LOS ANGELES — When is a stream not a stream? When the streams are played by viewers using a plethora of different devices.

That creates a measurement problem for streaming platforms and their advertisers. But one company that has been monitoring streaming quality for two decades says it has the answer.

In this video interview with Zach Rodgers at Beet.TV’s Beet Retreat, Keith Zubchevich, President & CEO, Conviva, explains why old streaming sensors are being used to solve a new problem.

Pivoting to measurement

Conviva was founded under a different name in 2006 to measure the technical quality of streams experienced by viewers, in what were still relatively early days for online video.

“Conviva has been doing audience measurement for 15 years in streaming,” Zubchevich says. “We just happened to start the company doing quality of experience (measurement).

“As people were putting video on the internet, it was all about ‘how are people watching it?’ and ‘was it buffering?’, ‘was it clear?’, ‘was it loading?’, all the quality issues.

“But to do that, we had to measure every single audience session and we had to collect every second of the session so that we could show them their census-level audience quality by session.”

Sensors for sense

It turns out the same monitoring process can now be applied to measuring video audiences more generally.

“We’ve now been distributed to five billion sensors around the world, across the premium inventory, across premium publishers,” Zubchevich adds.

“So now that same sensor opens up and now starts to collect things around content ads and ultimately an ID. Now it’s servicing a different group in the publisher.

“But the sensor and the platform is already under-foot, and it’s been there for 14 years measuring everything that they’ve been wanting to see – but now it’s all about audience measurement for us.”

The company claims to measure three trillion real-time events every day.

Measuring media

Conviva these days offers media and ad sales planners insight into audience segmentation and deduplicated reach and frequency.

It can also overlay data from sources like Experian and Gracenote.

Conviva applies a household identity model to assign its Stream ID to each household.

“That stream sensor is collecting everything as it relates to the stream,” says Zubchevich. “The stream sensor is essentially like a funnel that sits in the video player, and it collects everything around the session.”

Viewer journey

It’s no longer just about quality of service.

“As soon as you click ‘play’, the sensor starts measuring the video session,” Zubchevich explains. “It measures every second, and it starts to collect everything from events, like if you have a buffering event, or content and ad metadata, so it shows what content you’re watching, genre, where you get to, your engagement time, even the ad breaks.

“As you go into an ad break, it will tell you, ‘when did the ad breaks start and what ads played?’ and that the viewer got all the way through and got to the end of the content.”

The company claims to make clear the “viewer journey” across both owned-and-operated streaming platforms and social media.

Normalizing for fragmentation

But collecting the data is just the start.

Platforms like iPhone, Android, web and so on all call and render video a little differently. That poses a challenge to effectively capturing video viewing.

So Conviva uses what it knows about the differences to “normalize” viewing before it is turned into insight.

Zubchevich says it’s all privacy-compliant because streamers own their own viewer.

You are watching coverage from Beet Retreat Santa Monica 2021, presented by FreeWheel, IRIS.TV, Samba TV, TransUnion & Warner Music Group. For more videos, please visit this page.

By Robert Andrews on    @robertandrews