M&E Journal: Hail Hybrid! Welcome to Version 2.0 of the Recommendation Engine
By Matt Kollmorgen, VP Digital Strategy, Global Media Practice Lead, SoftServe –
Media companies focused on creating digital viewing experiences are standing at a fork in the road in regard to recommendation engines. These engines are a critical component of any user experience — increasing engagement, usage and time spent with content or products as well as brand. Frontrunners in the space have put a lot of thought and care into creating their engines and are benefiting from the results. However, most recommendation engines fall short, as does their value to the customer and the brand.
With all the abilities of technology to provide highly personalized experiences, it is not enough to simply remember what viewers have watched and present them with a generalized listing of “most watched,” “recently watched,” and “what others watched” categories. Yet this is how most recommendation engines are built.
Curation and supervised machine learning are proven methodologies with less risk, less time and less cost. These might be good short-term business reasons, but are not aligned to a forward-thinking, viewer-centric approach. And to compete today with higher customer demands, and increasing competition for viewership, businesses need genuinely put (and not only speak about the importance of placing) the customer first.
If looking for the business benefits, one needn’t look far. Financially speaking, the most profitable recommendation engine today lives at Amazon, a company that understands consumers will never be satisfied with the status quo. What is shiny today will be dull tomorrow so the only way to compete is to keep growing the engine.
This truth compels CEO Jeff Bezos and company to raise the bar perpetually, which in turn leads to greater retention and new business, which in turn leads to greater reach and advocacy, which results in greater profitability.
Just because a viewer has already paid for all-inclusive service with a subscription (PPV buys aside), it is not good enough to simply give them what they ask for. It would be far better to truly know the viewer, to anticipate his or her wants — based on more than viewing history — and go above and beyond to make new suggestions, the way a trusted friend would.
Idealistic? Definitely. Possible? Absolutely.
The right approach is simply a matter of being dedicated to customer experience excellence. This means considering and leveraging higher-level technology and methodologies available to develop a recommendation engine that acts more as a concierge rather than a coordinator.
If audience is king, then providing remarkable differentiation with a recommendation engine that knows the viewer as well, if not better than herself, should be the objective.
Building a truly personalized recommendation engine requires an approach that utilizes not one, but multiple learning processes. It should start with credible and perpetually-tested algorithms and, ideally, a containerized approach.
There is nothing standing in the way of an enterprise media company from delivering hyper-personal and proactively intuitive recommendations — if it is willing to think big and start small, today.
A smart(er) engine
The ideal road for recommendation engine development takes a hybrid approach. It leverages the strengths of both supervised and reinforced machine learning (ML) and adds an additional layer of artificial intelligence (AI) for perpetual self-learning. Finally, it combines curation data with O.C.E.A.N. (Open, Conscientious, Extrovert, Agreeable, Neurotic) personality trait data for customized tonality in messaging, sentiment and experience.
Such an engine was recently adopted by a news, entertainment and lifestyle curation website owned by one of the world’s largest IT companies. After only initial integration, readership increased by 17 percent. That’s a massive first step in the right direction given the hundreds-of-millions of adults aged 18 to 60 who visit the site on a monthly basis. With continued optimization and learning, the future looks bright for genuinely personalized recommendations that lead to pivotal moments in brand affinity building.
We’re proud to say that we built this recommendation engine specifically for enterprise media companies and the retail brands they serve. It is containerized, multi-faceted and agnostic, with hyper-personalization capabilities for both new and existing engines.