Deriving Value From Discovery Insights

The Discovery Insights information in Ooyala IQ can provide you with valuable information about your users and content. The content below describes how to derive value from the different Discovery metrics.

Discovery Clickthrough Rate (CTR)

Click-through Rate (CTR) is probably the most important KPI you can use to measure Discovery impact. Discovery CTR measures the effectiveness of Discovery recommendations by capturing how many times the display of the Discovery screen results in additional video plays. Discovery CTR is defined as the ratio of the number of clicks on Discovery screens to the number of impressions of Discovery screens. For example, if the Discovery screen is shown 100 times and a video was selected to play 53 times, a 53% CTR is recorded.

Discovery has a default countdown of 10 seconds for its auto-play feature, but can be configured to have a longer or shorter countdown. The countdown timer gets suspended if the user ‘leans forward’ and browses through the Discovery end screen to select a different video than the first one. However, we find that the majority of consumers prefer to lean back and let Discovery play the next video for them to watch. You will find this reflected in the Discovery End Screen CTR excluding Auto-plays metric.

The Discovery End Screen Impressions metric is also provided to give you an accurate count (when combined with the End Screen CTR) of how many plays resulted. Based on the average CPM for ads on your site and your (pre-roll) ad load, you can easily arrive at the ad revenue impact of Discovery.

Discovery Lift

Discovery Lift is a secondary KPI that helps you understand the impact of plays attributed to Discovery in comparison with organic plays on your site, the two adding up to become the total number of plays. An organic play is a video having been played by the user from any content carousel on the site except a Discovery screen.

Note: Discovery plays from a carousel using a Discovery API for recommendations are counted as organic plays, the metrics for these currently not being reported via IQ.

Discovery Lift measures the percentage increase over organic plays that Discovery brought to your site. For example, if the number of organic plays is 100 and Discovery brought an additional 15 plays, that would be a Discovery Lift of 15%.

The Discovery Lift may help reveal how users find and consume videos on your site. A site with a lot of carousels from which the user may choose a video is likely to have low “Lift” (and low CTR too) for each carousel (Discovery included), although Discovery does possess an advantage with auto-play.

A number of factors impact the Lift numbers, including average session length, how users find content (through sites that syndicate your videos or directly on your site, etc.), and length of videos, among others.

Similar to how the CTR metric is broken down by end screen (with and without auto-play) and pause screen, the Discovery Lift metric is broken down that way too to help understand the contribution of each Discovery experience element.

Discovery Time Lift

We believe that a good recommendations engine drives engagement, not merely clicks. The Discovery Time Lift metrics add a time nuance to the Discovery Lift metrics. Time Lift measures the percentage increase in time spent watching videos while Lift only measures the percentage increase in play counts. In other words, Time Lift helps measure the increase in average session length.

For example, if 100 videos were played organically and 20 through Discovery, that would be a 20% Lift. However, if the time spent watching the organically-played videos was 2 minutes each and that for Discovery-played videos was 2.5 minutes each, then Time Lift would be (20*2.5/100*2)*100 = 40%.

It is quite common to see a Discovery Time Lift that is higher than Discovery Lift, indicating that the impact on engagement by Discovery was higher than the Lift metrics might suggest. As with Lift, a number of factors impact the Time Lift numbers, including average session length, how users find content (through sites that syndicate your videos or directly on your site, etc.), and length of videos, among others.

Discovery Plays and Discovery Leads

Discovery plays and Discovery leads are available per asset in the catalog, and measure how many times the video was played through Discovery and how many times (when played organically) it led to a Discovery play.

The Discovery Plays metric is easy to understand, as it counts how many times a video was played after a recommendation by Discovery. You can compare this metric against the IQ metric of total plays to understand the impact of, say, a Discovery content promotion for that video.

Discovery Leads can be a valuable metric to help guide your homepage editorial strategy. Discovery Leads measures how many times an organically played video resulted in Discovery plays.

To understand the Discovery Leads metric, assume that a video (Video A) is promoted on the homepage of your site. When played directly from the homepage (or other part of your site or that of your syndication partner), that would be an organic play of Video A. The Discovery end screen, when presented at the end of all those plays (say, N times), may result in M (M lesser than N) additional (Discovery) plays, this being the Discovery Leads metric for Video A.

If you see the Discovery Leads metric being low for a heavily promoted video, then the inference follows easily: Video A isn’t a great lead-in for subsequent plays. When you compare this metric against the other columns, you may see other such observations which can come in handy for editorial decisioning.

You can also filter the Discovery metrics based on the labels that you use to categorize your content in Backlot, or filter down to specific players, thereby opening up additional layers of engagement growth insights.