Targeting individual website visitors with relevant personalized content is a tough task. Making that happen in the blink of an eye against terabytes of historical data is a challenge that few so much as contemplate; however, that’s exactly what we do at Youneeq.
One of the ways I like to describe the behavioral targeting of our services is simply to structure our algorithms in the form of simple questions to each individual, and from the answers look for the best indicators of what that person “might like”. We’ve built a system that can ask millions of these questions every second, and is flexible in how it records what might help to provide those answers. Technically this was no small undertaking, and in order to continue to be a leader in our field we tackle new research challenges every week.
Personalizing and placing content that appeals to the nuances of an individual is great, but only the start of the challenge. One truly offensive suggestion can have a ripple effect that can counter the impact of thousands, even millions, of compelling content suggestions. Consider Facebook’s mood experiment last year, whereby the company attempted to drive both happy and no-so-happy thoughts among the people unwittingly influenced (i.e. without their consent, express or otherwise). Whether I received the happy content or not, I for one wouldn’t be cheerful afterwards if I discovered a company used me in such a way. How to balance this out is only in part a technical challenge, taking a considerate view to people’s sensitivities goes a long way to building a truly compelling product.
