Delivering Real-time Personalization at Scale: Not Just a Pipe Dream
By INDG
If you tell 100 marketers: “Personalization is the next big thing”, chances are that you will get a few yawns and even some laughs. Everyone has been hearing the same story year after year. And yet, still personalized content has not reached its oh-so promising potential.
According to a recent McKinsey research study, only 15% of CMOs are confident in their path towards personalization. Interestingly, the overall belief that it is crucial to implement personalization within the marketing strategy is predominantly shared. Therefore, it is no longer a question of “if”, but rather of “how.”
What the research tells us is that personalization can increase conversion, improve customer retention, and even can make a company's marketing useful. Moreover, as the world of machine learning keeps advancing at a rapid pace, the benefits for customers become ever so more significant. AI, combined with the wealth of data, allows for more advanced logarithms in detecting and responding to customer behavior and emotional cues. As a result, (deceptively) human and personal experiences can be created - from recurring customers receiving a reminder for their upcoming anniversary together with ready-to-buy flowers (event-based triggers), to recommendations for vitamin C after a consumer purchases cough syrup (context-based triggers).
Another McKinsey study shows interesting insights into privacy concerns, with over 40% of consumers unconcerned about data that captures content consumption, purchases, and online searches. Most of the reluctance has to do with the kind of data capturing that is considered intrusive, such as face and voice recognition. These do much more than just classify you as part of a group that is likely to be interested in something, because they are considered to be unique aspects and therefore touch too closely upon your identity.
Those who agree that personalization should be integral in a company's marketing content, are faced with the question; why don't we see it at scale around us yet? Demand Metric's research uncovered reasons supporting the persistence of non-personalized content - which sheds light through two different insights. The results show that the need for personalization and understanding of its benefits is mostly appreciated and recognized, even though a minority is discouraged by failed prior attempts or by other reasons.
However, the biggest challenges that seem to be the main barriers have to do with limits on technology and data. In a world where technology is constantly allowing for new opportunities and where data is more readily available than ever before, the question has to be asked; are these reasons still valid?
It is impossible to doubt that Facebook, Google and Amazon would not have the data that can be used, let alone that this would be remotely the case with their Chinese equivalents. Perhaps the only concern is that of privacy, beyond the legalities of it, as that may skew consumers' faith in automated recommendations. Following the research, though, that is currently not what is holding back personalization.
The real struggle is in unlocking the treasure trove with more complex interactions and thus elevate the experience away from basic and mundane recommendations that fail to demonstrate anything truly relevant. A commonly cited example of such a valueless use is that of being retargeted consistently across channels for a product you have in fact just bought. Here we are witnessing only the technological capability of acting on some data, regardless of the device or network that you are on. Not very satisfying, at all.
As it is with many capabilities available to marketers everywhere, a too simplified and direct application does not bring the expected results, or at least, not to the full extent of its potential. With personalization there are multiple elements: (1) data and its derived insights; (2) intelligent experiences that leverage this; and most critically (3) the availability of a variety of content that can be used by such experiences.
This last element is where things get interesting. Of course, a lot of content can be prepared beforehand, such as what happens in games that feature non-linear plots that allow a player to make decisions that influence the outcome, so that multiple playthroughs can provide a different adventure each time. Last year Netflix tried its hand at such a concept with their Dark Mirror episode "Bandersnatch". Lifted straight from the ideas that go back to old CD-ROM games, the viewer is presented with a branching narrative. Choices made give access to various different parts of pre-recorded content.
What if such modular content did not have to be produced ahead of time? What if this could be generated dynamically, on the fly, as needed to deliver personalized recommendations based on data. That would radically change the consumer's digital experience from a simple and far too easily disposable incident (possibly even annoying due to its perceived dumbness in repetition).