Artificial intelligence is more than just a nice-to-have; it’s a necessity.
Fusion is pioneering the use of Artificial Intelligence in the world of optimization and testing. We use machine learning today in order to optimize our testing performance and deliver greater results and benefits to our clients.
Currently, we use Thompson sampling methods that allow us to optimize traditional A/B tests by using algorithms (multi-armed bandits) that drive more traffic to the best performing offer at any one time. The result is quicker test wins and higher commercial return and benefit from each test cycle we run.
However, we are moving into contextual personalization. This is where things get really exciting and takes testing to a whole new level! We are taking contextual transaction data (non-PII) and creating customer segments dynamically – offering a personalized product and message to an ever-changing variety of customer experiences. This eliminates the need for any pre-determined, pre-exisiting customer segments – as the rules for creating new segments are done “on-the-fly”.
Finally, we believe “Feature”-based optimization is the final step combining AI and optimization. This is where each element of an offer such as a header, copy, price, image, and layout becomes a ‘Feature’ – which is used to build an offer dynamically! You can have numerous versions of each feature which potentially gives you thousands of variations to test and explore.
Through machine learning and contextual personalization, Fusion is able to test thousands of combinations in parallel to determine the optimal offer to display to each individual customer.