, US radio has fallen woefully behind the UK in the innovation necessary to prepare radio for its digital transition.

The US needs a single radio measurement system that looks at radio in all the ways it is consumed. Having lent the Brits a hand in the past, maybe they’d be willing to help the US now.

We need it.

Creating Radio’s Next Killer App

One of broadcast radio’s greatest strengths is its availability, its ubiquity. Radio is everywhere.

Radio’s continued strength will increasingly depend on creating useful applications and making sure the apps end up on the screens of connected devices, like smartphones and digital dashboards.

Getting a user to download an app is easy. Apple users have downloaded over 10 billion apps alone.

Estimates place the average number of apps a user has downloaded somewhere between two dozen for some devices to over 60 for iPod/iPhone users.

The vast number of apps are downloaded, launched once, and then never used again.

So the key metric of success is not how many people have downloaded your app. It is how often it is used. Radio stations need an app that listeners go back to time and time again.

By that measure, we believe radio is headed in the wrong direction with apps.

The entire approach has to be rethought, and a good place to start is the 1970s.

Back then car radios had a limited number of pre-sets, often just two or three for each band. The push buttons were valuable and stations fought over who got a button.

Programmers knew that if they could get on one of the presets, people would come back over and over again. Listeners just found it easier to switch between a few buttons rather than use the dial.

That’s when liners like Make us your first button and Set your dial and rip the knob off entered radio’s lexicon.

The push button was radio’s first killer app.

Fast forward 40 years, and radio finds itself with a similar challenge and similar opportunity.

Except instead of the push button, today’s challenge is to become the screen’s killer app, the first app that listeners turn to.

With an Internet connection, one can potentially listen to thousands of radio stations from anywhere in the world. Yes, listeners can, but most won’t.

People haven’t changed much from the 1970s. While the number of options has exploded, most listeners still settle on a handful of stations and just switch between this small set of stations.

It means real estate on the screen is just as valuable as the push buttons were years ago. Get on the screen and you have a big advantage. Create an easy to use useful app, and you’ve got it made.

That’s why Pandora has worked so aggressively to create an easy to navigate app and then to get it on as many screens as possible.

Pandora has cut deals with Ford, BMW, Toyota, and others so that their app appears on the dashboards of the auto makers’ cars.

Pandora has a significant advantage over broadcast apps because Pandora’s single app gives a user direct and convenient access to as many as 100 personalized streams.

A user need not navigate through hundreds or thousands of stations to find something of interest. Having created every station on the app, a user wastes no time having to sift through a bunch of stations she will never listen to.

As we recently pointed out, Clear Channel also has a deal with Toyota so that iHeartRadio will share the screen with Pandora, which is a good start.

Harker Research believes, however, the corporate approach of bundling a group's stations into a single app will need to evolve if radio is to overcome pure-play’s app advantage.

There are two problems with the corporate radio group approach. First, only the company’s stations are available. (We should note that the Radio.com app includes AOL Radio.) Secondly, there are hundreds of stations on each app, most of which a person will never listen to.

With a corporate app, a listener has to use some other app or browser if she wants to listen to a station outside the company’s own.

There are other apps like TuneIn Radio. It offers access to 50,000 AM and FM streams, but the sheer number of stations can be overwhelming. It takes five taps just to reach a market's listings.

Does a corporate gateway help or hurt broadcast groups maximize TSL? We think it ultimately hurts.

Most people listen to three or four favorite stations owned by two or three companies. Groups offering access to only their own radio stations force listeners to use multiple apps, which discourages listening.

So we are back to the 1970s, when stations fought over a limited number of buttons, except in this case radio has the additional problem of pure-plays.

We believe most listeners won’t go through the effort of navigating multiple apps to listen to the radio. It is far easier to navigate within one app like Pandora's that offers enough choice.

One solution is for a broadcast group to welcome stations outside the company (for a fee) to be part of their app. Call it Radio USA.

