Online Video Analytics – KPIs

We (and that means every online marketer) have seen an extreme interest in Online Video (streaming video) because it simplifies the process of getting video to the consumer and with streaming video, unlike downloaded video, the content (the video and any interaction with it) can be tracked. AND if any business, that be B2C or B2B invest in online video – they are obligated to measure the return of that investment!

First I would like to apologize for the aggressive title – as this is more about essential Online Video Metrics than KPIs – I think it is fair to say that these online video metrics affect your KPIs, but are not KPIs by themselves, unless of course if you are in the business of delivering video as a revenue model. :-)

9 Essential Online Video Metrics

  • Online video started
  • Online video Pre-roll advertisement started*
  • Online video core content started
  • Online video Post-roll advertisement started*
  • Online video positive consumption action
  • Online video negative consumption action
  • Online video ended
  • Online video played, percentage of total
  • Online video played, seconds

*pre- and post-roll metrics are of course only of importance if and when a online video monetization model is in place. For some of us these metrics does not exist.

Essential metrics suggest that there is indeed more exotic metrics — and you are right — but I stoutly suggest getting the essential metrics up and running, as in being able to collect, report and analyze on them before working towards any of those more exotic ones.

Furthermore, surrounding metrics such as “send to a friend”, “rate this video” and other Online Social Networking activities are separate metrics that are important, but not distinctive online video metrics.

Consumption actions can be pooled together (dependant on your users) in positives and negatives; such as e.g. fast forward, slide forward and stop as negatives and rewind, slide backward, pause and play and positives. You have to be very careful about using these as it can be difficult to derive valid insight; AND you might want to split them into different actions, however, I am personally a fan of pooling them together.

All that said; we have to differ between (from a technical tracking point of view) between two ways of serving a video online.

Simplified Online Video serving methodologies

  • Server Streaming (e.g. played through windows media player)
  • HTTP Streaming (e.g. played through your own Flash player)

Both of them (and their sisters) require completely different technical tracking methodologies and I actually posted a Web Analytics HACK on how to track server streamed online video not that long ago.

And before you fall for any bullshit, patent pending, we are the only online video analytics provider – marketing spiel, I would like to conclude that you can collect and analyze most of the above mentioned Essential Online Video Metrics in all of the enterprise analytics packages out there. :-)

Online Video Optimization comments

The above metrics are to be collected on a per video basis and you can aggregate per visit (session), per visitor or any other specific grouping (segment) you want to analyze. It is of course a given that when collecting any of the Essential Online Video Metrics, such as Online video Started, you MUST collect information about which unique piece of content (online video title or other identifiable content ID) we are dealing with. That said, here is a couple of suggestions on how we can work with the metrics:

  • Conversion rate segmented by Online video started COMPARED to the site average
  • Conversion rate segmented by Online video ended COMPARED to the site average

Where the above two suggestions immediately shows the success of using video for conversion purpose. Imagining a positive scenario where you have 1% site average conversion, 2% on those who started a video and 3% on those who finished it (expecting that the more a visitor consume the more engaged and interested he is).

  • Compare the “Conversion rate segmented by Online video started” between visitors who had a positive consumption action and those who had a negative consumption action.

Refining a positive attitude towards online video as an element for conversion increase – hereby getting further insight on the importance of questions like; does it matter if they watch the video to the end (or do they “just” have to engage positively with it).

  • Trend the metric “Online video played, seconds” grouped by videos

Creating insight on WHERE people fall off; in the first 5 sec. Or in the first 30 sec? – essentially getting to know how long you can captivate your specific audience.

  • Create traffic source goals and alerts for Pre-roll advertisement started and Online video Post-roll advertisement started

This of course creating instant insight into which traffic sources send you online video engaged visitors. And not only as a comment to the point above; whenever you look at “Video view abandonment” (did NOT watch the video to the end), it is critical to distinguish between whether we talk about complete site abandonment or just that specific video.

But in conclusion; this post is more about creating awareness about the Essential Online Video Metrics and then making sure that you have a Web Analytics system available where you can utilize these metrics in retrospective – becoming able to answer marketing questions in regards to Online Video.


  • Ori

    Terrific analysis and advice, thanks for putting together!

  • Anil

    Dennis Great Post
    I have a admit that i’ve now learnt two new things! that’s consumption metrics and that really make a great KPI!

    Thanks again for the knowledge sharing, appreciate it :-)

  • Anil

    arghh!! i meant “consumption actions” really make a great KPI :)


  • Dennis R. Mortensen

    Hi Anil and Ori,

    I think the list is a great starting point – if one wants to do a bit more that just look at the number of online video “views”. Thank you very much for the comments/.


  • Eran Ben Sabat

    Hi Dennis

    Thank you for a great post. If you don’t mind me saying, I feel that there are more key metrics that I found to be important as well. Metrics such as buffering time and most repeated videos etc.
    I could not agree more with regards to the “video abandonment view”. When you run an online video monetisation it is very important to know where to put your advertisement. On another note, I think that in time the industry might need to think about bounce rate in the context of streaming.



  • Dennis R. Mortensen

    Hi Eran,

    I really like your comment about “Video Bounce Rate” and it will be fun to see how we could define it. Food for thought!


  • gbm2420

    My site creates a lot of video content (Flash progressive download) that gets embedded elsewhere. We use Google Analytics, but it seems useless for tracking embeds. Any suggestions for getting basic data (number of initiated plays, which URL a video is on) on embedded videos? Thanks.

