Exclusives

By Richard Hayami Z’Graggen, VP-Head of Experience Design, LVL Studio –

In our world of 500-plus channels and millions more video web sites, it’s hard to get noticed. And for our audiences, it takes a lot of effort to find that one video that they didn’t even know they wanted, but which is, in fact, exactly what they do want. Recommendation engines are supposed to solve it all, but here’s where it gets interesting. Add two more essential ingredients and you have a combo that just might solve all our discovery needs.

Let’s start with the recommendation engine because it takes us most of the distance. By tracking what you tune in to, and with some fancy algorithms and data sets, it can propose very good suggestions on what you are apt to like. Netflix turned the 70/30 ratio of new releases to older cheaper content on its head through multi-million dollar algorithms. Digitalsmiths analyzes videos and builds robust data sets around each asset to make even the most obscure connections possible. And Jinni contextualizes behaviors around time and mood, understanding that we don’t want the same genre at every hour of the day.

These and other recommendation engines are very good at pro-posing content based on past behavior… with some big data crunching to build relations between assets. But with a shared device like the family room TV, there is an essential flaw: your viewing habits might be mixed in with those of all members of the household. So in a family with young kids, you may tire of seeing all the Disney recommendations, especially at 10 p.m. after the little ones are tucked away. Your household did consume all those animated films, so it’s not the fault of the recommendation engine. But still, “Where’s my HBO?” you say.

User profiles: because it’s all about you

Add the ability to specify who is watching and suddenly the suggestions become all about you. Netflix lets us create individual “users” but many legacy TV systems are lagging. Good systems would go further – allowing multiple viewers to signal their presence. And we can expect to see more fluid interfaces on set-top boxes that react to the presence of individual nearby mobile phones, adjusting automatically to the comings and goings in the shared space.

It’s obvious of course: the more relevant the suggestions are to the people in front of the screen, the more useful they are. And the more useful, the more they will lead to transactions.

So user profiles are important but we’re still not addressing a third factor that can elevate suggestions based on the past to predicting what you will want before you even realize it.

Curation: because of course that’s what you were thinking

Every October 31st is a circus of costumes, candy and costume par-ties. In the week leading up to Halloween, operators would be negligent if they didn’t load up their featured lists with classics like It’s the Great Pumpkin Charlie Brown, A Nightmare Before Christmas, and the slew of horror movies everyone wants to watch. It’s a great time, by the way, to dust off the older assets in the library because that’s what audiences are looking for.

The same goes for other events like Thanksgiving, The Superbowl, Valentine’s Day, and yes, prom night. A search on this last topic found not only movies, but TV episodes too. Who knew this includes That ‘70s Show, Vampire Diaries, and even My Name is Earl with the uncomfortable theme of Earl organizing a prom in prison. I didn’t know this episode, and we should assume the average user won’t have the patience to spend time digging around for jewels like these. An hour spent by a curator/programmer can save millions of people from having to do the same. It’s a powerful equation.

Curation is not just about all the regular calendar events. It can also respond to current affairs, such as war, economic collapse, and celebrity news. On comedian/actor Robin Williams’ death, fans wanted to pay homage and watch their favorite movies like Good Will Hunting and Dead Poet’s Society. Good content marketers package the collections to make it easier for their customers to find. Can algorithms do this? The answer is yes, with enough data points and massaging and weighting. But the human touch can be quite effective, especially when it’s an expert in content marketing tuned to cultural markers and the mood of the moment – the zeitgeist.

Three legs of a stool

Recommendation engines, user profile management, and curation are like three legs of a stool. Take any one away and the seat is not so grounded. But if you have great data-driven algorithms to make intelligent connections, along with user profiles to make it relevant to the individuals in front of the screen, and finally curation to capture the predominant trends of the moment, you will go a long way to maximizing the impact of your content suggestions.

Click here to translate this article
Click here to download the complete .PDF version of this article
Click here to download the entire Spring 2016 M&E Journal