Twitter knows more about what drives viewer behavior than NBC

In my last post, I argued there’s more value in social media as its own content channel than in its ability to drive ratings or sales. The prime mover of that post was a quote in the Financial Times from NBCUniversal’s head of research Alan Wurtzel who said social media “’is not a game changer yet’ in influencing television viewing.” He also said “the emperor wears no clothes” which tells you just how much marketing people love a good cliche.

After taking another look at the articles written in the wake of his comments, it’s unclear how Wurtzel came to these conclusion but more on that in a bit.

If you’re going to draw conclusions about cause and effect, it helps to base them on the results of an actual study.

Here’s the latest: On Thursday, Twitter’s head of research Anjali Midha released a second batch of results from a study of 12,000 Twitter users that examined the effects of tweets on consumer action. Her first post focused on the relationship between the kinds of tweets consumers saw from/about brands and the type of action they took afterwards. It’s definitely worth a read.

Specific to this discussion is Midha’s second post which deals with the effects of TV-related tweets. In my post, I argued Twitter can be a complementary content channel all its own, but if you’re going to look to it as a call to action, it’s important to consider whether it affects awareness, consideration and brand education and not just ratings or sales.

It turns out tweets can do all of that:

90% of respondents took subsequent action such as watching a show they’ve never watched before, resuming watching a show that they’d previously stopped watching, and/or searching for more information about the show online. In fact, one-third of respondents reported having changed the channel to watch a show after seeing Tweets about it. [Emphasis mine]

Even when you consider the effects of tweets on consumer action when the show isn’t airing (“60% tweet about shows when they are not actually watching them” and “58% tweet about TV shows while they watch on time-shifted platforms like OnDemand, Hulu, iTunes and Amazon”) the data makes a powerful case:

Those who recall seeing TV-related tweets have…

  • Taken action to watch TV show content (77%)
    • 42% made a plan to watch the show later
    • 38% watched episodes online
    • 33% changed the channel to watch the show
  • Searched online (including video sites) for more information about the show (76%)

That 33% is a big deal. Those are people who stopped watching their current show and took an immediate action (the one networks desire the most) because they saw tweets about it. It’s really hard to change user behavior, especially while they’re in the middle of that behavior and data that says tweets can do that refutes just about everything NBC’s Alan Wurtzel said about social media not driving ratings.

(I will not be offended if you stop reading now. Honestly, I should have quit while I was ahead and stopped this post here. But I started going down the rabbit hole of the NBC Olympics data and determined it’s all hinky as hell which bothered me so I had to write about it to get it out of my head.)

How did Twitter and its research partners come to a conclusion that’s the exact opposite of Wurtzel and NBC? Perhaps because they asked the people seeing those tweets what they did after rather than comparing one set of numbers to the other which looks to be all NBC did. It’s worth taking another look at the initial story that started this conversation because near as I can tell, it’s based on some really weak correlation, not causation.

I re-read the Financial Times piece which first broke the story along with articles from Re/Code and the Associated Press. (I also read a few other stories that basically repeated all of the above.) All NBC/Wurtzel seem to have done is compare the number of people who watched the Olympics with the number of people who posted about it on social media and used those numbers to draw conclusions.

According to the Financial Times, NBC found “just 19 percent of Olympic viewers posted about the games on social media.”Where does NBC get this 19 percent number from? It’s not clear in the FT piece. The only total viewership number it cites is “the average 21 million” people who watched in prime time. But a recent NY Times piece provides more data:

An average of 21.4 million viewers watched the Sochi Games in prime time. But there were also 62 million digital users and 10 million live online video viewers.

So that’s 93 million. Nineteen percent of that number is 17.7 million. That number comes close to the other data in the FT piece:

The social media activity pales in comparison to the average 21m viewers who tuned in for prime time coverage of the game on broadcast television, Mr Wurtzel said.

On Twitter, there were 10.6m Olympic-related messages posted by 3m unique users. A total of 23m people saw tweets about the Olympics. On Facebook, about 20m people posted, commented, shared or liked something related to the Olympics, with those posts reaching a total of 150m users[Emphasis mine]

For the sake of argument, let’s say NBC’s data is correct and that using these numbers to come to a conclusion is a good idea (it’s actually terrible idea but I’ll explain why in a second).

If I was Alan Wurtzel, I’d look at these numbers and jump for joy.

In what marketing universe does 173 million people (23m from Twitter + 150m from Facebook) seeing a message about the Olympics “pale in comparison” to an average 21 million watching it? Or even 21 million people watching it versus 20 million posting about it on Facebook (nevermind how many people on Twitter are talking about it).

Put simply, how does 95 percent of your audience telling someone else about your product not make NBC break its proverbial arm patting itself on the back?

Part of the problem here is the reporting has been spare, at best. There’s little beyond the Financial Times piece that offers any insight into the data Wurtzel saw (no press release, no conference call transcript) or what he would expect the numbers to be if there was a direct relationship between social media and viewership. Was he expecting the social media numbers to be larger than the viewership? That wouldn’t really make sense. So he must have expected the social media numbers to be higher than 19 percent but how high should they have been to suggest that social media posts drove ratings?

Moreover, unlike the Twitter study, nothing I’ve read suggests NBC specifically asked people who watched the Olympics whether they posted about it on social media or even asked them anything at all. Perhaps the Olympics audience isn’t inclined to post in social media about the Olympics because of demo- or psychographics.

Based on my read of the articles linked above, it looks more like NBC and Wurtzel just looked at one set of numbers (the people who watched the Olympics) and then looked at another set of numbers (the people who posted about the Olympics in social media) and based its analysis on the numbers themselves, not the relationship between the two sets of actions (watching and posting) about which it seems to know nothing.

There’s definitely nothing in their data to support the assertions that social media is “not a game changer yet.” As opposed to Twitter’s which suggests it very much is.

And when you consider the 33% of people who can be convinced to change the channel based on seeing tweets about a show, if there’s a naked emperor in this scenario, it’s definitely Wurtzel, not Twitter.

Leave a Reply

Your email address will not be published. Required fields are marked *