This inclusiveness eliminates the need to navigate between apps, but the number of stations in the US alone creates navigational challenges within the app. 

A better solution would be for entrepreneurial aggregators to create market apps, combining stations in a single market regardless of ownership.

Most people spend the majority of their time listening to local stations. Enabling users to install an app that offers all the local stations will increase over-all broadcast listening by making it more convenient.

The aggregator could pay stations to be included, or app advertising revenue could be shared among stations on the app.

This approach is an improvement, but Harker Research believes that to maximize broadcast TSL and marginalize pure-plays an entirely different approach is necessary.

We believe to successfully compete against pure-plays the ultimate solution will be an app that combines local stations with a Sirius-like lineup of format channels along with personalized channels. It will transform the broadcast app into a supermarket of audio options.

The app will have several local morning show channels offering different mixes of local services and information.

The app will also include a wide range of national generic music stations and syndicated talk programming. Customizable personal stations will round out the offerings. All on the same app.

It will be a package that offers a mix of stations to keep listeners on the app most of the time.

Over time we expect that all broadcast groups will create similar packages. Ironically, radio will end up pretty much where it was in 1970, but with groups rather than stations fighting over a couple of push buttons.

And broadcast radio will continue to be the dominate source of audio entertainment.

Pittman versus Pundits: Who's Right?

There are two types of radio people, Talkers and Doers.

Talkers spend all their time complaining. They react to new competition by whining about all the reasons the new competitor is going to succeed.

Radio doesn’t get it!
Radio is moving too slowly!!
Broadcasters just don’t understand!!!

Doers, on the other hand, focus on their side’s strengths and advantages. They weigh them against the competition’s weaknesses and vulnerabilities. They then set about implementing a strategy to win, confident of ultimate victory.

We can see the contrast between the Talkers and the Doers in how radio people are responding to Pandora.

The Talkers are radio pundits who have all but conceded radio’s defeat. They point to Pandora’s extremely high awareness, exponential growth, its apparent unstoppable momentum and declare radio hopelessly outclassed.

For pundits, the matter is closed. No room for debate. Anyone who disagrees is a relic of radio’s past glories, a digitally ignorant luddite racing towards an analog abyss.

Bob Pittman is a Doer. He doesn’t buy the fatalistic pundit talk.

He understands radio’s strengths. He believes Pandora is a one-dimensional product. He offers concrete evidence that broadcast’s resources are far greater than Pandora’s. He is confident that Clear Channel can successfully compete against digital pure-plays.

So what is punditry’s response? Apoplexy.

To them, Pittman is the poster child of what’s wrong with radio. Proof that radio is full of simpletons who just don’t get it. Here’s what they say here, here, here, and here:

Maybe instead of choosing the denial route about Pandora, broadcast radio operators ought to go to school about this brand and what it means to consumers.

How much more evidence do broadcasters need that time is short, and opportunities are fleeting? Exactly what are you waiting for?

If terrestrial broadcasters are going to survive, it won't be by dismissing Pandora. So many broadcasters are resolutely incurious as to the reasons for Pandora's success (sic), and it is that lack of curiosity that some broadcasters may not live to regret.

Pandora (is) the customizable radio service (that) radio people and Bob Pittman love to underestimate.

So what is it that Pittman says that so infuriates pundits?

Here’s a few comments from his recent Los Angeles Times interview:

"The perception is that radio is somehow in trouble. The reality is that we have the same percentage of the population listening to radio as we did in 1970."

Pundits jump all over this and claim that the real issue is TSL. They make every effort to perpetuate the myth that radio TSL has evaporated, but there is simply no credible evidence that it has.

Arbitron looked at PPM trends and found no decline in TSL during Pandora’s meteoric rise. None. Harker Research analyzed both PPM and diary trends and found no evidence that broadcast TSL had been impacted.

