Ratings Carnage and the Heavy DVR Theory

If you're reading this, you probably know it's been a fairly brutal week in the TV ratings world, even though there's no clear explanation in the overall viewing levels. Frankly, I find the "10-15% of TV viewers just suddenly decided they hate broadcast TV and love cable" theory about as ludicrous as the conspiracy theories.

But there is one interesting one that's been floated recently. And it's wrapped up in Nielsen's relatively new definition of "viewing level." Here's Nellie from Deadline:

I think a footnote in the HUT definition may hold the key to the puzzle. In addition to live viewing, it also accounts for people watching pre-recorded shows on playback. The theory is that a blockbuster TV event like the high-rated Grammys on Sunday created a “tsunami effect,” with people catching up on previously recorded shows as well as possibly replaying the Grammys, instead of watching that night’s programs. That perpetuated for a few days, especially with Valentine’s Day in the mix when many people were out, creating an additional backlog of shows on their DVR. The theory is supported by measurements that registered a higher-than-usual level of playback viewing this week.

And I believe my own commenter iggy mentioned this a couple days back and I didn't exactly grasp what he was talking about. It does kinda make sense; the Grammys are typically the biggest non-sports event on TV against which the networks go "full-strength," and this year's was wayyy bigger than usual. That leaves a lot of potential entertainment programming (including The Walking Dead, remember) sitting on the DVR shelf heading into the week. I'm not endorsing the idea, but I am gonna try to explain it a bit and provide some numbers that are potentially in its favor.

She and iggy aren't the first people to notice the problem here. My very first post in the creation of the True Strength number was largely about this. But since those posts weren't too well-read, lemme give a little refresher course:

Nielsen's current definition of "viewing level" is essentially "everyone who is watching TV right now." Seems simple enough, right? Yes. It's very elegant. I'm guessing that's why they went that route. But the problem is that it gives you very little info on what they're watching. They could be watching live, they could be watching DVRed shows from the same day, or they could be watching DVRed shows from any time in the past. All anyone (at least in a public setting) gets is one overall number that encapsulates everything.

Typically, the percentage of DVR usage is relatively predictable; in True Strength, I came up with a rough conversion to the Old Methodology PUT and it's almost always gotten me pretty close to even on a year-to-year basis. But if something strange does happen, we have no way of knowing.

At least not directly. (I don't have access to her "measurements that registered a higher-than-usual level of playback viewing.") But I have noticed that there's this guy Silvio in the TVByTheNumbers comment sections who has access to the % Same Day DVRing numbers (correction: it's the Live Only numbers he was getting (see comments)), which he applies to TVBTN's broadcast finals to break the shows down by live and same day DVR. I'm guessing the numbers are a bit off since TVBTN's finals are only precise to one decimal, but they're still close enough to paint a pretty clear picture about some of these declines:

How I Met Your Mother Last Week: 2.91 Live / 1.09 SD | This Week: 2.33 Live / 1.07 SD | Net: -20% Live, -2% SD
2 Broke Girls Last Week: 3.27 Live / 1.03 SD | This Week: 2.76 Live / 1.04 SD | Net: -16% Live, +1% SD
Two and a Half Men Last Week: 3.44 Live / 0.86 SD | This Week: 3.10 Live / 0.80 SD | Net: -10% Live, -7% SD
Modern Family Last Week: 3.72 Live / 1.78 SD | This Week: 3.23 Live / 1.47 SD | Net: -13% Live, -17% SD
Happy Endings Last Week: 2.50 Live / 0.40 SD | This Week: 2.02 Live / 0.38 SD | Net: -19% Live, -5% SD
Law and Order: SVU Last Week: 1.71 Live / 0.29 SD | This Week: 1.40 Live / 0.30 SD | Net: -18% Live, +3% SD

The Voice Last Week: 5.57 Live / 1.13 SD | This Week: 4.87 Live / 1.13 SD | Net: -13% Live, -0% SD
Smash Last Week: 3.31 Live / 0.49 SD | This Week: 2.43 Live / 0.37 SD | Net: -27% Live, -24% SD

Thanks to Silvio!

The top six were the shows whose Live+SD drops appeared most suspicious in the first three days of the week. And with the exception of Modern Family, they have one big thing in common: the drops were mostly Live. I didn't find The Voice that suspicious but it turns out its drops were basically completely Live too. So it does appear that there are a bunch of people bailing on live TV this week, yet there are no declines in viewing in those live moments. With no clear spikes on cable, they're likely ending up on the DVR.

Nellie seems to think this is dissipating as the week progresses, which would make intuitive sense, but there were still some weird drops on Thursday. (Love how she says "at 9 PM, things begin to stabilize" and then her next words are about Grey's Anatomy losing 23%.) Will be interested to check on those later.

I'm not saying this is what it is because I don't have the numbers to know that. But a lot of these drops do happen to be hugely skewed toward live. That means it's definitely not impossible under the new methodology that we could be seeing a PUT inflation as a result of previous-day DVRers. (This is why I always thought the old "tendency to view Live+SD" methodology was better. We have to go back, Nielsen! We have to go back!)