44 comments

[ 3.5 ms ] story [ 148 ms ] thread
Seems odd to use Youtube views to determine whether the Super Bowl spending was worthwhile. You have to choose to watch a YouTube video, while the Super Bowl ads are pushed to everyone.
The idea here is that if people connected with the video while watching the superbowl, they shared it online. Commercials were uploaded to YouTube by their respective company, and included a hashtag- deliberate pushes to form a conversation online. There are a lot of ways to measure engagement/impact of a commercial, this is one of them.
We thought that Youtube was a good gauge of general internet interest in the ad. If you liked the ad or ended up talking about it after, you probably pulled it up on Youtube to see it again.
Though, the only real data point that probably matters is what sales were like for the companies. I'd imagine they are at least a little correlated, would be neat to see some actual numbers.
We probably could get Clash of Clans downloads, interesting they're not necessarily selling anything.
Really? Have you played Clash of Clans? They sell gems!
(comment deleted)
Sales have a longer cycle than these numbers, and while the Superbowl is probably a big enough spend that it's noticeable for some companies, I doubt Budweiser or Nissan saw big bumps in sales from these spots.
Youtube views can be and are purchased just the same as TV views. The total views shown on a youtube video is inclusive of paid youtube advertising and isn't a very reliable metric of "interest" (when it comes to ads)
That makes sense, but I'm sure it's skewed between younger/more tech savvy target audience (ie, Clash of Clans) vs older or less savvy viewers.
The pac-man ad is a counterpoint. It's an ad with suspense and a twist. During the Superbowl, you have to watch the whole thing. I saw it on youtube, the suspense was curious, then there was the reveal - okay, got it now, no need to watch the rest of the light show. On Youtube, you can't recapture interest, but if someone has to watch the whole thing, they can re-engage later in the ad.

This doesn't mean that the youtube analysis isn't useful, but it does highlight a difference in the mediums.

Very nice analysis of the effectiveness of advertising. Things like this are really needed in the field.
Except I take issue with this statement from the article, especially as it applies to Superbowl Ads: A successful video will have more views, higher engagement, and resulting lower cost per minute.

The only thing that matters when putting forth a Superbowl commercial is that the viewer stood around long enough to figure out who sponsored it, even if they didn't stay for the whole thing. On air this is complicated by several factors.

The Youtube attention statistics are a nice proxy for it but it didn't take into consideration the likelihood that a potential viewer lost interest before they knew what it was 'about' (a particularly poor example of this would have been the Nationwide commercial-- people tuned out before the reveal and if they did find out who sponsored it after-the-fact, they heard it in a negative context).

And of course none of this translates directly into brand awareness, but having a talked-about Superbowl commercial is usually the point; the question is did you achieve the buzz you were looking for.

I feel a better measure of the utility of the commercial (answer the question: "what was my bang for buck vs. competitors?") would be to measure the trending of hashtags / mentions / youtube views of the commercial in the days following

>The only thing that matters when putting forth a Superbowl commercial is that the viewer stood around long enough to figure out who sponsored it

>The Youtube attention statistics are a nice proxy for it but it didn't take into consideration the likelihood that a potential viewer lost interest before they knew what it was 'about'

Great points, and agreed. This is what the attention plot ventures to take into account

>measure the trending of hashtags / mentions / youtube views of the commercial in the days following

This would definitely be useful to see.

I'm not sure how well this metric works for Supercell and Clash of Clans. I sometimes play that game, and I can tell you that during the Superbowl, there was an in-app notification sent to all it's players to watch the youtube ad, which greatly increases the number of views of the ad beyond what the other company ads could do.
Yea, that surprised us that Clash of Clans was #1, but figured it was something to do with it being a mobile product. #2 is Budweiser though.
> there was an in-app notification sent to all it's players to watch the youtube ad

Mobile platforms really need a way to report spam. Notifications to watch an ad should get apps removed from the push notification system.

I wonder if that's even a part of Embedly's data set. Is Clash of Clans using Embedly?
Recalls a SB ad years ago, during the original Internet bubble: E-trade showed a dancing monkey, then the text "Well, we just wasted two million dollars. What are you doing with your money?" https://www.youtube.com/watch?v=BnQMq5wtZcg

Gotta wonder what that cost them in lost business (you gonna invest with a group that brags about throwing money away?).

Gotta wonder what that won them in new business (you gonna invest with a group that had an add people were talking about more than a decade later?).
Kinda amazing they still are producing Super bowl ads. I'm guessing they're not spending too much on production with babys and cats.
Don't they have an 'embrace your inner micromanager' ad campaign now. Isn't that basically the opposite of what you are supposed to do?

