The good news for the radio industry is that Nielsen just reported the medium’s audience rose substantially in October, nearly recovering all of the erosion it experienced during the COVID-19 pandemic to date. The bad news is that it comes with a caveat: Nielsen made a methodological change in October to factor people who listen with bluetooth headphones that effectively boosted radio’s audience by 4%.
Nielsen called the headphone adjustment a one-time factor that would not be broken out in future months, but the inclusion fundamentally alters the bases for Nielsen’s estimates for U.S. radio audiences.
Nielsen had been planning to make an adjustment for some time, because it’s radio audience measurement technology -- portable people meters, or “PPMs” -- cannot measure listening done via a Bluetooth headphone, so it developed a method for factoring it based on surveys of former Nielsen PPM panelists who left Nielsen’s panel.
That said, Nielsen executives recapped how radio’s audiences have been “recovering” month-to-month since March, when it began eroding due to the displacement of work and social lives following the U.S. COVID-19 pandemic.
Effective with its October estimates -- including the headphone adjustment -- Nielsen says U.S. radio audiences are 95% of what they were in March.
To understand the impact of the headphone adjustment, Nielsen executives noted that October’s radio audience was up 6% over September’s and about 2 of those percentage points came from “organic” listening, while four of those percentage points came from the headphone adjustment.
During a briefing with clients today, Nielsen executives indicated the headphone adjustment had a bigger impact on some demographics and markets more than others, noting that some big urban cities such as San Jose, CA; Washington, DC; and Austin, TX, had disproportionately bigger gains from the adjustment vs. markets like "Greensboro and Las Vegas."
Joe, I gather from your interesting report that Nielsen is not measuring bluetooth headphone usage but is merely applying some sort of general percent add-on ratio---based on other research---to whatever "listening" the PPMs report. I wonder how they handle those adjustments for each station---there are many types of formats and personalities out there---and various locations where radio listening takes place. And what about irregular programming---such as baseball game coverage? Do the "other surveys" provide a valid indicator for how many bluetooth listeners they reach for each market ---no matter how crucial the game happens to be or what teams are playing?
Good questions Joe. Having worked on the AU currency radio ratings for years, I have to say that solving this issue is extremely complex. It falls into the "Known Unknown" bucket. That is, we know the listening is missing (from electronic capture) but we don't know by how much.
You have two choices. (1) Ignore it and under-report the market (2) use secondary data to 'in-fill' for the missing data using a different data capture method.
Here in Australia we use hybrid methods. We use paper diarires. electronic diaries and are also collecting metered usage. The problem with metered usage is once you don't personally have the capture-device the listening is lost forever, so carry-rate is vital. At least with a diary most respondents will 'back-fill' using the recall (albeit could be faulty recall) - a proxy for their usage.
Electronic capture is also affected by tech-change. For example, using mobile phones for data capture is highly dependent on a stable OS (typically iOS or Android). A change is made an the app all of a sudden isn't functioning at 100%. But as consumers do OS updates at their own pace it may take days, weeks or even months to find out that what used to work no longer works. Also some of the OS may block certain 'features' such as headphone listening. And finally there is an inherent assumption that because the device can 'hear' the radio content then that is a user listening to the radio. Have you ever been called away from your desk into a boring meeting and have your device racking up listening? Gyroscopes can assist in identifying probable 'unaccompanied listening' and filter it out but that is also far from perfect.
Given all that, adding a BT factor to the sampling ratio is probably as accurate a method.