Lapsation curve: As different businesses have different customer latencies, defining the lapsation period for the brand being analyzed will help in categorizing the customers. We just need to see what cumulative % of repeaters repeats in what time interval. Usually, the threshold is 80%.

Drop rate: The retention curve or the Drop rate is a visual representation of the average retention of some dimension. Here we will study the retention curve based on the number of visits of a user.

  • Retention of Brand/Drop rate of the brand: The customer loyalty of the brand can be checked by looking at the drop rate curve. This is one of the vital signs of a business as it tells how a customer stays with a brand (in terms of visit).

Customer snapshot: This gives an overall view of all the existing customers of a brand by grouping them in different buckets like One timer and Repeaters (based on visits). This also provides a snapshot of the customers' demographics captured.

Steps to run Notebook

The Databricks notebook has different commands where codes are written. The Cmd{number} represents that command line that you need to refer to.
  1. Open the link for the Notebook relevant to your cluster- SEA cluster, India cluster, EMEA cluster
  2. Clone the Notebook into your workspace.
  3. Cmd 1: Import all the required Python libraries.
  4. Cmd 2: Read instructions on how to use the notebook.
  5. Enter the following data into the text box
    Org_id, Start date, End date, Active Period (Initially, one can input any random number between 0-300 to get the lapsation curve, giving us the lapsation period of the brand. After getting the output, you have to run the notebook again with the lapsation period obtained earlier.)
  6. Cmd 6: Run the single view command.
  7. Cmd 10: Visualizes the lapsation curve, provides the lapsation period as per curve (Point of latency where the customer% reaches 80%)  Read comments on the command cell.
  8. Cmd 11:  Define the lapsation period of brand and set rules for active, lapsed, and lost status.
  9. Cmd 12-15: Checking Retention of the brand
    1. Cmd 13: Data manipulation.
    2. Cmd 14: Visualizes the drop rate curve. Provide comments on the graph as per visuals.
    3. Cmd 15: Provide overall comment on the brands' retention.

Understanding the Sample Output

In this graph you can see that 80% of customers return to shop within 300 days from last visit. Which means if a customer is supposed to repeat with the brand, there is an 80% probability that he or she would return in 300 days. Only 20% of customers have a gap of more than 300 days between their visits. Hence, 300 days is considered as the lapsation period for the brand.

Retention curve indicated the stickiness of the brand, that is how long a customer stays with the brand. The following curve indicates that the brand has bad retention of customers. Only 25% of customers stay with the brand after the first visit. Only 4% of customers stay with brand for at least five visits.