Start tracking your progress
Trailhead Home
Trailhead Home

Refine Your Recommendations for Email and Web

Learning Objectives

After completing this unit, you’ll be able to:

  • Create standard data extensions for segmenting.
  • Build templates for Email and Web Recommendations.
  • Preview rule manager and reporting tools.

Put Einstein to Work

Now that you’ve successfully set up your catalog and are collecting data from your website, it’s time to use that delicious cookie data you’ve collected. There are a lot of cool things you can do with this data, including using the created data extensions for segmentation or for building behavioral triggers, such as an abandoned cart campaign. For this unit, we focus on segmentation and how to create recommendations for your emails and website. 

Note

Note

Behavioral triggers are built off of the Collect Tracking Code and require you to use Professional Services. Questions? Ask your Account Manager.

Segment with Einstein Data Extensions

Before you start creating segments, you need to enable data extensions for Personalization Builder. Navigate to the Status tab under Email or Web Recommendations. Personalization Builder status console.

 Next click the Setup icon Setupand select Data Extension Settings. Click Enable Einstein Data Extensions. Enable Einstein data extensions.

Once enabled, a variety of data extensions are created based on the type of catalog you set up. Why is this so important? Because it allows you to use the rich data collected through the Collect Tracking Code to create filtered data extensions that can be used for segmentation. 

Here’s how shoe retailer Cloud Kicks might use these data extensions to target specific customers. Brandon Wilson, digital marketing manager for Cloud Kicks plans to create a sendable data filter, based on the content views data extension. Cloud Kicks has an annual end of summer sale, and he wants to send a targeted one-time coupon to customers who viewed sandals. Einstein makes this possible.

There are tons of ways you can use your data to create targeted campaigns. So, go on, experiment with using these data extensions in your campaigns, but don’t update the fields or add custom fields to these—to make sure everything works properly it’s important to leave them as-is or make copies of the data extensions or filtered data extensions.

Note

Note

Learn more about the data extensions created on the help page,  Personalization Builder Integration with Contact Builder.

Einstein Email Recommendations

Einstein Recommendations provide customers with fast, accurate recommendations when they open their email. But first, we need to build those recommendations using templates. On the overview tab of Email Recommendations, notice there are two distinct areas: Logic and Displays. Let’s start there.

Overview screen with logic and displays identified.

Logic (1) is where you create the logic or rules for each type of recommendation. Display (2) is the what or how your content is visually displayed in your email. You can use an existing display or create a new one. You should have unique logic for each email template, but displays can be reused across multiple logic models. 

Let’s walk through the steps to create a recommendation. (Reminder: you need to be an admin to complete these steps. But even if you’re not, you can still follow along.)

Create the Logic for Your Template

  1. Navigate to Email Recommendations under Personalization Builder.
  2. Click Create New Logic. Create new logic screen.
  3. Enter a Unique Name for the email template.
  4. Select the type of catalog you are using and click Create.

Choose Your Display

Next, determine what display (or recommendation layout) is used for your recommendations. A display shows an image of your product or content, along with the additional info that you want the subscriber to know about that recommendation. Choose display and edit screen for display selection or create new.

  1. Select an existing display or click Create New.
  2. You can select identifiers or use the existing identifiers (displayed attributes) by clicking Add.
  3. Click Preview to review your recommendation.
  4. To save your display without publishing, select Stage Changes. Or click Save and Publish.
  5. Click Exit.
Note

Note

Not sure what to include in your recommendation? We suggest starting with these key identifiers: Image, Product Link (URL), Product Name, Regular Price, and Sales Price.

Select Your Scenarios

Remember scenarios from earlier? Yep, those prebuilt algorithms that drive Einstein Recommendations. They’re back. Now’s the time to select a scenario based on the type of communication you are building. Let’s review some common uses.

