CCRM Data Analytics 3 — Questions and Answers
Question 1: Which technique is used to reduce the number of variables in a client dataset while retaining the most important information?
- Random sampling
- Principal Component Analysis (PCA) (Correct answer)
- Cohort analysis
- A/B testing
Correct answer: Principal Component Analysis (PCA)
PCA is a dimensionality reduction technique that transforms correlated variables into a smaller set of uncorrelated principal components.
Question 2: A CCRM professional wants to evaluate the effectiveness of a new onboarding program. Which analysis framework is most appropriate?
- Funnel analysis comparing pre- and post-program conversion rates (Correct answer)
- Market basket analysis of client purchases
- Sentiment analysis of social media posts
- Cluster analysis of demographic segments
Correct answer: Funnel analysis comparing pre- and post-program conversion rates
Funnel analysis tracks clients through sequential stages, making it ideal for measuring onboarding program effectiveness at each step.
Question 3: In client data analytics, 'data wrangling' refers to:
- Presenting clean data to executive stakeholders in visual dashboards
- The process of cleaning, transforming, and organizing raw data for analysis (Correct answer)
- Encrypting client records to ensure GDPR compliance
- Archiving historical client interaction logs
Correct answer: The process of cleaning, transforming, and organizing raw data for analysis
Data wrangling encompasses the preprocessing steps needed to convert raw, messy data into a structured, analysis-ready format.
Question 4: A cohort analysis of clients acquired in Q1 shows lower 12-month retention than those acquired in Q2. What is the MOST actionable next step?
- Eliminate Q1 acquisition campaigns immediately
- Investigate differences in acquisition channel, onboarding experience, and early engagement for each cohort (Correct answer)
- Increase advertising spend for Q1 to compensate for losses
- Archive Q1 client records and focus exclusively on Q2 clients
Correct answer: Investigate differences in acquisition channel, onboarding experience, and early engagement for each cohort
Diagnosing why cohorts differ requires examining the variables that distinguish them, such as channel, onboarding, and early behavior.
Question 5: Which key performance indicator (KPI) directly measures how willing existing clients are to recommend the firm to others?
- Customer Effort Score (CES)
- Net Promoter Score (NPS) (Correct answer)
- First Contact Resolution Rate (FCR)
- Average Handle Time (AHT)
Correct answer: Net Promoter Score (NPS)
NPS measures the likelihood that clients will recommend the firm, classifying them as Promoters, Passives, or Detractors.
Question 6: When building a client risk model, overfitting occurs when:
- The model performs well on training data but poorly on new client data (Correct answer)
- The model is too simple to capture patterns in the training data
- Training data contains too few client records
- The model is applied to the wrong client segment
Correct answer: The model performs well on training data but poorly on new client data
Overfitting means the model has memorized training data noise rather than learning generalizable patterns, causing poor performance on unseen data.
Question 7: A relationship manager uses a decision tree model to classify clients as 'at-risk' or 'stable.' The model has 90% accuracy on training data but 62% on test data. What should the analyst do?
- Deploy the model immediately since 90% accuracy is excellent
- Apply regularization or pruning techniques to reduce overfitting (Correct answer)
- Replace the decision tree with a pie chart visualization
- Increase the training dataset size by duplicating existing records
Correct answer: Apply regularization or pruning techniques to reduce overfitting
The large accuracy gap between training and test sets is a classic sign of overfitting, which can be addressed through pruning or regularization.
CCRM Data Analytics 3
Which technique is used to reduce the number of variables in a client dataset while retaining the most important information?