  • Dennis R. Mortensen

    Hi gbm,

    >>elsewhere. We use Google Analytics, but it seems useless for tracking embeds

    Google Analytics actually does have a very ok “EVENT” model that will give you the opportunity to collect the data you are asking for. I would go explore that if I were you.

    Cheers Dennis :-)

  • gbm2420


    Thanks for the response. The event tracking works fine on our own domain, but we can’t get it to work if our videos are embedded elsewhere (MySpace, someone’s blog, etc.).

    Here’s how a technically savvy friend explained it to me: “When a video is embedded, the video itself has no control over the HTML of the page on which it is embedded, which means that trying to load and use an Analytics Javascript object from inside of Flash will fail, since it will not exist or will exist for someone else’s domain. Including the Analytics code as part of the embed code itself will not work either, since most sites strip Javascript from this code.”

    Does this look accurate to you? I’m desperate for a cost-effective solution. Thanks.

  • Dennis R. Mortensen

    Hi gbm,

    I am sorry to say that your friend is somewhat right. Content consumption outside your domain is not something that is measured easily. It of course all depends on WHERE and HOW your content is embedded, but a video of yours on YouTube or a picture of yours on flickr cannot just be tagged.

    I actually described this issue in my post about the Online Business Measurement Quadrant.



  • Pingback: Youtube Insight, první krůčky video analytiky • Dlouhý chvost

  • Oren Harnevo

    Hi Dennis,

    Indeed, great insights into video analytics. Especially the part you speak about video as a conversion tool.

    I am the CEO and founder of EyeView. EyeView’s sole purpose in life is to measure the effect of video implementation in online businesses to their conversion rates. All the discussion on: “video bounce rate”, “time viewed”, “video popularity” and much more are analytics we deal with on a daily basis.

    We also built technology around it and we have the first online software as a service for optimizing conversion rates through video.

    I’d be happy to share more details. You can view our blog:


  • Gerry White

    @ gbm2420 – I know this is a bit late, we are trying to do something similar as well so excuse me if this is wrong as it could be – the NEW google analytics flash plugin will allow you to do it all within the flash – (look at the Google Analytics Flash Blog) – and this should allow you to capture the information you are looking for …

    @ Dennis – no chance your doing this (flash only type solution) into Index Tools / Yahoo Analytics ?

  • Dennis R. Mortensen

    Hey Gerry,

    The new Flash opportunity from Google Analytics is indeed a great and very useful new feature, that will let you do essentially native flash tracking. AND to answer your question we are about to do the same, so IT IS ON THE ROADMAP :-)

    But, when looking at the comments from @ gbm2420 – I don’t think this actually solves anything, as his issue is around the fact that media (video) is consumed outside his area of control (thus no ability to tag it, native flash or not).

    Thank you very much for your input.


  • Kedar Thakar

    Hi Dennis

    Great post. I have some thoughts -

    It seems that for ad-funded online videos, the business model is similar to TV broadcasting. Advertisers want to reach as many customers as possible (unless site owner has detailed customer subscription data that can be anonymised and segmented and shared with advertisers).

    So advertisers would be interested in most watched videos – most views, unique views, unique viewers. But I am sure there is time dimension – max. unique viewers watching a video at a particular time of day. Is this important for advertiser? OR they are ok with just most popular videos? I am comparing here with prime time of TV. If unique visitors stats are generated for every video on time dimension, will advertisers pay a premium for a particular time slot?

    Of course, I am assuming here that ad insertion and delivery system can insert max. bid ad into the video stream dynamically.



  • Dennis R. Mortensen

    Hey Kedar,

    Thank you very much for your thoughtful insight, and I don’t necessarily disagree with you, I do however not think that the above points change the actual online video KPI’s that one would have in place.

    In regards to premium time slots, and increased revenue from them, seems to be mostly about the fact that, that’s where the most volume is. I believe the buying pattern is a tad different on the web – or at least with the advertisers that I have been working with. I see some using time slots as part of an optimization process, but thats more to make sure they reach the right segment, and usually due to weak advertising systems where there are little or no targeting opportunities.

    my 2 cents anyway

    d. :-)

  • Leticia E. Colon

    Dear Dennis,

    Thanks so much for elucidating online video analytics, its metrics and complications.

    Looking forward reading more of your related blog posts and white papers.


  • Pingback: Yahoo! Web Analytics Blog » Online Video Analytics – An Introduction

  • betsy kuriakose

    Hi Dennis,
    Do you have information on KPI for tutorial videos? Are there any industry standards for this type of video?


  • Dennis R. Mortensen

    Hi Betsy,

    I have not come across specific research on tutorial videos, neither have I been able to actively work with this type of videos from a tracking point of view. Sorry! DO share when you conclude on setup though.

    d. :-)

  • http://none doug firr

    Hi Dennis,
    Any suggestions for metric(s) that would show propensity to watch an additional video?

    Put another way I’m seeking actionable insights on how to increase our 2+ video views per session.

    Any ideas?

  • Dennis R. Mortensen

    >>Put another way I’m seeking actionable insights on how to increase our 2+ video views per session.

    I would spend my time on “Content Selection”. Such as upfront pool and the end of play, here-is-what-we-suggest-you-see-next, recommendations.

    d. :-)

  • Pingback: Most Popular Blog Posts of 2008