Email Recommendations is an offering within Constructor's Cross-Channel & Offsite Discovery suite, enabling companies to proactively engage shoppers with emails that contain attractive, personalized content, such as item recommendations.

Constructor’s product discovery learnings from across the company’s website drive these recommendations. For example, a shopper who browsed running shoes last week but didn’t make a purchase might open an email and see similar shoes or featured pairs on sale.

Since the shopper might not visit the company’s website every day but likely does look at their email inbox daily, the company can deliver timely recommendations that are guaranteed to be attractive and of interest to the shopper.

How can I use Email Recommendations?

• Target known users or registered shoppers based on actions.
  • Send alternatives to out-of-stock items shoppers have viewed, suggest items they may like, or send a reminder email with products they left in their cart or items they viewed and did not cart.
• Send promotional emails to specific segments or your entire audience.
  • Promote timely item recommendations, like holiday items, or showcase best-selling items. While the promotion is broad in scope, items shown to a user are personalized based on their shopping and behavioral history.

How Email Recommendations works

Email Recommendations integrates with email service providers (ESPs), allowing companies to design campaigns and specify how recommended results appear to recipients.

Sending an email campaign requires action in both a company’s preferred (ESP) and their Constructor dashboard.

⚠️ Please note: Companies must implement Email Recommendations before they can send recommendation campaigns. There are two options: data feed templates and HTML snippets. Learn more about these implementation methods here.

Step 1: Design an email campaign

Where? In a preferred ESP

Companies must decide the type of campaign they wish to send: either a broadcast campaign or a campaign triggered by a shopper’s action (e.g., cart or browse abandonment).

Not only will this inform the copy and design of the email, it will dictate the type of recommendation pod the company will need to create in Constructor.

Step 2: Configure the recommendation pod

Where? In the Constructor dashboard

Merchandisers will need to create a recommendation pod and assign the appropriate strategy for the campaign (e.g., abandoned in cart, complementary, bestsellers, etc.). Alternatively, merchandisers can use an existing recommendation pod.

Merchandisers can also searchandise the recommendation pod, creating boost, bury, and other rules that can influence results to meet a short-term promotion or goal (e.g., boost a specific brand for promotion, bury out-of-stock items so they do not appear in pods, etc.).

Step 3: Create a notification campaign

Where? In the Constructor dashboard

Next, companies will create a notification campaign in Constructor. This provides Constructor with essential details about the campaign, such as the campaign type (e.g., broadcast or triggered).

Companies will also indicate the desired send time, choose the recommendation pod to use in the campaign, and specify the number of results to return.

⚠️ Please note: If you are creating a broadcast campaign, you must complete this step 12 hours before you plan to send the email campaign. Any last-minute changes can cause non-personalized results to render in a recipient’s email.

Step 4: Generate HTML snippet

Where? In the Constructor dashboard

If a company is using HTML snippets, then once they create a notification campaign in Constructor, they can proceed with copying the HTML snippet.

Merchandisers can add this snippet to HTML code blocks in an ESP’s drag-and-drop email designer or add the code to a code editor. Which option they choose will depend on how they are choosing to create the email.

⚠️ Please note: If you use data feed templates, then you established an API connection during implementation. This takes the place of this step. Please proceed to step 5.

Step 5: Design email

Where? In a preferred ESP

Next, merchandisers will need to work with their own teams to design the email.

If using HTML snippets, then merchandisers will need to add the style_id associated with the desired product card design they wish to display. The Constructor team designs these product cards with direction and input from a company during the implementation of Cross-Channel & Offsite Discovery.

If using data feed templates, then merchandisers can design the product cards to appear however they wish for them to appear when designing the overall email.

Step 6: Schedule or send the campaign

Where? In a preferred ESP

Last, merchandisers will schedule or send the campaign. It’s important to note that the different implementation methods call the Offsite Discovery API at different times, impacting when Constructor generates and sends item recommendations:

• Data feed templates
  • Depending on the ESP you use, recommendations can be generated at send or open time. ESPs such as Movable Ink and Braze allow for generation at open, whereas other popular ESPs (e.g., Epsilon, Emarsys, and Mailchimp) only allow for generation at send.
  • ⚠️ Please note:
    • With send time, if the shopper visits your website and interacts with more items before opening the email, it’s possible the recommendations won’t be as current since the new behavior came after generation.
    • If you are sending a broadcast campaign using data feed templates that call the API at send time, we strongly recommend you send in batches of 5,000 per minute.
• HTML snippets
  • Constructor generates item recommendations when the recipient opens the email.
  • This method allows for the most current, up-to-date recommendations based on the user’s behavioral history and clickstream data.

🚧

A note on email tracking

Constructor does not include email recommendation data in Constructor analytics as merging the no event requests events from these emails could significantly affect Recommendations on-site metrics. This can result in high volumes and a large number of un-tracked clicks and conversions, skewing on-site Recommendations data.