customer churn and scoring analysis
TRANSCRIPT
Churn Analysis
How can you prevent losing your customers with Microsoft Azure?
Dawid Detko and
Wiktor Zdzienicki
Churn Analysis
Dawid Detko • Principal Data Architect • Microsoft MVP Data Platform • Data Science Project Manager • Power BI Coach
• E-mail: [email protected]
• Twitter: DDetko
• Blog Predica
• Advanced Analytics Associate Project Owner • Data Scientist
• E-mail: [email protected]
• Twitter: @wizden5
• Blog Predica
Wiktor Zdzienicki
• Customer churn • It’s a crucial metric for a growing number
of companies in a variety of industries.
• Traditional consumer behavior patterns
disappear, along with the decrease of
customer loyalty and retention.
• Research shows that the cost of
acquisition of a new customer is much
higher than the cost of retaining current
clients - therefore, churn analysis is of the
utmost importance.
• Companies must make use of the data
to analyze not only the probability of
churning but also combine this with an
evaluation of customer value.
• Challenge • How to use analytics to determine the
probability of customer churn?
• How to combine it with scoring and
estimation of customer value?
Prevent losing your customers
Make use of advanced analytics
• Predictive analytics and machine learning
models allow us to extract patterns and
insights from numerous sources of data.
• The result of such analysis are customer
segments with churn probability calculated
for every one of them. This combination
provides valuable information for
businesses.
Machine Learning models used for analysis
Logistic regression
Random forest
Neural network ANN
Boosted decision trees
Featured solutions
Churn analysis
Lower cost of generating campaigns1 Specifying the
negative effects of potential churn2 Tailor-made and
targeted marketing actions3
z Identification of the reasons which made a customer leave.
z Final effect – estimating probability of situation in which a customer stops using our products or
services.
Benefits
RFM analysis
RECENCY FREQUENCY MONETARY
The freshness of the customer activity, be it purchases or visits.
The frequency of the customer transactions or visits.
The intention of customer to spend or purchasing power of customer.
• E.g. Time since last order or last engaged with the product
• E.g. Total number of transactions or average time between transactions/engaged visits
• E.g. Total or average trans-actions value
RFM analysis
Simple and effective customer evaluation
1 Predicting marketing campaigns effect2 Better targeted
marketing actions3
z Estimating how much is the customer worth for the company, based on customer activity and
relationship with the brand.
z The final effect is the customer segmentation, which allows choosing
the most appropriate marketing strategy for every segment.
Benefits
Customer scoring
More accurate customer classification1 Enhancing knowledge
about clients and generating personalized recommendations by discovering similar groups
2 Better targeted marketing campaigns3
z Estimation of customer value from the business point of view taking into account multiple variables.
z The result of the analysis is a single number used to classify a customer.
Benefits
The process of customer churn and scoring analysis
Sample definition
Determine what exactly churning means for the company
“The main goal of the project is to detect which customers are likely to churn, meaning that they won’t purchase any service or product during the following 12 months.”
Stages of the analysis
Data integration and analysis
3 weeks - 3 months approx. 1 month for 3-5 models
Model training
Generating predictions
Accuracy validation
Confusion matrix analysis
Model tuning
Integration of data sources such as
CRM or data lakes
Data cleaning and feature engineering
Statistical analysis and selection of variables
used in modeling
Data visualization in a BI tool
Interactive dashboards and storytelling
Multidimensional analysis
RFM analysis
1 month for every 10 pages of reporting
Machine learning modeling
Data visualization
Project stages
Churn analysis and customer scoring
A powerful combination
Combine churn analysis... with customer scoring for more actionable insights regarding
the risk of migration of your most valuable customers
Recency Frequency Monetary analysis model
K-means clustering Agglomerative
DivisiveHierarchical clustering
Analytical methods used for customer scoring and segmentation
Customer segmentation
Cost optimization
Optimization and
control of costs related
to cloud operations
Event Hub
Logs (unstructured)
Media (unstructured)
Files (unstructured)
Business/custom apps
(unstructured)
Azure Storage
Power BI
Azure SQL Database
Azure Cosmos DB
Azure Machine Learning
Azure Analysis Services
Azure Stream Analytics
Azure Databricks
Azure DataFactory
IoT Hub
INGEST STORE PROCESS & MODEL
EXPLORE
Azure services used for customer scoring and churn analysis
Key takeaways
Churn analysis
• It’s a crucial metric which
needs to be taken into
account by companies
in a variety of sectors
• Effective use of multiple
sources of data
leads to tailor-made
marketing actions
What does it cover?
• Estimating the probability of
customer churn
• Calculating customer score
• Applying RFM analysis
Azure Toolkit
• Azure provides the entire
toolkit to cover churn
analysis process from a
technical point of view
• Apply Azure tools to speed
up deployments and use
state-of-the-art advanced
analytics solutions
Interested? Contact us