It’s that time of year again. Days are short. Champaign sales are up. Inboxes are full of end of year deals. And every analyst is doing a year in review post. The year in review posts are always interesting. Every so often it’s important to take stock of where you are and what you’ve done in the past. The post that I like the most each year is from Google. It focuses on the searches over the past year – here’s a link to the 2018 Search in Review report.

Instead of spending my last post of 2018 looking back or making a prediction for the future I wanted to discuss personalization, especially personalization through automation. Yes, this is on most of the analyst’s list of what happened in 2018 and what they see expanding in 2019. Infact, I believe it’s been on those lists for a few years.

No surprise here

I hope if you’re reading this blog you’ve heard about personalization and how it can increase digital conversions. The power it can have to sway users/customers who are on the fence. Everyone’s seen it in action. Take a second and look for the, “You might like” sections of most digital properties. Open your inbox and see the messages from any number of brands that you’ve purchased from recently. All of these are forms of automated personalization.

Personalization’s the process of tailoring a digital property or digital elements to individual users (read more about personalization here). Personalization uses two basic methods.

1. Rule-based personalization
2. Predictive personalization.

These two methods are not mutually exclusive. They actually should work together to help build create efficient and effective personalized experiences. The two methods are pretty self-explanatory.

Rule-based personalization means that there are rules or conditions that provide the personalization. For example, I buy a new car and then see ads/content about upkeep on my new car. This can take different formats. Maybe my new car has leather seats and I get served ads/content about keeping leather clean. Another example is in three months I get an email reminding me to get an oil change. The personalized content that I am being shown is based on an action I took. This is why I sometimes refer to rule-based personalization as reactive personalization. This has been around for a long time and is widely used.

Predictive personalization is where there’s a lot of growth. Instead of requiring a user to do a specific task and then reacting to it, predictive personalization makes assumptions about a person based on analytics. This assumption is not directly tied to a specific user or user action, but instead is based on larger usage trends. So, if I have a digital property that caters to three different audiences – Students, armchair investors, and hedge fund managers I could show different content in prominent places based on data. If an IP address is from a university I could assume those users are students and therefore displays information that data shows they’ve interacted with the most. If a user comes back to the data section that targets hedge fund managers, the assumption could be made that the user is a hedge fund manager and doesn’t need the same information as a student.

The examples above are rudimentary, but I use them as conversation starters. Personalization can become more and more complex (and efficient) as more data is collected and automatically analyzed by various tools. So, why isn’t everyone jumping into the personalization pool?

The personalization pool

Often when I talk about personalization with clients/potential clients I get some pushback. Especially with smaller clients and organizations. Yes, everyone’s interested in personalization, most people believe it will improve their conversion rates (and have actually taken advantage of it on other digital properties). So, why isn’t everyone jumping two feet into the personalization pool?

From my experience, there are a few of topics that cause organizations pause when it comes to automated personalization.

1. Not knowing where to start.
2. Giving up control.
3. The misconception that it’s pricey to implement and maintain.

Yes, I put these topics in order. Usually, when I have discussions about automated personalization the three topics above are brought up as barriers (usually in the order I’ve listed them).

The above examples of personalization demonstrate that personalization doesn’t need to be super complex. In fact, when you’re new to it, I recommend starting simple. Ask a simple question or propose a simple problem. In the above examples, this is the approach that was used. For the car example, the problem was how to help customers maintain their car (preferably at a partner garage). In the investing example, the question was how to show value to different users by making sure they see relevant information quickly.

Everyone loves to have control. Once we have it, it’s often hard to give it up. So, when I say an algorithm is going to make the decisions, it can be a tough pill to swallow. I’ve found the best way to get around this is to ensure that the algorithm is fully understood. Control is not being lost, it is just being programmed. Making sure the naysayers have a stake and a voice in the process can help alleviate their concerns. The main point that requires getting buy-in is the idea that data can help improve user experience and that in turn benefits everyone.

Automated personalization can be expensive, but when implemented using an iterative approach, the cost can be spread out. This allows the benefits to be seen and validated during the process. It is a process. Automated personalization is not something that happens overnight, it takes time and effort. It really is like walking. First, you crawl. Then you walk. Then you run.

Come on in, the water in the automated personalization pool’s just right. There’s always be a reason not to do something. In the case of automated personalization, there are just as many reasons to do it (at least to dip your toe in).