AI service
Amazon Personalize
Builds personalized recommendations and rankings.
Key points
- Creates real-time personalized recommendations using user behavior and item data.
- Supports use cases such as user-item recommendations, similar items, ranking, and promotions.
- Uses managed ML so teams do not have to build recommendation algorithms from scratch.
- Requires relevant interaction data such as clicks, purchases, views, or ratings.
- Outputs recommendations or ranked lists for applications to display.
When to use it
- Choose Personalize for product, content, media, or offer recommendations.
- Use it to rank items differently for each user based on behavior.
- Use it when the scenario resembles retail, streaming, news, or personalized marketing.
Exam tips
- Personalize recommends items; Forecast predicts future numeric time-series values.
- Personalize is not for generic search; Kendra handles enterprise semantic search.
- Good interaction data is essential for recommendation quality.
- For custom recommendation research or unusual algorithms, SageMaker may be the better platform.