Amazon Personalize – Improve the user experience on your ecommerce site

In this blog, we will dive into Amazon Personalize and see how it influences users’ experience on sites. To begin, let’s understand why we need a personalized experience. We often tend to buy things that we commonly see used by other people. It may be fashion trends, exclusive offers, or new gadgets. Sometimes, time plays a constraint which pulls us back from making a purchase but what we need is lost in history. So a personalized experience will provide you with the list of products that you have seen in the past and also the new ones which you are likely to purchase. We can say that personalization is like a mind reader and saves a lot of time in deciding what is the ideal product for us.

What is Amazon Personalize ?

Amazon Personalize is a machine learning service that makes it easy for developers to add individualized recommendations to customers who use their applications. You can use Amazon Personalize in a variety of scenarios, such as giving users recommendations based on their preferences and behavior, personalized re-ranking of results, and personalizing content for emails and notifications.

How does it work ?

To make recommendations, Amazon Personalize uses a machine learning model that is trained with your data. The data used to train the model is stored in related datasets in a dataset group. Each model is trained by using a recipe that contains an algorithm for a specific use case. In Amazon Personalize, a trained model is known as a solution version. A solution version is deployed for use in a campaign. Users of your applications can receive recommendations through the campaign.

A dataset (contains product csv, users csv and interactions of users with product) can grow over time and your models can be retrained on the new data. The data can come from new metadata and the consumption of real-time user event data.

Amazon Personalize workflow

  1. Export the product , users and interactions data from the site for which you need to implement personalization. Sometimes when we don’t have interaction data, the same can be purchased from the datasets available on different platforms. This may not be accurate interactions that you can expect on your site but it may hold true to the level of user behaviour across the web.
  2. Please make sure you set up all required prerequisites
  3. Upload your dataset in Amazon S3
  4. Create dataset group in AWS console
  5. Create the schema for user interaction data
  6. Import the dataset and click next
  7. Create a solution using existing recipe based on your requirements
  8. To apply this recipe and training the models takes some time
  9. Once the solution is active, proceed to create campaign to use this solution
  10. You can have multiple campaigns for each solution and the same can be used on your site or email subscription newsletter
  11. Once the campaign is active, use getRecommendations API to get list of product IDs for users with their ranking score

Detailed guide to follow the above steps can be found here

The beauty of Amazon Personalize is that you don’t have to know what is under the hood. You don’t interfere with AI/ML directly and most of the things are already set up for you. One can directly select the recipe and get the desired results. One thing that needs to be kept in mind is having enough data so that models are trained properly and results are more relevant to the users. 

Is Amazon Personalize only for Retail?

Well, the focus of this blog was more towards product recommendations but Amazon Personalize can be used for a variety of business solutions. It can be used for suggesting relevant blogs to premium members. If there is video streaming content, then the same can be used along with Personalize and the list goes on.

What will be the cost of the overall system?

When we consider AI/ML, we always think of having to spend too much on infrastructure, building the algorithm, maintaining the underlying hardware cost, and using the large chunk of data. With Amazon Personalize, you pay only for what you use, and there are no minimum fees or upfront commitments. You are charged based on the amount of data processed and stored, on the compute hours used to train your models, and for the throughput of recommendations.

There is a long way to explore Amazon Personalize as it has many algorithms and recipes which can be used for various recommendations. Hope to share more interesting recommendations and recipes in days to come. Stay tuned.

Siddhesh Shirodkar
Siddhesh Shirodkar
SJ Innovation LLC
Where employee happiness generates client success.
Follow us on Facebook or Twitter