Measuring the impact of Amazon Personalize recommendations
As your customers interact with recommendations, you can measure how Amazon Personalize recommendations are helping you achieve your goals. You can identify which campaigns and recommenders have the most impact on key performance metrics. For example, you can identify which resource generates the most minutes watched, the most clicks, or the most purchases. And you can compare the performance of Amazon Personalize recommendations to those generated by third-party services.
When you know which campaign or recommender is generating the most impact, you can take actions to further benefit from its recommendations. For example, you might increase the prominence of the recommendations on your site to drive more engagement. Or you might feature the recommendations in a marketing campaign, such as personalized emails or targeted ads.
If you identify a resource that isn't having the expected impact, you can take actions to improve recommendations. For example, you can use the Amazon Personalize console to analyze the training data used to create the resource, make the recommended data improvements, and then import data again. For more information about analyzing data, see Analyzing quality and quantity of data in datasets.
The following can help you measure the impact of recommendations:
-
Metric attribution: An Amazon Personalize metric attribution creates reports based on metrics that you specify and the item interactions and items data that you import. For example, the total length of movies watched by users, or the total number of click events. After you create a metric attribution, Amazon Personalize automatically sends metrics on events from the PutEvents API operation and incremental bulk data to Amazon CloudWatch. For bulk data, you can choose to publish reports to an Amazon S3 bucket.
-
A/B testing: Performing A/B testing with Amazon Personalize recommendations involves showing different groups of users different types of recommendations and comparing results. You can use A/B testing to help compare and evaluate different recommendation strategies, evaluate model performance, and measure the impact of the recommendations.