A Customer Lifecycle Campaign is utilized to maintain and increase a customer’s engagement with the brand through multiple marketing strategies and offers at various times from their first interaction. Such types of campaigns are useful in creating personalized offers for different customers based on their recency. It also helps in reducing campaign costs while increasing customer retention and ROI, which can be increased by choosing the optimal time to target and which customers to target based on their previous transactions.
- Overall Repeaters lifecycle.
- Lifecycle based on the Recency segments.
|The logic and methodology used are the same for both sections, just the Recency segments lifecycle could have more touchpoints. Any one of these should be chosen as the final Repeaters Lifecycle.|
In the case of the Repeaters Lifecycle, the aim is to keep customers Active and reduce their latency. To find the touchpoints, we look at the behavior of the previous customers who have shopped with the brand before and have made at least 3 visits. The days taken between their nth and (n+1)th visit (latency) is taken as the reference (where n >= 2) and plotted in a graph (on the x-axis) with the corresponding number of customers (on the y-axis).
Section 1 (upto cmd 12)
The first step is to identify where the peaks are coming in the graph. These peaks correspond to the latency values which should be the ideal touchpoints for the Repeaters. The graph below clearly shows a peak at 180 days, thus becoming one of the touchpoints.
Cmd 8 in the notebook has the latency graphs. There will be 6 graphs in the output as the main graph has been broken down into buckets of smaller time periods to make the reading of the graph and locating of the peaks easier (the above graph shows one of these graphs for the days between 150 and 240). Please keep in mind the values on the y-axis as they would be different for the 6 graphs.
After getting some initial touchpoints from the graphs, make use of the table in cmd 9. This table has the data of the same customers divided into 10 percentile buckets based on latency. For each percentile, the mean, min, and max latency values are shown. The column with the mean latency values is the most useful.
Combining the graph touchpoints with this table will give the final lifecycle touchpoints. The use of the table is to see how many percentiles of customers would come in the target range. It’s important to not choose touchpoints close to each other because contacting the ‘yet to transact’ customers too quickly again is not advisable.
The final touchpoints are:
D is Day of purchase.
The graph in cmd 12 is a visualization of the max latency value for some important percentile values. It shows what is the maximum number of days that many percentages of customers take to return. It is used just to understand the repeater's natural behavior.
Section 2 (cmd 13 onwards)
Same methodology as used in Section 1, where the peaks are located and combined with the percentile table.
In Cmd 17, based on the recency segments provided the graphs will be shown for each segment based on the latency days and divided into 2 halves. So, each segment will have two graphs divided into equal days. An example is shown below for a part of the graph (Lapsed customers - 480 to 730 days).
We can get some initial touchpoints from these graphs based on the peaks as 480.540 and 650.
For each recency segment, around 3-4 touchpoints should be found. So if there are 3 recency segments then there will be 9-12 final touchpoints.
These touchpoint values are then combined and seen with the percentile latency table in cmd 18. The table will have the data of each recency segment combined column-wise. The final step is to follow the same process done for the Overall repeaters in section 1 and individually combine the graph and table for each recency segment to create the touchpoints. An example is shown below for Active Repeater and then all touchpoints combined.
- Don’t contact the customer too soon from the last touchpoint. Some brand knowledge is required here to determine the frequency of touchpoints. For example, a supermarket brand will have a higher number of touchpoints with fewer days in between compared to an Apparel brand.
- The y-axis values for the graphs (Customer counts) need to be seen carefully while determining the peaks.
- Customer counts will reduce as the latency increases, but don’t just consider the first few peaks as the touchpoints since those customers are going to be coming back to shop without needing a push so no need to start targeting from very early. The best way is to first find the prominent peaks in the higher latencies and work backward from there.
- The first 1 or 2 touchpoints offers can simply be product offers or product catalog information based on their previous purchases. Then the % or flat amount discount offers can be given with the aggressiveness of offers increasing as days since the last transaction increases. The avg. ATV and ABS of the customers/segments can be used to figure out the aggressiveness. DVS campaigns can be run for Lifecycle.
- Org_id of the brand.
- Start Date and End Date: Duration for which the notebook is to be run. Format: yyyy-mm-dd. (example: 2020-01-16)
- Lapsation Cuts: The cuts of the recency segments are comma-separated and without any spaces. Start from 0. Example: 0,365,480,730
- Lapsation Period Names: Names of the recency segments comma separated and without any spaces. Example: Active, Dormant, Lapsed, Lost.
The lapsation cuts and names in the examples above correspond to Active - 0 to 365, Dormant - 366 to 480, Lapsed - 481 to 730, Lost - >730
- Cmd 8: Latency Graphs.
- Cmd 9: Percentile table of Customers based on Latency.
- Cmd 12: Latency curve.
- Cmd 17: Latency Graphs.
- Cmd 18: Percentile table of Customers based on Latency.
Open your cluster-specific link provided for the Notebook.
|Lifecycle Campaign Touchpoints - Repeaters||India, SEA, EMEA|