One-to-one: CRM campaign

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

The execution of a recommendation leads to the hiper-personalization of the CRM campaigns. Every user gets a specific recommendation of products based on their behavior and their propensity to buy.

2. Create a recommendation

To create a recommendation, it is necessary to fill out a form with the following aspects:

• Name: Name of the execution. It must be unique and it can't be a previous one. It is useful to put a name that contains the date, the hour, and the minute at the moment of creation.

• Model: It is compulsatory to fill this part. The desired model needs to be selected from the list.

• Business Rule: The desired business rules need to be selected, it can be added as many as it is needed, but is necessary to put at least one.

• Number of recommendations: Number of products offered for each customer.

• Default products: This part requires to be filled with the ID Reference of the recommended products by default. It is necessary to be the same as the number of recommendations.

• Segments: This option allows to put the user segments that need to be recommended.

• Number of exploration days: This refers to the amount of data (in time value) that the neuronal network is going to use to create the recommendations.

• Creativity: It evaluates how creative the recommendation is going to be. It requires a number between 0 and 1. If it is closer to 0 it is going to be less loyal to a particular user based on the buying propensity and the variation between different customers it is going to be lower. If it is closer to 1, the option given will be more different.

• Name and location of the data extension: Once the recommendation is executed a new .csv file is created in the FTP in the location defined in this field.

3. Data visualization

The created recommendations are previsualized on this side as a list, it contains the name, the status, and the creation date.

When a recommendation is ordered, it takes a time to process all the data. While the process is in progress a blue RUNNING button will appear until it is done. If the process takes some time maybe it is necessary to refresh the page.

Once it is finished executing the word SUCCEEDED is going to appear in green.

If the word is not visible it means that it is in process, or something didn’t work.