Embedded Files
One-to-one: Real-time
One-to-one: Real-time
One-to-one product recommendation engine. Recommend the most suitable products to each user using a deep learning engine trained only with your data.
One-to-one product recommendation engine. Recommend the most suitable products to each user using a deep learning engine trained only with your data.
1. Introduction
1. Introduction
The one-to-one recommendation of the products offers spectacular results in terms of conversion and uplift of revenue and number of orders.
You can use the one-to-one recommendation engine using the real time API on your eCommerce or the bulk recommendation for your CRM campaigns and journeys.
Click on One-to-one at the sidebar to access to the recommendation engines. Here you can manage your engines (train, use in real time or for a customer segment, create business rules, update with new data...).
Click on Create new engine at the top right button to create a new one-to-one recommendation engine.
Click on the engine name at the main list to manage the recommendation engine.
2. Create a new one-to-one engine.
2. Create a new one-to-one engine.
Follow these steps to train a new recommendation engine:
Write the recommendation engine name.
Define the Customer information to be used in the training.
Define the Product information to be used in the training.
Create engine clicking on Create engine top right button.
3. One-to-one engine
3. One-to-one engine
When your recommendation engine is already trained and ready to be used, you have these usage options:
Real time: Get the product recommendation in real time using the API.
Bulk recommendation: Get the product recommendation for a Customer's segment in a CSV file.
Business rules: Create new business rules to be used on the recommendation (Real time and Bulk).
Neuronal network: Manage new trainings for your Artificial Intelligence.
Datasets: Upload new data to the Artificial Intelligence.
Events API: Register the user interaction to be used on the next recommendations (Real time and Bulk).
Use the Update button on the top-right corner to refresh the information and check any process status.
4. Real time recommendation
4. Real time recommendation
Click on the first tab (Real time) to access to the recommendation engine in real time.
To test the one-to-one recommendation on real time follow this steps:
Select the deployed model you want to use.
Select the business rule you want to use. You can manage all the business rules clicking on Business rules tab on the header.
Write de user reference id or the user email. You can get the user reference id or the email on the Datalake module. Click here for more information.
Get the number of recommendations you want to get.
Click on Get recommendation.
This is the recommendation result:
Theses are the fields shown for the 10 items recommended:
Product Pleasepoint id.
Product reference id. The product identifier.
The product name.
The product price.
The recommendation score.
Also is shown the previous customer purchases.
These are the fields shown for the previous customer purchases:
Ticket Pleasepoint id.
Product reference id.
The product name.
Quantity, is the number of purchased units for each product.
The product price that has been purchased the product. This value is for one unit.
5. Recommendation API
5. Recommendation API
Use the getRecommendation API call to get the one-to-one product recommendations in real time.
Click on Show API call button, near of Get recommendation button, to get the API call example.
Use the Get recommendation topic on API documentation for more information.
6. Events API
6. Events API
The events API allows to introduce the customer navigation and interaction data to the Artificial Intelligence to be used for a more accurate recommendation.
There are four events to be registered:
Impression: A product impression.
Click: When the user clicks on the product. Is a visit on the product view.
Add to cart: When the user adds the product to the cart.
Purchase: The conversion moment.
1. Create Event manager
1. Create Event manager
By default there is no Event manager created.
Click on Create event tracking on the top right in the event management box.
Follow these steps to create a new Event manager:
Write the event manager name.
Click on create event manager.
2. Register event API
2. Register event API
With the Event manager with the Active status you can start to register new events with the registerEvent API call.
More information on the Register Event topic from the API Documentation.
Page updated
Report abuse