This is the final installment of a four-part series.
1) Make basic content recommendations
Next, let the machines do the work. Ooyala Discovery is one of the many content recommendations technologies that will work with your existing video player to suggest new videos based on content velocity, viewer behavior and usage patterns. Ooyala Discovery gave publishers an average 10% lift in viewership
in our most recent Global Video Index.
Ultimately, we find the best personalization approach is a hybrid of machine-learning and manual curation.
2) Make session-based recommendations
With a bit of knowledge about context, you can update recommendations dynamically. At Ooyala, we look at session-oriented details like (1) which device the viewer is using, (2) the time of day, and (3) the length of the clip.
Let’s say a viewer visits your site on her smartphone at 9 am on a weekday. A reasonable conclusion is that she is on her commute to work and probably wants short-form video nibbles rather than a long-form meal. Recent Ooyala Global Video Index data concurs: 66% of AVOD views on phones
are short-form clips (0-5 minutes).
Source: Ooyala Global Video Index Q1 2016
Let’s say that same visitor arrives on your site on their iPad at 9pm that night, which according to the same Ooyala report is the most popular consumption period for tablets
. Now she’s probably not pressed for time and wants something more substantial.
That’s why it doesn’t make sense to keep a static site that features the same videos for all devices all day long. With just a few data points you can make your site fluid and ever-changing to match shifting viewer needs.
3) Make data-based recommendations
Data reveals viewer preferences in many ways:
Data Management Platforms (DMP)
Ever logged into a site using your Facebook, Twitter or Google credentials? That site probably learned things like your location, likes and social behavior. Viewer logins like these can benefit you as a publisher and be useful and convenient for the viewer.
Another way to know what viewers like: ask them! Make it quick and easy once they’re logged in, and viewers are usually happy to tell you what you need to give them a better experience on your site.
With data in hand, group your viewers into interest segments, like female viewers aged 17 to 25 who like sports. Then it’s easy to highlight each group’s preferred content.
Start by asking basic business questions like these:
-- Which videos perform best and why?
-- Do some genres perform better than others?
-- Which videos have the highest drop-off?
-- What is the ideal clip length for each video type?
-- Are there certain geos we should target?
Locate the data that will answer those questions. Create a central repository for that data and decide on a cadence to update it. Then do so -- consistently!
You’ll find that reviewing these important metrics regularly will spark ideas that turn into relevant experiments that lead to greater success with your viewers. There’s no shortcut here: You have to experiment constantly and see where the data leads.
Make sure you define metrics in a standard and consistent way across all platforms. If you define a video “start” differently than a video “play,” be sure to stick with these definitions across all devices and processes. Use industry-standard terms whenever you can.
We find it’s helpful to create a dashboard where stakeholders can get up-to-date versions of key metrics. If you have the resources, make it an automated dashboard. Simply outline the metrics required and map them to the APIs for each data source.
There are many ways to reach success with video and AVOD, and every publisher’s strategy is unique. Above all: “always be experimenting.” We hope these notes will give you some great ideas for your next set of experiments and, in turn, your next set of successes!
For more information about Ooyala’s Strategic Media Consulting Team, contact us: firstname.lastname@example.org
Download “Four Stories of Video Success: Best Practices for AVOD”, which includes case studies about these recommendations, here.