Commentary

Challenges To Linear TV: Q&A With IRIS.TV's Field Garthwaite

by Charlene Weisler , Op-Ed Contributor, February 23, 2017

The need to discover new content has grown as both content choices and delivery options continue to increase. IRIS.TV boasts that it uses “artificial intelligence and adaptive machine learning to surface the most relevant video for each individual site visitor on desktop, tablet, mobile phone, even if the video is delivered OTT,” according to Field Garthwaite, co-founder and CEO of the company.

Garthwaite has an eclectic background, including experience at data-oriented companies like Rubicon Project, along with content creation as the assistant editor for the Emmy Award-winning PBS documentary "Girls on the Wall,” to a gig as research assistant at Universal Pictures. His current company just came off a strong 2016 by expanding into eleven countries across five continents including Canada, the UK, Australia, New Zealand, Japan, Panama and Chile.

Charlene Weisler: What are the biggest challenges to linear TV today?

Field Garthwaite: Consumers receive hundreds of channels from their cable providers, and the latest research shows that they watch less than 10% of the channels. But even then discoverability is a problem for networks and other video providers.

Meanwhile consumer viewing habits are trending toward on-demand and TV-centric services like Netflix, HBO Go, Amazon and Hulu.

So now linear TV programmers are competing with websites, apps and social media platforms that utilize interactive recommendation engines like IRIS.TV.

It is a big leap from traditional TV to personalized viewing experiences with business rules to guide the video viewing experiences. Traditional TV will not go away, but it will merge and become more interactive as consumption continues to move to digital platforms.

Weisler: How does IRIS.TV content curation work?

Garthwaite: Let's say the most compelling next video for a particular individual is sitting in the publisher's vault, untouched for months. Because it knows that video will be of genuine interest, IRIS.TV’s personalization engine, Adaptive Stream™ goes and finds it and "programs" it into a continuous stream (teasing it in a preview screen in the corner of the current video.) Suddenly, a video that the publisher hasn't monetized in a long while generates revenue. IRIS.TV can also provide “Netflix style” content recommendations to support user content discovery.

Weisler: How do you use data?

Garthwaite: To optimize video assets' discoverability, we optimize data structure and taxonomy by ingesting asset metadata. We also track viewer behavior and each video's performance so that publishers know what content works across a variety of parameters such as category, device, time of day, and in-stream (knowing what assets generate greater follow-on viewing). All of this happens in the background, automatically.

We also utilize data tools to help customers learn from their audience engagement and consumption patterns. When the Cubs won the World Series, every sports customer of ours had a dozen or more videos. But the reality is that one of those videos outperforms all the rest, and by helping our customers be more data-driven, we enabled them to put the best-performing video on their home page or on the trending article on the Cubs win.

Weisler: How does your company measure TV?

Garthwaite: We look at viewer behaviors such as the device the user is using, what time they are engaging, how long they’re engaged, bounce rate, geolocation, demographics, as well as market data, including advertising spend and revenue.

Weisler: How will viewing and measuring TV change with connected TVs?

Garthwaite: TV sets remain the number-one source for streaming TV and video content in America. If publishers invest in innovative video platforms for their channels and apps, connected TVs will enable media companies to personalize video programming using machine learning and give viewers the “sit-back-and-watch” experience that they love.

Connected TVs would make it possible to use data sets to structure and organize TV programming categories such as news, sports, reality TV, documentaries, and to pull content from top sources based on the individual consumer’s preferences. This will also clearly have a positive impact on their ability to get the right commercials to the right viewer.

AI-based personalization essentially lowers costs of presenting customized video programming and increases advertiser revenue, a win-win for video publishers.

streaming, tv, video

6 comments about "Challenges To Linear TV: Q&A With IRIS.TV's Field Garthwaite".

Check to receive email when comments are posted.

Ed Papazian from Media Dynamics Inc, February 23, 2017 at 1:14 p.m.
Charlene, I continue to find comments like the greatest challenge to "linear TV" is that we get 200 chanels but only watch less than a tenth( 17? ) of them a major "challenge" for this medium rather amazing. Over longer than one week time intervals, the average person---who, btw, lives somewhere in south west Indiana---- watches----or samples---- 25-30 channels and heavy TV viewers who account for the lion's share of TV viewing, watch considerably more channels than the averages suggest. But letting that pass, so long as total "linear TV" usage---live plus delayed----remains at around five hours per day per person, why is this an issue? Do we really believe that if the average person can get 200 channels then he/she should watch most---or all--of them on a regular basis? That's silly.

As for "connected" sets, it's fine to speculate on their potential, however, merely having one---as we all will, eventually---doesn't mean that they will be used five hours a day per person to watch content in a "connected" manner. Current stats tell us that despite fairly high "connected" penetration levels, most viewing still takes place the old fashioned way. Whether this will change remains to be seen. People will go where the availability of worthwhile content takes them.
Reply

John Grono from GAP Research, February 23, 2017 at 1:55 p.m.

When I started in TV measurement in 1991, people watched around 10% of all channels. So what has changed apart from the sheer volume of available channels?

There is massive fragmentation in the supply side, but nowhere near that amount of change in video consumption on the demand side.

Charlene Weisler from Writer, Media Consultant: WeislerMedia.blogspot.com, February 23, 2017 at 3:56 p.m.

Hi Ed,
I don't want to speak for Field but part of his answer regarding discoverability is certainly a challenge when there are more and more viewing source options.

Charlene Weisler from Writer, Media Consultant: WeislerMedia.blogspot.com replied, February 23, 2017 at 3:59 p.m.

Hi John, In my opinion the ratings erosion in linear that many ascribe to fragmented viewing choices is a challenge. Beyond linear, the issue of completely measuring cross platform and thus get credit for all airings is another challenge.

John Grono from GAP Research replied, February 23, 2017 at 4:21 p.m.

Totally agree.

My comment was primarily in reply to Field's remark that "the latest research shows that they watch less than 10% of the channels (they receive)."

It has basically been thus for decades.   According to data Ed has previously provided, the US linear TV is holding up better than here in Australia.   The average person dipped below 3 hours a day for the first time two years ago (it was 3 hrs 14 mins when I first started in 1991).

And yes, cross-device and cross-platform is a big one.

Our broadcasters release "VPM" (Video Player Measurement) utilsing a plug-in via their websites, so any first-party streaming is being captured.   It is then converted to 'Average Minute Audience'.   This has been beneficial in educating the market that big streams <> big audience.

At the moment the system assumes one person per browser (i.e. the 'lowball number) and has no demos but you have to start somewhere.   It is also not de-duplicated temporally or by device but the numbers are low so we have time to work on that.

This can be tracked as it is a 'sub-set' of the universe where the content is 'known'.   Moving wider into the content world where the video content is unkown apart from the URL string or bitly, and the streaming structure of the video player can change without notice is a huge challenge.   There is a MASSIVE need for video (and audio) centralised meta-data on a global basis.

Doug Garnett from Protonik, LLC replied, February 23, 2017 at 4:51 p.m.

Thoroughly agree, Ed. This kind of weak logic is often accompanied by a pitch for some shiny new thing. And instead of offering some value in the shiny new thing, the pitch seems to need to tell us what exists is bad.