A/B testing has been around since as early as the 1920s, when statistician and biologist Ronald Fisher began conducting randomized controlled experiments in farming (source: Harvard Business Review). He might, for example, test two different levels of fertilizer on a plot of land, controlling for environmental factors, in one of the first documented A/B tests.
The practice became the foundation of medical clinical trials in the 1950s and has, in more recent decades, become a foundational practice for digital marketers. A/B testing has conceptual value, but its shortcoming is its simplicity. Today, there are far better ways to scale testing, thanks to massive amounts of data and technological resources.
A better way
When creating digital experiences, there’s a better way to optimize. What if you could quickly identify the best of 20 options and, within the most successful of those 20 options, determine the exact components of a landing page that tested well?
We call that continual optimization, and it involves constantly running experiments across all of our clients’ landing pages to identify the strongest possible conversion scenario.
We use the MAB (multi-armed bandit) testing methodology to dynamically scale traffic to winning variations, with low-conversion-rate variations being replaced by new variations throughout the process. Winning variations continue to be updated based on real-time data and ongoing experimentation.
Here’s a diagram of our process at work:
As you can see, the most successful variation within a cluster competes against variations in other clusters until the best prevails. The process keeps going, continually optimizing for the best possible conversion rates.
Clearly, the MAB methodology has enormous potential for continually optimizing digital experiences. It can:
- Identify and present the best possible online experience for landing page visitors on an ongoing basis.
- Ensure the experience is highly customized to each specific audience.
- Aggregate insights from across users’ customer behaviors.
Our data models are built from more than 20 billion data points, 3 million landing pages and 800,000+ conversions. That’s why Postclick customers see an average 53% conversion lift using the continual optimization methodology built into our platform.
Receive a free conversion analysis
Receive a free conversion analysis on your landing pages and see what the MAB methodology can do for your company. The analysis will include a live page review, an audit of your campaigns, page performance and speed insights, and a competitive benchmarking report. Request your free conversion analysis here.