Hi there @ejb,

Welcome to the Community and hope to provide some guidance and info! 

Our personalization engine tailors product recommendations to recipients based on their past behavior, depending on the metric that you have chosen. This is a key concept when creating a win back series.

There are three metrics:
Viewed- this option uses their browsing history. Klaviyo will generate recommendations based on the products that customers have browsed in your store.
Ordered- Klaviyo can infer each recipient’s preferences, and recommend products that are purchased by other customers who have similar tastes, styles, or needs based on your customers’ past order data.
Blended- will generate recommendations based on a combination of Ordered Product and Viewed.

The product recommendations are specific to a recipient past behavior.  For Ordered Product our engine will infer recipient preferences and  suggest products that have similar styles, customer's who have like tastes, or needs based on your customer's past order data. If a person bought items in one color, say a scarf, hat, dress our system might suggest the same color accessory. A customer must have significant purchase data for this. 
If you select Viewed Product, then our engine will make recommendations based on the browsing history in your store. If a customer does not have existing purchase data, they will receive recommendations  similar to the ones they already viewed.  New products take time to appear in product recommendations with this metric. They must first accumulate a purchase history. A recipient must have an existing profile to make a personalized recommendation based on viewed product.
Blended will make the recommendation based both the Ordered Product and the Viewed Product metrics. 

Please read more detailed information on set up and other concerns here!

Knowing this info, I would recommend checking out our article going over how to Create a Winback Flow!

There are also documentation on how to create an Upsell/Cross sell Flow as well as a Product Review Flow that could be of future use :)
Lastly, as supplemental info, I would recommend looking into our post purchase flow community post too!

Let me know if this helped!
Alex