Maybe they are targeting dumb money.

I don't invest with E*Trade because of that ad.

I don't buy Belkin products either, for comparable reasons a decade after. (Made routers that would randomly inject ad pages. I still won't so much as buy a cable from them.)

Coke's #makeithappy campaign ended up being pulled thanks to an effort by Gawker, that's gotta have some marketing folks upset.

Here's wondering if the gossip mongers like Nick Denton might have to pay the price for their tomfoolery.

http://www.jta.org/2015/02/07/news-opinion/the-telegraph/whe...

I thought Coke's commercial was a great response to last year, when there was such a backlash against including a gay couple in their advertising.

It's one thing for 4chan to come in and ruin the day, but for an actual company with actual managers and employees to ruin another company's advertising campaign by quoting Hitler? That's crossing a line, unfortunately one that Gawker has crossed many times in the past. How have they not been sued out of existence...

The commercial last year was the one with 'America The Beautiful' and the backlash was having it sung in other languages besides English.

Not really sure where you're pulling the gay thing from.

Is this an example of "professional trolling" ? Gawker trolled coke into cancelling their campaign, and the whole fiasco became a story in and of itself.
Has very little bearing and a lot of sample bias. How many of you actually viewed the commercials via Youtube?

The is a connection, and the data is applicable, but only casually.

The real gold here is how Embedly tracks content engagement all the way through the video, and how that information can correlate to content success. Regardless of how casual the connection may be.

Disclosure - I run a competitor to Embedly

As a side note, Clash of clans makes 5 million dollars per day so what they spent in this ad is like pocket change for them.
Good analysis of available data, but there lots of flaws in how the conclusion is drawn:

1. Using YT as a proxy for overall attention. There are so many variables here at work, it's hard to begin to start why this number isn't a perfect science.

2. Advertising isn't a straight line number of attention to sales (even though most geeks in the FB era seem to think so). There are many of factors at play here (branding, image, communication, loyalty, engagement etc) and many are immeasurable.

Believe it or not, effectiveness (and therefore "waste") of ad campaigns are not just how many eyeballs see the ad. I reckon most CMO's wouldn't take this type of napkin math very seriously.

Note - I understand this type of content is for inbound marketing for Embedly, so this is probably quite effective. I'm personally pedantic about the data science behind it so others don't feel betrayed by it.

Also people on YouTube will just quit watching an ad, whereas someone on the Super Bowl is almost certainly not changing the channel.
You bring up great points!

>1. Using YT as a proxy for overall attention

'Overall' attention is never a claim. This is simply building upon the views data with what is measured around how far people watch in a video.

>2. Advertising isn't a straight line number of attention to sales

There is no mention of sales in the post. It is comparing the price of the superbowl spot with the number of attention minutes. There are many factors which can serve as a measure of success, in this case it is looking at how people are watching the video after the ad spot.

>most CMO's wouldn't take this type of napkin math very seriously

Attention minutes is a fairly powerful measure. Ex. Upworthy uses it to determine which videos are most likely to be shared. The goal here is to show one way it can be used. In practice, it would be used among other measures an agency or company has- as you describe above.

> In practice, it would be used among other measures an agency or company has- as you describe above.

If this is the case, then why is the conclusion that their super bowl ad dollars are a waste?

> We’re going to use this to see who got the most bang for their buck, and more importantly — who messed up and what lessons there may be to learn.

Maybe I'm just being pedantic about your misleading title and lede....

PS - I assume you're the author. This is very good inbound marketing for your company...I do applaud you for that.

One of the points people don't get here is that you work with the data you have. And it is not perfect data.
I've got a major disconnect with the "facts" at the beginning of this article:

   Super Bowl ad spots are expensive —
   ranging from $20M to $135M.
   ...
   The cost refers specifically to the
   Super Bowl ad spot, and not to production
   of the commercial.
But Wikipedia[1] (and countless other links e.g.[2]) claim that a 30 second commercial cost about $4.5 million for the most recent Super Bowl.

So how does an ad cost $135 million? Did I sleep thru a 15 minute ad? The numbers just don't add up. (pun intended)

[1] http://en.wikipedia.org/wiki/Super_Bowl_advertising [2] http://www.forbes.com/sites/onmarketing/2014/01/29/yes-a-sup...

You still have to create the ad, and it can be a campaign which may cost a lot of money.
Wow, bank error in your favor. It looks like we had everything inflated by 10X. Appreciate you pointing this out and surprised no one else caught this including us. We're updating the post to reflect and will throw a mention in there.
Sick, and tight as well. Interesting to see that this kind of data surface too often -- either no one is doing it or they're playing it close to their chest.