Soundience has wide experience gaining customer insight from data, using modern analytic methods and tools.
There are several key areas where we have found that data and analysis lead to valuable, actionable insights:
- Customer lifetime value: we use survival analysis and machine learning techniques such as decision trees to understand expected customer lifetime and value
- Churn and retention: we have been able to measure and predict customer churn in subscription and consumer businesses
- Segmentation: in addition to traditional segment analysis using pre-defind segments, we have been successful in using machine learning to discover segments based on behavioural and transactional data
- Cohort analysis: by basing analysis on period cohorts of new customers or subscribers, we have been able to detect and predict changes in retention and activity of the customer base over time in a rigorous and insightful way
We have used these approaches to help companies in a number of consumer-facing sectors, such as financial services and packaged goods, to create more effective marketing budgets, saving significant money and increasing sales.
Tools and Approaches
We use a portfolio of tools and approaches to gain customer insight along different dimensions:
- Statistical analysis for finding basic relationships and trends
- Machine learning for more advanced pattern and trend recognition, as well as automated segment discovery (clustering)
- Survival analysis, for measuring the expected lifetime of customers, and discovering the drivers of different retention behaviour
- Simulation (system dynamics) for evaluating how a customer base will grow or change over time under different scenarios
- MarketPulse, our proprietary tool which discovers customers' response to marketing channels, campaigns, and offers, and helps to measure the ROI of campaigns
Most of our work is done using R and Python, two open-source environments that allow for rich and rapid analysis, and that have access to a wide variety of add-on packages for advanced analytics work.