At first, print service providers (PSPs)—including mailing companies and direct mailers—might wonder what they have in common with Netflix. Although the service started out as a direct mail offering, it has since become more associated with online digital streamlining. Nevertheless, a recent technology blog from Netflix has shed some additional light on how the company is pushing its personalization practices forward, making them more advanced and better synced to each individual Netflix user. While this level of personalization is quite advanced now, it will likely become increasingly commonplace in the years ahead.
Many firms, including PSPs, can benefit from absorbing and understanding how Netflix continues to personalize its offerings in today’s digital age.
First off, PSPs cannot hope to replicate Netflix’s success without data analytics. While the Internet of Things (IoT) has made it easy for companies to absorb record amounts of data, all this innately amounts to nothing more than absurd amount of information on its own. If a company takes in a wealth of information but has no way of understanding it, then that information loses all its value and becomes static and unusable.
Data analytics software solutions are designed to help organizations separate all this data into digestible, actionable chunks. Netflix has been using data analytics for years—since its launch, in fact. From the get-go, the company asks each customer to create an account and then asks them to rate every movie or show they watch using the service. All of these preferences are tracked to develop an understanding of user behavior. Netflix also monitors the shows and movies that users watch without rating, as well as the programs they add to their queues (intend to watch).
An array of different artwork designed to appeal to different types of users.
Source: Netflix
By tracking all of this information, Netflix is able to personalize each interface so it becomes unique to the user—displaying movies and shows that they are more likely to watch. But that’s not all. Now, Netflix is taking personalization to new heights by personalizing artwork for users. What this means is that two different people may see a popular show like Stranger Things presented in a unique way. This is designed to make both people more likely to watch the show. This level of customization is based on customers’ past viewing histories and preferences. For instance, customers who like horror will see artwork designed to heighten scary elements, while customers who are Adam Sandler fans would be more likely to see his face highlighted in each one of his movies.
Imagine this level of customization in the direct mailing world. If direct mailers can track their products with even some degree of the success that Netflix has experienced, they can start personalizing mail and other products to an extent that has never been seen before.
One interesting insight that came from Netflix staff members is what it means to “stand out.” Since the company works with such a visual interface, image overload is quite possible. A vibrant and colorful image will stand out on its own, but it would be more difficult to distinguish within a field of similar images. This is an important concept for direct mailers to consider. Right now, almost every direct mailing campaign—mass or personalized—will include a splash or color to add contrast and differentiation to what would otherwise be a pile of dull while envelopes. If most direct mail pieces became saturated with vibrant colors, the attention-grabbing effect would be diminished.
Netflix is working to balance its software programming so that its interface knows not to place too many like images together. By keeping things in rotation—and developing a diverse variety of image options—Netflix can keep all of its offerings fresh and increase the likelihood of user interaction. Mailers and marketers can take the same lessons from this approach. It doesn’t matter how much color is used or how good the print quality is; variety remains the spice of life and is crucial for effective communication.
Netflix has moved beyond simple mass customization and into a level of personalization that is only possible through machine algorithms. For lack of a better term, these algorithms can learn to perform their jobs more effectively. In contrast, traditional batch testing is slow and keeps a percentage of clients on an outdated infrastructure while the company scrambles to innovate.
Using a type of algorithm called contextual bandits (also known as associative bandits, multi-world testing, or learning with partial feedback), companies like Netflix hope to better control which content is displayed to which user, which content should be prioritized for which user, and how best to optimize search results and other user inquiries.
The term “bandit” comes from gambling world. Slot machines with pull levers are known as “one-armed bandits” because they take money and only rarely return the favor. That said, all slot machines have different probabilities. In theory, a person could go to a casino and pull a minimal amount of levers/arms with a maximum payout potential. Through experimentation (and by favoring the machine with the best payout), this person would be able to reap greater rewards than the average casino guest.
A/B testing prohibits many users from seeing any improvement. Multi-armed and contextual bandit testing bring more uniform change to a much wider batch of users.
Source: Dynamic Yield
For Netflix, these algorithms optimize using previous observations as well as situational information. People who like comedies will see more comedies, while people who like drama will see more drama. In the direct mailing world, an animal shelter might send dog lovers envelopes with dogs on them, while cat people would see cat-clad mailing pieces.
These advanced machine learning algorithms are helpful for pushing the level of personalization beyond its traditional boundaries. Without them, companies will often find themselves stuck at surface-level customization. For example, Coke’s “Share a Coke” campaign—which has already proven a huge success—might be even more profitable if the company was able to send Cokes labeled “Jim” to areas including a higher-than-average share of people with that name.
Although Netflix primarily operates in the digital space, it can still provide some applicable lessons for companies considering the use of personalization to boost sales. Netflix has established itself as a pioneer for customized interfaces and user-product matching. Now, the company is pushing even further, diving deeper into the potential of personalization. Those in other industries that are hoping to increase the potential of their own customer communication channels can use these principles—data analytics, contextual bandits, and image diversification—to create more effective marketing and communication efforts.
For more information on evolving customer communications technology and trends, please check out the InfoTrends report store or become a subscriber today.