I have spent the better part of the last decade working in the marketing personalization space and I can say with confidence that personalization that can be described as intrusive, creepy or offensive is not personalization. As a rule of thumb, personalization should make things appear more relevant—but should never feel overt or in-your-face.

Netflix recently came under fire for altering the featured movie cover actors depending on the viewer, in an attempt to  make the films more appealing to specific audiences. They claim they don’t track race or gender of their customers, yet the images being served up to certain customers would make you think otherwise. Not only that, but the image swap often resulted in  a clear misrepresentation of the actual content. This is even worse than bad personalization: they have also layered on a bait and switch. “Watch this” and see how little the people that have been portrayed as having leading roles are in the actual movie. I am always up for pushing the edge, but this is just a complete fail on so many levels and I think the backlash is absolutely justified.

Idea vs Execution

You could argue that Netflix was just trying to put the right content in front of the right person. It’s also likely that much of the content chosen was actually relevant for the customer. Netflix has so much content these days, that without their help most of us would never discover the hidden gems that lie within the pages of options in any given genre. However, taking that extra step of modifying the images was going too far. Misrepresenting the content greatly increases the chance that the customer will feel betrayed shortly after they click play—which, judging by the backlash, is exactly what happened.

Use my data for good

At this point, most of us are aware that businesses have access to significantly more data about individuals than ever before. I, for one, have no problem if someone takes advantage of my information to promote more relevant products or convenience features based on my location. In the Netflix instance, I think they may be actually telling the truth that the data they have does not include race or gender. What they don’t say is whether or not they use predictive models to determine things like race or gender. There is no doubt that race or gender may play a role in a person’s viewing habits, but instead of using that as a single data point amongst many it became the focus point when the imagery was changed. We are more than a single data point and exploiting a single point so dramatically is more likely to offend than pleasantly surprise.

Good personalization should make an experience better.

Netflix’s biggest error was missing the entire point of personalization—which is to make a customer’s experience better. Personalization is about playing the long game, knowing that a great customer experience can trump all in the long run. Netflix instead focused on immediate impact and, unfortunately, left a lasting negative impression on many customers.

It doesn’t have to be that hard

The Netflix example is made more egregious by the fact that, for the most part, they use my information to provide me with significantly more relevant viewing options. I would qualify them as very good at personalization—but they just took it too far, and it failed. Start simple. Use your customer’s information to make their lives easier. The Starbucks mobile app saves my preferred location and drinks, making it easier for me to place my morning latte order. Amazon does a great job at reminding me to reorder products that have a defined lifespan. My bank asks me for my preferences to make sure they don’t bombard me with irrelevant offers. Sophisticated data models that use multiple customer attributes are a great way to target and deliver a personalized experience. Just remember to put yourself in your customer’s shoes and make sure the experience will deliver the desired result.