Campaign Type
Campaign Focus
Suggested Scenarios
Upsell
Email includes a suggested product based on a customer’s previous or current transactions.
Bought Bought Merge
Bought Bought Last Purchase
Bought Bought Last Cart
Engage & Retain
(promos, loyalty, discounts)
Email includes a product or content that is based on a user’s interest or affinity.
User Affinity
Last Views Merged
Recently Viewed
Evergreen
(newsletters)
Email includes recommended content or the most popular or most viewed item. You can also highlight new products.*
New Releases
Top Enjoyed
Top Grossing
Top Selling
Top Views

*Add a rule for products released in the past 7 days—or whatever time frame you choose.
Category or Brand (holiday sale, new brand launch)
Email that includes on-site behavior specific to a category or brand being highlighted.
Category Bought
Category View
Tag Recent
Tag Top Rated
Tag

Let’s look at an example of the selected scenarios for an upsell campaign in a weekly email.

Configure scenarios with enable waterfall recommendations circled and the cursor on Bought Bought Merge.

You can reorder the scenarios as needed and, if desired, select Enable Waterfall Recommendations.

Note

Note

Waterfall recommendations is a recommended setting that shows items from the highest priority scenario first and then fills in the lower priority scenarios until the total number of items for the recommendation block is met.

Choose a Layout

Next, determine the layout for how your recommendations appear. Highlight the number of columns and rows needed, like this.

Selection of 3 columns with 2 rows of content.

Note

Note

Be sure to consider how the display looks in your email. It may be best to have fewer columns and more rows to better accommodate mobile displays.

Get the Code

Now that you’ve completed the set up and previewed the content, click the Get Code tab within Einstein Recommendations. Copy the code, click Save, and then click Exit. Here’s what that code might look like.

Embed code for email recommendation

Navigate to Content Builder and your selected email. Insert the copied Email Recommendations code into a new content block within email. Format your email as needed and be sure to Preview and Test before sending.

Tips:

  • To customize how recommendations render in an email use AMPscript. Want to learn more about AMPscript? Check out the Trailhead module, AMPscript for Non-Developers.
  • Place recommendations above the fold in your email design—in other words, don’t make subscribers scroll to see this awesome content.

Einstein Web Recommendations

Einstein Web Recommendations are similar to the email recommendations we just covered, but are displayed within designated placeholders on your website in real time. It’s magic. API magic, that is. 

You can pull any catalog field into the web recommendation using either a JSON response or HTML/JS. JSON is the preferred delivery method due to its flexibility. So, let’s see what it’s all about.

Get Started with Web Recommendations

Check out this video to get started with web recommendations. 

Rule Manager

Once you have your base configuration for both Email and Web Recommendations, you can use the tab Rule Manager to refine your recommendations. You can create new rules or modify existing content from this page. Rule Manager with Configure New Rule circled.

You might create rules to:

  • Remove a specific SKU from a recommendation based on limited availability.
  • Not show products until a specific release date (from a release date field in the catalog).
  • Add a greater-than or less-than rule to show only products that have prices higher than $25 or to leave out products that have low ratings.
  • Filter on categories that are popular, like clearance.

Every business is different and has unique needs for recommendations. Custom rules can be created, but keep in mind: Rules can override top-performing algorithms and can have a negative impact on predictive performance. We recommend only using them if you have a custom use case that the provided scenarios or algorithms don’t address. Otherwise, stick to the magic.  

Reporting on Your Recommendations

Once your AI engine is up and running, how do you know how it’s performing? You can view the performance of your recommendations on the Reporting page. Click through the tabs to view recommendation performance by channel (email or web), scenario, contact, or item type. Reporting dashboard with session performance and recommendations funnel

This reporting tool gives you insights to do two important things.

  1. Measure the effectiveness of your recommendations in driving top-line conversions and revenue.
  2. Analyze what content is working and what content or scenarios need adjusting to optimize performance.
Note

Note

Learn more about reporting on the help page, Personalization Builder Reports.

There’s no stopping you now. The magic of personalized recommendations is in your hands. In the next unit, we focus on analysis and ways Einstein can improve deliverability and overall performance. 

Rights of ALBERT EINSTEIN are used with permission of The Hebrew University of Jerusalem. Represented exclusively by Greenlight.

Resources