Choose an approach to website optimization
1. Consider A/B tests if you can limit your website to a single version shown to all visitors, and you have the time and resources to test one idea for long enough to reach statistical significance.
A/B tests compare the conversion rate of a new version of a page against the original by showing each variation to a random 50% of your audience. Whichever version performs best is declared the winner, and becomes a standard part of the website.
2. Consider multivariate tests if you want to find the single best version of a page and understand how different elements interact with one another, and have the high traffic and longer time to support them.
Multivariate testing (MVT) is like running multiple A/B tests in parallel. This approach uses many of the same principles as A/B testing but instead of testing one element at a time, MVT tests multiple elements simultaneously (e.g. headline, image, and CTA). Similar to an A/B test, MVT will allocate traffic evenly among every possible combination of your ideas. For example, if you were testing 2 headlines, 2 images, and 2 CTAs, MVT would allocate traffic evenly among all 8 (2 x 2 x 2) possible combinations. In other words, each combination of headline, image, and CTA would receive one-eighth of the traffic.
3. Consider rules-based personalization if you have a lot of time and resources to spend on deciding the right rules for each segment, setting them up, and then maintaining and iterating on them as visitor behaviors change.
With rules-based personalization, you set up one or more rules to define what a segment of visitors will see on your website. Each rule is like an if this, then that statement. For example, If the visitor is from a B2B SaaS company, then show them case studies from other B2B SaaS companies. It makes the most sense to use rules-based personalization when it’s inappropriate or brand unsafe to show particular content or a specific experience to an audience, like showing content about the Boston Red Sox to a New Yorker.
4. Consider Continuous Conversion™ if you want to deliver the most relevant, personalized page experience to each unique site visitor in real time and are okay with each visitor having their own experience.
Continuous Conversion uses machine learning to automatically test a virtually unlimited number of variations simultaneously on the same page and decide which combination of these variations to show each unique visitor in order to maximize conversions. It also automatically adjusts the page experience over time as visitor mix or visitor behavior changes - for example, when you run a promotion or change ad targeting.