Providing a TV-style linear stream is a challenge for any media company, but especially for web native publishers who more used to posting content that their users can consume on an à la carte basis. Going linear requires you to have a continuous stream of content in the pipeline, all of which must be produced/acquired, scheduled and monetised. Then there’s the fact that most web publishers own short-form content, so stitching together five minute videos into a linear stream would be a costly and time consuming operational headache. However, Iris.TV, a Los Angeles-based start-up just might have a linear solution that works for the web.
Iris.TV’s technology analyses a publisher’s library of content and uses the metadata to create a personalised linear stream. Put simply, it’s dynamic video recommendation, but the content plays automatically. Robert Bardunias, CRO at Iris.TV, explains,”‘We take an archive of content from a publisher and connect the right piece of content to the right viewer, and we do this dynamically to create a linear viewing experience. So if you think about clicking on a video asset and watching a pre-roll ad, followed by the actual content, we will select the next most appropriate piece of content and stream it automatically right after that first piece. This turns the viewing experience into one where the user has to opt-out rather than one that is dependent on the discovery, searchability and location of the content. So we’re an add-on to your content delivery infrastructure.”
Everyone a Winner?
The tech sits on top of whichever online video platform (OVP), content management system (CMS) or content delivery network (CDN) the publisher is using. So on the business side of the fence, in theory everybody wins: publishers and advertisers get more inventory (one case study suggests they can expect approximately 47 percent more viewing time per user), the OVP/CMS/CDN gets to deliver more content, and the ad tech vendor gets to serve more ads.
But what about the users? How do they respond to on-demand content suddenly becoming a dynamic linear stream. “I wouldn’t say we’ve had negative feedback,” says Bardunias, “But with any new product the users need time to familiarise themselves with it. It’s going to take time for some users to realise that another piece of content will flow, although we have designed the buttons in such a way that they’re similar to buttons found on social networks and elsewhere online, so we’ve used things like thumbs up and thumbs down. Once you get users used to it – which usually takes a month or two – that’s when you start to see real increases. During the first three months there’s a certain amount of QA to be done on the system, tweaking the way the recommendations are done and how the metadata is processed for different locations etc. One approach might be for the publisher to arrange for a pop-up to appear explaining to the user that they’ll have content recommended to them.”
Iris.TV insist that the set up process is relatively straightforward, “We set up some automatic API calls directly into your OVP, which enables us to capture the video metadata, so when you add an asset to you library, we will ingest the metadata into the system and do a couple of different things with it. Firstly, we’ll clean the data and make sure there’s no imperfections, before using data mapping techniques to create a robust data set for every asset. So you’ll have things like time, date, the different people featured in the content, the context, the source etc. Then we can also look at what ad networks the publisher is using and how their sales are working, and deliver all of that metadata back into the system. Then, when when a video is selected by a user, it can pull up an asset id or a small calling card for us. We will then do the processing on our end and select the most relevant piece of content to send back to them. ”
The result is a personalised viewing experience, so if you’re watching a video about Arsenal, it’s possible to tweak the parameters according to a publisher’s requirements, so the next video might be about Arsenal or another Premiership team, and you could also stipulate that the next video should be one created in the last day/week/month/year. The technology can also pick out your duds, as Richie Hyden, COO at Iris.TV, explains, “If for example you come across a piece of content that for whatever reason cannot be monetised, we can create a rule that ensures that ‘for all subsequent views, do not recommend any unmonetisable content, thereby helping them not only generate views but also contributing towards the publisher’s bottom line.”
The potential for partnerships with the data-driven ad tech world are obvious, and Iris.TV are keen to explore the possibilities. “For example, as we collect information about a user, we might be able to come up with a smart way to ask what the value of advertising really is, so if a publisher was getting $15 CPM, we might be able to get it up to $18 or $20 with full transparency. So if you imagine being able to customise the experience across mobile and desktop in the way that that linear experience is targeted with advertising and content,” added Hyden.