Product Predictions Logic
In order to create our Product Predictions, we look at two types of logic:
Bought together: We analyze historical orders that include items with the product a customer bought to look for logical compliments.
Bought X, then bought Y: We analyze the orders for customers who started with the same products to determine what the most likely next products are in the following order.
We also take into account order number (products commonly purchased on 2nd purchase vs. 3rd, etc.)
Recommendations update over time as we have better data globally, and recommendations update per customer as they place more orders.
The following will not appear in product recommendations:
Out of stock items
Unavailable items (products not available on the Repeat sales channel)
Items that have already been purchased by the customer
Note: To add or remove products from the recommendations, add or remove them from the Repeat sales channel. You will want to exclude any products you wouldn’t want to sell to a customer (for example: free gifts, package protection, etc.) Keep any real products on the Repeat sales channel because this influences all of our automations and dynamic content.
How to see Repeat’s Predictions in Action
In order to view your Product Predictions, navigate to the flow template for your Primed for Cross-sell Flow on Klaviyo. After selecting an email of your choice, click to edit the template and enter the “Preview & test” mode. Here, you can surf through recent events to see what your predictions look like.
If your brand does not already have a Primed for Cross-Sell Flow template, reach out to the Customer Success team to clone a template for you via [email protected].
How Product Predictions are different than traditional recommendations
1. Repeat Product Predictions are tailored to specific Moments, including:
Cross-Sell Product Predictions
Replenishment Product Predictions
Subscription Upsell Predictions
2. Repeat’s Product Predictions are consistent across channels
Our Product Predictions work in both email (Klaviyo) and SMS (Attentive/Postscript) for a unified customer experience
3. Repeat’s Product Predictions are flexible in how they display to the customer
You can show customers dynamic Product Predictions in any email flow or campaign by including a pre-built reusable component
Product Predictions are available across all customer profiles in Klaviyo, so you can also use them to build powerful email segments for flows or campaigns
4. Repeat Product Predictions work with Repeat's Product Mapping
If you have any replacement/sample/bundle mapping set up inside of Repeat, the Product Prediction engine takes that into account when choosing what products to display to your customers
5. Repeat’s Product Predictions are real-time
The entire engine gets smarter as your store grows and each customer’s predictions get smarter as they place more orders
6. There are various settings that allow Repeat Product Predictions to create a hyper-personalized experience:
Choose from a variant vs product “prediction type” depending on if you want to recommend other variants of the same product or not
Use our goals layer to choose what products get prioritized in the recommendation engine (newness, order value, by category, etc.) [Coming Soon]
Using Product Predictions with Repeat
There are a variety of Plays you can utilize to put Product Predictions into action, including: