Is AWS recommendation engine Amazon Personalize right for you?

How Amazon Personalize works

Personalize is pretty straightforward, and while it takes some technical capabilities, it doesn't require a team of developers. The code is minimal and simple.

The first step is uploading the data to Amazon. This is the part that differs from other personalization options. Most other offerings reside in your existing IT infrastructure, but Amazon Personalize sits outside of it in AWS. You can load the data in bulk or as you collect it over time.

The second step is analysis and training. You map the transactions with your customers' visits, and Personalize begins to analyze the trends. The more data you feed it, the better recommendations will be made. As you map the desired features and algorithms, you gain a better picture of what your customers are doing.

The final step is to define the API calls that you will use to retrieve your recommendations. You can send a request to Amazon about the customer or a set of behaviors, and Personalize will make a recommendation based upon past activity. These recommendations can be real time and built into the user journey or in bulk, such as when they authenticate into the site.

Understanding if Personalize is a fit for you

The first thing to understand is that Amazon Personalize targets e-commerce sites. If you are not running an online store, then Personalize is not for you. If you are selling products on your site, then you need to ask yourself if you're ready for personalization.

To start, you need to make sure that you've mastered the basics of e-commerce.

The problem is that many firms invest in personalization technology before they are ready to implement it. This is especially evident in the number of organizations that purchase full-stack solutions, such as Sitecore or Adobe. These businesses use the full suite for basic web content management and don't use the advanced experience management tools. They usually have grand plans to do so, but when reality hits, they are too busy trying to get the basics right.

Before you get started

To start, you need to make sure that you've mastered the basics of e-commerce. Otherwise, time spent on personalization would be a waste. You need to have the following:

• the right content in place;
• a search engine optimization strategy that works;
• data to feed the personalization engine; and
• effective privacy and security controls.

If you have all of that in place, you are ready to customize the customer journey. Part of that is personalizing the journey through recommendations. However, understanding the people that come to your site and why is also important to do beforehand.

Of course, unlike those full-stack providers, trying Amazon Personalize is inexpensive and doesn't commit you to anything over the long term. Like all AWS offerings, it is a pay-as-you-go model that you can stop at any time.

Moving forward

The ideal profile for Personalize is a vendor with a sizable product selection that manufactures what it sells or has full control of its market. For example, Yamaha sells products on Amazon, but Yamaha makes a profit regardless of whether the products are sold on Amazon or its own site. But, if you directly compete with Amazon, you may want to think twice.

If you don't compete with Amazon and already have your e-commerce basics under control, give Personalize a try. It isn't expensive for you to start, but like any cloud offering, you should monitor your costs. Test it on a few of your customer journeys, and see what happens.

It may be just the boost your e-commerce site needs.