Case Study
Knime is a versatile no-code/low code tool with a rich set of built-in nodes for data transformation and analysis. Today we use KNIME across multiple business units in delivering quick, robust solutions. From ETL jobs to data- modelling and advanced analysis, there is a Knime node to help you do the job with ease. In cases where you feel the need to use a custom Java Library, custom Python Library, or a specific function in R, the custom Java Snippet, Python Snippet and R Snippet nodes come to the rescue.
A US-based telecom industry giant wants to better understand the competitive presence within their footprint to develop targeted strategies for customer acquisition and retention.
The Marketing data sources within the company did not have a census-block level tagging for the prospect and customer universe. Census-block level data provides a more accurate representation of the competitive environment as it considers local competitors and avoids the potential issue of averaging out data across larger zip code boundaries. Summarizing competitor-presence data at zip level over-estimated the total competitors and the maximum speed provided.
Census Block tagging for the Customer and Prospect Universe; thus arose the need of using Geo-spatial Tagging in R. Using Lat-Long mapping with Census Block Shape Files provided by US Census Bureau.
Some prospects did not have Lat-Long mapping available. The challenge was then to generate the latitude and longitude mapping for these addresses and then overlay competitor data at census block level.
Extrapolating the Competitor-presence data to the entire footprint. There was a need to expand the universe to cover the entire footprint.
R has a rich ecosystem of libraries for handling geographic data and geo-spatial tagging. We extensively used Knime for pre-processing and summarizing the data. R Snippet node provided by Knime came in handy here. We could use `sp`, `rgdal`, and `rgeos` to work with shapefiles in the R snippet node easily.
To generate Latitude and Longitude for records missing this information in the data, we used a combination of Knime and traditional Java functionalities. Java has custom pre-defined functions for standardization and address clean-up. Embedding Java Snippet in Knime was a quick solution to standardize the addresses. We then used Google API to generate latitude and longitude for these addresses.
We used Nearest Neighbour algorithm and extrapolation in Python to expand the universe to neighbouring census-blocks.
Overlaying competitor presence data with the Marketable and Customer universe helped the company to identify focus- zones for planning their acquisition and loyalty strategies. The recommendations we shared with them helped reduce customer churn by over 30% month over month. Analyzing competitor presence at the census block level offers a more nuanced and accurate perspective on the local market and the ability to embed Java, Python, and R in KNIME helped us deliver this solution in record time and was very well accepted by the stakeholders.
Staying informed about Competitors is essential for sustained success and growth of a company. From proactively responding to market changes and potential threats of a new competitor entering the market to identifying gaps in the market and developing custom solutions to meet customer demands, understanding the competitive influence becomes key to many go-to market strategies.
At Saarthee we have helped our stakeholders identify focus zones for targeted marketing for acquisitions and retention programs. Knime is a versatile no-code/low code tool with a rich set of built-in nodes for data transformation and analysis. In solving this business case, having the ability to incorporate custom Java Library, custom Python Library, or a specific function in R within the tool was helpful.
Get Started
Looking to team up? Connect with us through email or send a line.
