Data explanations and considerations

Retention Science's data schema has been optimized to power Cortex AI performance and ensure our client's success. Whether you are setting up Flat File Imports to feed us your data or already installed our Shopify or Magento plug ins, this guide will help your understand our data schema and key details.

• How is this Data Received?
• Magento/Shopify clients
• Custom integrations
• Data API
• What does this data do for you?
• Items
• Users
• Categories
• Orders
• Order_items
• Subscriptions
• Item reviews
• User wish lists

See how it's done: Looking for more help? You can also watch our YouTube videos for similar explanations of data, JS, and other topics.

How is this Data Received?

Magento/Shopify clients

• Categories/Items/Users/Orders/Order_Items are "magically" pulled from the respective systems via our standard code installs.
• Subscriptions/Item Reviews/User Wish Lists have to be FTPed over.
• For things like Subscription data we may have additional resources to support you, such as our Recharge Script which covers 90% of the DEV work for a Recharge integration!

Custom integrations

• All data would be sent via FTP as "Flat Files"
• Simple TSV files.
• Usually our clients get their query setup to extract data in a certain manner, then schedule a job to execute the query and transfer the file to our FTP.
• Each client's execution of this task is going to be different depending on tech infrastructure.

Data API

This method is not really used to fill large amounts of historical data, but can technically be used to add orders/users to our system in real time.

• You can reference our Data API documentation here.

What does this data do for you?

Items

These would be the products that you sell and want our AI to optimize as recommendations to your users in emails. For example:

• Standard e-commerce clothing company - Shoes, Shorts, Tops, etc.
• Subscription Business - The items that can be given in a subscription offering/box.
  
  

Context on select fields:

• active - Controls whether items are allowed to be recommended in your emails, effectively serving as your "on" and "off" switch for items.
• item_url - If an item is recommended in an email, it typically needs to be clicked upon, and this URL is where the user is directed to.
• item_type - How items can be grouped together for use in our item_targeting feature (similar to Categories), and for special AI customizations this field could be leveraged to serve certain items only to certain users.
• Example:  If item_type = "Canada", these items would only be served to Users that also have a corresponding datapoint tagged "Canada"
• If you are a Subscription business offering stand-alone products vs subscription products (for example, a monthly box), you can leverage this field to distinguish between your item types.
• Discuss with your Onboarding Engineer how to best set this up.
• image_list - Items are in your emails, so make sure they look good and you use the best/consistent images.
  
  

Users

These are all of your purchasers and newsletter subscribers, essentially a list of every person that can and cannot be emailed.

• Example: Even if a person is unsubscribed, data about them is still valuable and they may end up browsing onsite and giving our AI information.
  
  

Context on select fields:

• birthday - Required to use Birthday stage in Cortex.
• registration_source - Required to get lead scoring features inside of Cortex.
• unsubscribe_link - Typically not needed because RS has it's own unsub link, but clients looking to be the unsub system of record can use their own instead.
• custom fields - We can support custom attributes and dynamic segmentation of these values, but it does require additional setup/implementation on the RS side.

Contact your Account Executive if you would like to add these to your Cortex platform.

Categories

Similar to "item_type", serves to help our AI models understand how your items are grouped together and enables you to use category based recommends in Cortex.

• Example: A makeup company may have Categories of "eyelashes", "face", "lipstick", etc.
• Only consideration is that the record_id in this table is used in Items/Order_items columns when designating what categories Items fall into.

Orders

The information about what people have bought, fairly standard.

Context on select fields:

• user_record_id - Be sure this matches your record_id in the Users file.
• order_status - In your order data you may have various status strings "returned", "canceled", "pending", etc.
• You have the option of telling RS which ones should be revenue attributing in our dashboard vs. not.
• Example: Make sure "canceled" orders don't count towards any $$$ totals.
• Your Onboarding Engineer would then make sure your system is setup correct based on your unique statuses.
• For Magento/Shopify this is already automated.
  
  

Order_items

Similar to Orders, but even more straightforward as this is just giving more details and quantity of the orders.

• Just make sure your order_record_id maps properly to your Orders file and item_record_id maps properly to your Items file.
  
  

Subscriptions

• Only useful for companies that are pure Subscription (e.g. only offering a monthly box) or Hybrid Subscription (box + one off products).
• Required if you want to be on our "Subscription Cortex", which is a specialized version of Cortex that caters to needs of Subscription companies.
• Think of this as a subset of the Users table as all Subscribers here would exist in the Users table, but everyone in the Users table would not exist here.
• Example: Not everyone in the Users table has bought the product.
  
  

Context on select fields:

• user_record_id - Be sure this matches your record_id in the Users file.
• status - This can be used as a flexible field for segmentation, as you can put any value you want and group your subscribers into them for specific targeting.
• Typically values such as "Active", "Paused", "Canceled."
• trial - If you don't do any trial offerings, just default to "0."
• churned - Typically related to the status of "Canceled", set to 1 if the subscription is terminated.
  
  

Item reviews

• Not required, consider this a bonus table to make our AI even smarter if you have the data!
• Most of these fields are straightforward, and similar to other tables just make sure that user/item record_ids properly link to other files.
  
  

User wish lists

Similar to Item Reviews in that it is not required, but can enhance performance of our Price Drop and Back in Stock campaigns.

• Data essentially tells us which users have expressed explicit interest in what items, and fields are straightforward similar to item reviews.