In truth, broadcast radio consumption is increasing, not decreasing. Terrestrial usage has remained steady while broadcast consumption by way of streams has grown, something we proved some time ago.

"Pandora is not really radio in that it's not curated. It's more like a playlist that you put on shuffle. They don't have local information or local personalities."

Take a look at the graph at the left (click to enlarge). It shows Pandora session TSL compared to the leading broadcast streams. Broadcast stream sessions last 55-81% longer than Pandora’s sessions.

 

For a service that pundits claim is light-years ahead of broadcast, Pandora doesn't look very sticky.

Of course, Pandora has a high proportion of mobile listeners. As we recently explained, mobile listening hurts TSL. However, even when mobile isn’t included, Pandora still only has an hour of session TSL.

Longer broadcast sessions could be a consequence of radio’s curation and local content. Lack of curation could be Pandora's greatest vulnerability, a point that we made in 2009.

"We have roughly 30 million unique visitors each month to our online radio stations. Pandora has 25-plus million."

Pandora portrays itself as an enormous service serving vast numbers of Americans.

We’ve shown that in fact consumption of Pandora isn’t particularly high, especially when you consider that Pandora consumption is spread across 1.4 billion streams, while broadcast consumption is spread across only 10,000 stations.

"The good news is that we have billions of dollars in advertising revenue from our terrestrial radio business. We can afford to invest and grow our online audience. We have the deep pockets to wait it out."

Last year Pandora billed $138 million. Local  broadcast  radio  stations  billed $16,000 million. This last quarter Clear Channel alone billed $650 million.

Do we really need to debate who has deeper pockets and what deep pockets mean in a competitive battle?

"Later this summer, we'll be adding a feature that will let our users build custom radio stations just like they do with Pandora."

This is the point where the pundits’ true colors show. Critics first complain that radio isn’t moving fast enough, but then belittle Pittman for announcing that Clear Channel are building their own Pandora.

As we noted last year in the post Thanks Pandora, We’ll Take it From Here:

Nothing that Pandora does is proprietary. Every aspect of Pandora can be replicated, and ultimately improved upon.

What’s wrong building your own Pandora? Every broadcast group should have their own music genome project. As we pointed out recently, now there’s even an app for that.

It's a long tradition in broadcast radio to watch the competition, copy key elements of what they are doing better, then improve upon it. It has served radio well. Isn’t that what Pittman is doing?

But to give Pittman credit would be an admission that maybe radio isn't brain-dead. That would screw-up punditry's narrative of broadcast's self-inflicted imminent demise.

Pundits want to appear visionary, so they create false straw-man arguments they can easily win. Only they understand the true threat that Pandora presents to radio. That’s why they are much smarter than we resolutely incurious radio people.

What’s surprising is that we at Harker Research don’t run into these clueless radio neanderthals that the critics see everywhere. We work with radio people who are vitally interested in preparing for radio’s digital future.

It’s just that real broadcasters have to keep an eye on the bottom line while doing it.

What too many pundits fail to understand is that the goal of radio’s digital transition is not to find a way to beat one service that bills 1% of radio’s revenue.

The goal is to find a strategy that first protects radio’s piece of the advertising pie, and ultimately helps grow radio’s share.

On this subject, the talkers are strangely silent.

Talkers versus Doers. Time will tell who’s right. We’re betting on Pittman. In this debate, he's the only one with skin in the game.

Slacker: Mobile's Double-Edged Sword

It’s not often that we get a chance to see the impact of mobile listening up close. The latest Triton Digital numbers offers us one of those rare opportunities.

Slacker signed-on to Ando Media in July of last year.

For the first few months, only Slacker’s computer-based listeners were counted.

Then sometime around November, Ando began including mobile listeners in the service’s count.

The graph above shows what happened.

The orange line shows the session TSL trend. When only computers were counted, average sessions lasted about 1:45. Now with mobile added, average session length has plummeted to less than 45 minutes.

The green line shows Session Starts in millions. After several months hovering in the mid four million range, Slacker “took off” in November, settling over the 20 million mark.

While some portrayed the 4.5 million to 20 million Session Start gain as growth, it's likely there was very little real growth.

The “spurt” beginning in November simply shows that mobile listening makes up a great deal of Slacker listening. Before November Ando didn’t count it. Beginning sometime in November Ando started.

The warning for those who believe that mobile will be the real breakthrough for streaming is the precipitous decline in Slacker’s TSL once mobile listening was counted.

You can see it in Slacker total consumption measured in hours-tuned.

While Slacker’s Session Starts increased more than four-fold (+355%), total hours-tuned grew less than two-fold (+84%).

What it tells us is that Slacker’s 15+ million mobile users are spending virtually no time with the service. Slacker’s TSL is coming from the service’s computer-based users.

Most agree that streaming growth is going to come from a wave of new mobile users. If that is the case, we will see more Slacker-type growth, lots of new sessions spending little more than a few minutes of drive-by listening.

For subscription-based services, that’s no big deal. A paying subscriber is a paying subscriber no matter how little she listens.

For advertising-based services, it’s a much bigger deal. Very short listening spans are problematic when you’re trying to sell advertisers.

Advertisers want to have some confidence that users are going to see or hear their commercials. The most common way to do this online is to have a pre-roll position, one or more commercials that users have to sit through to hear the service.

The problem is that with short listening spans, all the user will hear is the pre-roll--not a good way to build an audience, or a business.

Can We Call Internet Radio Mainstream Yet?

“So can we call Internet radio mainstream yet?” The question, recently posed by an internet radio supporter, is an interesting one. Has internet radio become mainstream?

Maybe one day, but it looks a long ways off.

Internet radio claims to be the future of radio. According to its supporters, one day soon few people will be listening to traditional radio. They will be listening to new personalized streams of Internet companies like Pandora, Slacker, and others yet to be created.

The gold standard then is broadcast radio. If the goal is to replace broadcast radio, then Internet radio should be measured against broadcast radio.

The comparison is somewhat complicated by the fact that internet and broadcast radio ratings are not directly comparable. Internet radio measurement counts IP addresses and reports things like sessions, while broadcast radio measurement counts listeners and reports things like quarter-hours.

Fortunately, with a little math, the two can be compared. In a June post we showed that both broadcast and online ratings can be converted into Hours-Tuned, a metric that tells us how many hours people spend “consuming” radio.

For example, people in just New York and Los Angeles consume about 334 million hours a week listening to local radio stations in the two markets. Nearly half of the hours are spent listening to just the top 20 radio stations.

The latest national Arbitron estimate puts total broadcast consumption at over 3.5 billion hours a week.

How does that compare to Internet services?

The top 20 Internet streams generate 377 million hours of consumption. The number is somewhat misleading, however. Over 200 million of those hours go to Pandora alone.

This means that the consumption of broadcast radio in just two markets nearly equals the national consumption of the top 20 Internet radio services even when we include Pandora. Back out Pandora and New York broadcast consumption alone is greater than the remaining national Internet services.

There’s really no comparison between broadcast listening and Internet listening. For example, consumption of just the Clear Channel stations in New York exceeds the total national Clear Channel streaming consumption.

Nine New York broadcast stations have greater hours-tuned than Slacker’s national numbers.

Were it not for Pandora, Internet radio would barely exist. The graph above shows weekly consumption in millions of hours according to Ando Media’s July domestic full week ranker. For scale, we’ve included the leading station in New York and Los Angeles (in bright red) for comparison. (Click to enlarge.)

So can we call internet radio mainstream yet?

For a business that hopes one day to be broadcast’s successor, Internet radio is falling well short of being a serious threat. Yes, Pandora is a phenom, but its success is the exception, not the rule.

There are no other break-out successes that suggest Internet radio has gone mainstream. Maybe one day, but not this day.