We have updated our Privacy Policy and Privacy Options

Got It

How to Use Personalization to Enhance Your Existing Optimization Program


We know an A/B test can lead to powerful insights. However, the information gained from traditional A/B tests tends to be focused on what’s best for the majority of users – not every individual user. That’s where personalization comes in.

Personalization enables you to leverage the specific wants and needs of each individual user on your site. This can lead to even more substantial results–higher conversion rates, deeper engagement with your site, and increased revenue.

Many of the traditional testing tools–Adobe Target, Optimizely and Maxymiser–have personalization capabilities available. There has also been an emergence of companies like Dynamic Yield and Evergage, which offer personalization technology as their core focus.

As technology in this space improves, personalization has become a major focus for many Brooks Bell clients. However, the question we’re often asked is not whether to implement personalization alongside existing optimization efforts – rather, its how to do this.

Luckily, there are many ways to do just that. For the purposes of this blog post, we’ve outlined two relatively simple strategies for implementing personalization alongside your existing optimization program.

Strategy 1: Rule-based targeting

Rule based targeting is a personalization technique that’s available on most A/B testing platforms. Instead of targeting all users, you select a specific segment of users to target an experience to: new or returning users; mobile or desktop users; or users in a specific location.

Because these different types of users are interacting with your site differently, you’ll likely see higher returns by personalizing your content to each group.

You can also apply rule-based targeting after running a traditional A/B test, by breaking down your results by those specific user segments. In doing so, you may find that a “winning” homepage experience performed very well among new users, but was flat for returning users.

Though pushing the winning variation live to all users would increase revenue, you might see a bigger increase if you were to push it live to new users only. This gives way to additional opportunities to test different strategies for returning visitors.

Strategy 2: Predictive personalization

Many testing platforms now offer predictive personalization, which works in real time to learn which experiences are ideal for certain types of users.

A predictive personalization “test” runs indefinitely – and adjusts as users’ preferences change over time, showing the optimal experience to each user.

Predictive targeting technology is exciting for many reasons. It accounts for the fact that a winner from a year ago might not be the best option for your users now.  

The technology also makes it easier to figure out the best option for short term website changes, like a holiday promotion–for which A/B testing is not a viable option due to time constraints.

Additionally, having the ability to step back and leave the analysis to the computer – instead of spending the time analyzing data yourself – is a huge benefit to experimentation professionals and the companies they work for.  

There are, of course, potential pitfalls to this form of personalization.

When you run a traditional A/B test with a clear winner across all users, it’s easy to make the decision to build the winning code into your site. However, with predictive personalization, you may have many different versions of a page for different segments of users, and continue relying on the testing tool to deliver the code, never building it into your site.

This can be risky for a few reasons: it can increase load time; and if, over time, other updates are made to your site, those updates could break the experience.

Additionally, you’ll also want to make sure you trust that the machine learning algorithms are actually making the best decisions for your users. To that end, many platforms offer a control experience which segments users randomly. You can then compare metrics from the control against the personalized segments to ensure the algorithm is working optimally.

Personalization offers the opportunity to gain new insights about your users and deliver the most valuable content for each individual. Incorporating personalization into your testing program is certainly worth the investment, with the potential for huge rewards.

At Brooks Bell, our Personalization Jumpstart program enables enterprise optimization teams to incorporate and scale personalization strategies into their existing optimization programs. To learn more about our services, contact us today.