Online Video Analytics – An IntroductionIn recent years the Internet has seen an explosion in the use of streaming video. This has simplified the process of getting video to consumers as well as giving content providers the opportunity to measure the consumer’s engagement. Online Video Analytics allows for content providers to quantify user engagement in new ways that should help them measure returns on their video investments.
Essential Online Video MetricsIf you host videos using your own flash player (or stream through Windows Media Player) then many different user activities can be recorded for analytics purposes, such as pressing the pause button or sliding the video position slider bar forwards or backwards. Some of the key video metrics (or KPIs) are:
These can be measured using a combination of media player events, methods, timer events and tracking code written in JavaScript or Adobe ActionScript. Video SegmentationFor the purpose of making measurements we can choose to conceptually split the video into a fixed number of sections, say 10 segments of 10% length. These segments can be used to measure which sections of a video have particular high end user engagement activity. We can also define another way of splitting the video into slots, where in this second method, we can choose a fixed lot length in time of say, 10 seconds “time sections” or “time slots”. These can be used to measure engagement in terms of fixed time slots of 10 seconds length. This type of measurement can also show the total playing time of groups or individual videos. We can further group the videos into categories such as chapters, channels (yahoo, msn, etc..), types (sports, sci-fi, etc..) or any other categorization the publisher may wish to use for analytics purposes. Video Engagement CharacteristicsA video engagement characteristic is a representation of the audience’s interactions with the video. Content providers can use this characteristic to rapidly see which of their videos are most compelling to their audience. It can also be used to analyze the strengths and weaknesses of a particular video. For the engagement characteristics we will define two types of audience video consumption: Positive Consumption Actions Negative Consumption Actions Positive consumption actions are ones that associated with a positive user experience. One positive consumption action could be the act of rewinding the video to re-watch a specific section. This indicates that the user saw something that they want to see again. Therefore each time an end user rewinds the video a positive consumption action is recorded for the section that is rewinded to. We can also decide that certain activities constitute a negative consumption action. One such example can be the act of fast-forwarding the video (or moving the slider bar) to a position further forward in a video. This can indicate that the audience is bored with the current section they are fast-forwarding from, and so for this case of fast forward, we use the section of video moved from, and not the video section moved to, as the grouping for negative consumption action. The positive consumption action reported by the analytics system then be used indicate which sections of a video are particularly compelling. Conversely the negative consumption action sections can be removed, reworked or edited to make them more positive. Videos and LinksAnother type of video analytics could be the correlation between video sections and clicks to links for product information. It could be deemed useful to measure at which stages in a video your audience was compelled to actively seek extra textual information sales or specification information about a product. This may also be considered as positive consumption as it shows active interest in the product, and such data can be gathered about audience behavior. ConclusionWe’ve just scratched the surface of the types of data content providers can mine from their online videos and why it can be useful. In our next post we will show the nuts and bolts of how to implement online video analytics using Yahoo! Web Analytics. 3 Responses to “Online Video Analytics – An Introduction” |
|
[...] How do you quantify the success of your online video? It’s been a conundrum for a lot of advertisers for some time. But Yahoo! is here to help. We’ve just launched a new Yahoo! Web Analytics Blog. In the blog’s second post, Yahoo’s Tim Hampshire helps you figure the ins and outs of tallying the results of your video efforts. [...]
Tim,
Great article… Great information…
When you speak about segmenting a video into percentages or in seconds, how are you tagging these segments? Are you adding onclick commands to each segment with custom fields in the YWA javascript? Please explain…
Thanks Tim
Hi Eric,
Thanks for the comments :)
The way the video segments or sections or chunks are being tagged is described below:
A javascript periodic timer event (every 1 second) runs a piece of code that checks 10% and 10 second boundary crossings – using total length of video from player api and current play position from player API;
Then if a boundary has been crossed then the code reports this boundary crossing data back to YWA data collection.
The data itself is recorded in action based custom fields for both the 10second chunk number seen 1,2… (labeled video_which_time_chunk), and a separate custom field for which 10% chunk 1,2..10 (labeled video_which_chunk) , as well as action based custom fields for which video , channel etc, and actions for both a 10 second boundary crossing event and 10% boundary crossing event.
Using these data a custom report can then created using groupings of video title, subgroup video 10 % chunk played , or subgroup 10 second chunk played; and metrics of corresponding 10% action or 10 second action, and thus report popularity of each individual chunk of each video title (also channel , genre etc can be added as groupings) .
A full description of how I coded the YWA Video source code js include file and html/js coding for embedded Yahoo YEP player and embedded YouTube Player examples will be appearing shortly in the next blog.
Kind Regards
Tim Hampshire
YWA