Last week, I reviewed some history and offered general thoughts about Nielsen’s plans for the radio diary. This week, I’ll comment on the specifics.
First, the cringe-worthy: The deck included the point, “retain unimpeachable truth set.” The word “truth” should never be used with survey research. Nielsen could be the best survey research company on the planet (and the company is good, no matter how much we attack them), but it will never know if it has truth. Period. Referring to the diary service in 2026 as “unimpeachable” should include a laugh track.
Next, Imran Hirani, who gave the presentation, referred to Scarborough and measuring newspaper readership. He noted that Scarborough adjusts its data to make the results match readership audit figures from the Alliance for Audited Media (AAM). Imran suggested that this approach might be usable for radio. Funny, but in all my years in the business, I’ve never heard of anything equivalent to AAM for radio or audio, and frankly, it’s impossible. That’s Nielsen’s role, so Imran’s point made no sense.
Questions About Current Methodology
With that out of the way, most of his points about how current diarykeepers perform their task were on the money. A large percentage do not write down their listening as it happens, and plenty fill in the entire diary at one sitting, whether the day after the survey week or even during the week. If the postmark on the diary is after the end of the survey week, the diary can be intabbed.
And Thursday is the big day for listening because it’s the first day of the diary week. Take me back to 1970 — nothing has changed.
He mentioned a non-response study from “several years ago.” I’m not aware of one, although I’ve been away from day-to-day dealings with Nielsen for six years. Perhaps he was referring to non-response studies from last century, which would be a new definition of the word “several.”
There were two such studies. One was publicly released because it showed that non-responders were similar to those who filled out the diary with respect to radio listening. There was also a “Soviet history” non-response study from 1984 that showed bias. “Soviet history” means that the study never existed, even if it did. If there is a more recent study, Nielsen should release it, for better or worse.
The presentation covered the use of modeling, and Imran was correct that modeling exists in the diary service today. I was directly involved when we implemented a small amount of modeling around 20 years ago. Diaries with two “missing days” of listening — meaning no listening was recorded and the diarykeeper did not check the “no listening” box — could be counted.
Previously, Arbitron accepted one missing day without the box checked and assumed no listening on that day.
The Future of Nielsen Audio
This model was different because listening from another day was used for the missing day. While I can’t remember the exact system, as an example, if the diarykeeper left out Monday, listening from Wednesday may have been inserted. However, if the two missing days were Saturday and Sunday, the diary was not counted. If only one of the missing days was a weekend day, that day was assigned no listening. Got it?
Why do it? Pressure for higher response rates. It was cheap and easy to implement.
Here are the key items in Nielsen’s plans for the service. First, give up on probability sampling. The cost of probability sampling in 2026 is outrageous, and response rates are awful. Nielsen goes through a multi-stage process just to get households into the sample. What the company appears to be proposing is the use of databases to recruit respondents, especially harder-to-reach groups such as young people and minorities.
This is a big change and, as I noted last week, opens the door for any other company interested in jumping into the radio ratings game. What was left unsaid — and perhaps Nielsen hasn’t decided yet — is whether that piece of the overall sample moves from “everyone 12+ in the household” to “single person per household,” as commercial databases are typically made up of individuals rather than households.
Further, the Nielsen deck noted that because respondents opt into these databases and can choose the surveys they want to complete, this will result in “high interest and high quality data.” On one hand, this may mean radio listeners will be more likely to participate, keeping listening levels stable.
Reshaping That Future
Most of you know the term “contest pigs.” People in these databases complete surveys for one major reason: the incentive. I’m in a few myself. In fact, my flight to the Barrett Media Summit later this month was paid for through surveys I completed in exchange for airline miles.
Next is the move to a one-time survey instead of seven days. This approach is not new, just new for Nielsen. I was director of research for Birch/Scarborough’s radio ratings service before the company was shut down at the end of 1991. The Birch system was a probability-based phone survey, which was feasible at the time, and the questionnaire was straightforward:
- What stations did you listen to in the past week?
- Using four dayparts covering 6 a.m. until midnight, what stations did you hear yesterday?
- For each station you heard yesterday, did you hear that station the day before yesterday during that same daypart?
That was it. The duplication allowed Birch to model a seven-day cume, which was typically close to the Arbitron seven-day diary cume. Also, because Birch was sold to what is now Nielsen, the company owns that intellectual property, which probably languishes in a file drawer somewhere.
I expect Nielsen will take the data from the single questionnaire and model it using the remaining seven-day diaries it claims it will continue to field. It makes sense and would work methodologically.
While you won’t find odds on it at Kalshi or Polymarket, it’s a safe bet that this is the second step (mSurvey was the first) toward a completely online Nielsen Audio service using convenience samples. It’s a sensible move, but it differs from the method we’ve used for the last six decades, one that has withstood challenges from multiple competitors. The results will be different.
Here’s the question: Will another research company have an opportunity to overthrow King Nielsen? Will radio and audio measurement change? I expect we’ll see a new landscape by the end of the decade.
Let’s meet again next week.
Barrett Media produces daily content on the music, news, and sports media industries. Sign up for our newsletters to stay updated and get the latest information right in your inbox.
