Video Popularity – Views or Links

Update: I’ve updated this post with further comments from Bri at Holt Labs

For a while now, I’ve been using a couple of video tracking sites for work-related reasons, both to see what is popular and to track some of our own videos. The two systems use different tracking methods, one is on views and the other is on links. Both have their place when understanding the popularity of the videos from a marketing perspective. I’ve blogged about them before (and have been tested their services), but here’s a comparison, plus a look at why you should use both methods.

The first is is VidMeter, from San Diego based Holt Labs, a company that seems to consistently produce interesting systems.
Key elements from their general service are:

• Top 100 videos, based on views across 13 networks
• For each video provides total views, views per network, traffic history
• RSS feeds for most viewed of all time and most viewed per day.
• Widget to add to your blog/site to provide updating lists of the popular videos.

In addition, they also provide a tracker service for you to sign up and track your own videos. We’ve been using this in the office for a few weeks to track some brand videos we have out in the wild, and it makes a difference in making everything a lot easier when you’re reporting. The tracker service also tracks links to your video, so you can tell who is linking to your videos. Holt Labs are busy adding even more services to their offering, everything aimed at making it easier for agencies to understand what is happening with their content.

The second one is the Viral Video Charts, from London based Unruly Media. This system rates the videos on inbound links.

• It details where the video was found and how many links are being made to the site
• Embed code to add video to won site
• popular videos by categories
• RSS feeds for Top 10 daily
• Widget for your blog for top 10 (now moved from sidebar of this blog, but added in post)

The Viral Chart also accepts advertising, providing sponsored video links and blurb.

But the 2 charts give completely different results; as it stands at the moment there are no videos in common between the two of them. So as a content producer which metric should you use. In my opinion, both. Both methods give you a different insight into how your content is spreading. I asked the 2 teams why they had chosen their methods.

Bri from HoltLabs

“Ranking videos by links is good method for determining what the “blogosphere likes” and it’s usually cited on as a method for finding “good” video. Those sites are also competing with the “social video links” sites which allow users to vote on what videos are “good… Vidmeter’s goal is different. [from either link or voting sites] We’re trying to provide the most objective analytics possible about a video’s traffic and in doing that we’re able to determine what the most viewed videos of the day are. We don’t claim that these videos are “good,” but we do know that they are hot.

Scott from Unruly Media

We measure linkage and embedding in preference to views because views can just be bought or, worse, easily faked. For instance, you only need 30-50,000 views in a day to make one of YouTube’s most viewed pages or to make Vidmeter’s top 50. And we know agencies routinely fake out that many views as part of their seeding strategy….there are lots of smaller specialist shops doing viral work and it’s pretty wild west. We want to track social acts of recommendation and to find out which pieces of content are inherently viral and therefore likely to achieve high numbers of views regardless of campaign budgets or black hat techniques. We think measuring linkage and embedding is a good way of doing this. It’s still subject to spam, gaming and other forms of abuse, of course, but we find that pretty easy to detect at the moment. I’m not wanting to big up our method. It’s simply a question of what you want to find out.

Ultimately, our interest is in helping agencies to identify strongly mimetic strains of content and to map out the the routes via which specific strains of content are typically diffused. Put commercially, to provide planning and evaluation tools and to offer bespoke research and consultancy.

We added three new video sites to our roster last month – Metacafe, Dailymotion and Break.com – in addition to YouTube, MySpace and Google. To be perfectly honest, the incentive to track other video sites is not particularly strong. For instance, in February we saw 700 blog posts embedding or linking to Break.com videos. And 500,000 blog posts embedding or linking to YouTube videos. Linkage is much more skewed towards YouTube than viewing data!

So, as a marketer, how do you read these numbers? The views is very easy to explain to a client – how many people have watched their content , usually to the end. It’s easy to explain, it’s directly analogous to metrics they are used to when it comes to print or TV (ignoring the accuracy of those systems), it’s a metric that can be used across all the video distributors. It’s also more the default mode of exploration on the web. That is what YouTube and the other sites have bought, easy entertainment along a broadband pipe. Point and click, no technology knowledge required at all beyond being able to type youtube in a search box*. (something that Joost and other distribution clients need to think about). It can be gamed, and gamed easily. As a client, I hope you would never find an agency that would do that. As Bri says, raw numbers give an indication of who has seen the content.

On the other hand, if you know your audience is relatively tech-savvy, likely to have their own web pages (and I’m not talking MySpace) then links is another measure that is useful. It’s a percentage of those who watch you can take the content – and not all sites allow you to do that anyway. Linkage and embedding is a strong indicator of whether people loved your video enough to do something about it and want to share it with their readers. The choice of linking reflects on their online brand, what is in it for them- I can go and watch all the trash I like with no consequence but when I embed or link then I express something about myself that reflects on how I like to be portrayed. So linking is a reflection of this as well as engagement with the content.

So using both measures, (along with comment numbers) gives you different levels of engagement. Use them all to express the success of distributed content. If looking at 2 videos on a site with a similar view count but different link numbers, i would tend to regard the higher linked one as a better success.

And the two services? Vidmeter allows a DIY approach, giving you the numbers but look to be building out a commercial offer with more information as well. Unruly Media provide the top numbers and a commercial and consultancy service to help improve your content. Use as required!

Update: Further comments from Bri from Holt labs:

We do take the issue of “faking” views seriously, but it hasn’t been a problem. In order to register a view on YouTube, a view must come from a unique IP address. Knowing that, there are 3 ways to quickly register views:

1. Placing the embedded player on a website. Since registering views in this way still requires that unique eyeballs are watching it, we think this is fair. From your perspective as a marketer, you WANT to know if the video you embedded on a site is getting played.

2. Buying traffic. Through various ways a person can send lots of people to a YouTube page, however this still requires that unique people see the video so we also treat this as fair game. From your perspective as a marketer, you WANT to see if the traffic you bought actually watched the video so you want to know this.

3. Putting a hidden iframe to a YouTube page on another website. This method would surreptitiously allow a website with a lot of traffic to add a large number of views to a YouTube video without having people actually watch it. We consider this cheating, however there are three obstructions that prevent this:
a. A cheater needs to fake at least 70,000 views to get on the Vidmeter home page.
b. We have TWO HUMAN DATA ANALYSTS who scan the list daily to look for this
c. As you stated in the post, this isn’t relevant when tracking your own videos because you aren’t going to fake your own traffic.

In the event that a person cheats 1,000 views and then gets listed on YouTube’s most-viewed to receive 60,000 additional real views (as the case with “Spiders on Drugs”) we actually consider the 60,000 legitimate views as 60,000 people did actually watch it.

The proof that faking is not an issue is on our site. Thus far, no one has been able to point to a video that faked it’s way to the top. All the top YouTube videos are either from major providers, featured, or blogged about.

Also, I think it would be fair to note that blog links can be faked as well. The same way that SEOers would create fake sites with links to their own in order to “Googlebomb” and increase their page rank, a person could easily create multiple fake blogs and embed their videos in.

Bri makes some good points about how video views can be skewed and what they do to prevent that on the charts. However, i still feel that there is the possibility for some unscrupulous agencies to skew the results to present them to their clients – they do not refer to the charts at all, just to the video site with the view counter. This is in the same way that there are SEO agencies, buzz agencies and all other kinds of advertising agencies may do things to promote positive results. Look at Henriette’s comment for some examples of what is being done.

*that’s my default assumption of how the average person finds sites these days, after watching relatives. No need to know what the address bar is, just where the search box is and you type the URL.