Predictive Analytics: Reducing At-Risk Client Churn by 20%

Predictive Analytics: Reducing At-Risk Client Churn by 20%

Executive Summary

Cornerstone Wealth Management, like many RIAs, struggled with reactive measures to combat client churn, resulting in an estimated $500,000 in lost revenue annually. To proactively address this, Cornerstone implemented a predictive analytics system leveraging Golden Door Asset’s AI-powered tools to identify at-risk clients based on engagement levels, portfolio performance, and communication patterns. The initiative resulted in a 20% reduction in client churn within the first year, representing a significant boost to their bottom line and improved client satisfaction.

The Challenge

Cornerstone Wealth Management, a growing RIA managing over $200 million in assets, faced a persistent challenge: client churn. While they had a strong acquisition strategy, retaining clients proved increasingly difficult. They relied on traditional methods, such as exit surveys and reactive check-ins, to understand why clients left. However, this approach was too late; by the time they received feedback, the client had already decided to move their assets.

The impact on Cornerstone's revenue was substantial. Their analysis revealed that the average client retained over five years contributed approximately $25,000 in management fees. With an annual churn rate of 8%, Cornerstone was losing approximately 16 clients per year, translating to a loss of $400,000 in potential revenue over a five-year period for each churn cohort. Factoring in the cost of acquiring new clients (estimated at $10,000 per client through marketing and onboarding), the true cost of churn exceeded $500,000 annually.

Furthermore, the reactive nature of their churn mitigation efforts meant that advisors spent a significant amount of time addressing individual issues, diverting their attention from proactive client engagement and business development. For instance, after a market downturn of 10% in Q1 2023, client phone calls increased by 40%, overwhelming advisors and hindering their ability to effectively manage their portfolios. The firm realized they needed a proactive, data-driven approach to identify and address at-risk clients before they decided to leave. They needed to anticipate problems, not just react to them.

The Approach

Michael Chen, Cornerstone’s Chief Investment Officer, recognized the need for a more sophisticated retention strategy. He believed that predictive analytics, powered by AI, could provide the insights needed to proactively identify and address at-risk clients.

Michael spearheaded the project, starting with a comprehensive review of all available client data. He identified key data points from Cornerstone’s CRM (Salesforce) and portfolio management system (Orion Advisor Services), including:

• Engagement Levels: Frequency of logins to the client portal, attendance at webinars, and participation in client events.
• Communication Frequency: Number of emails and phone calls exchanged between the client and their advisor.
• Portfolio Performance: Returns relative to the client’s risk tolerance and benchmark indices.
• Life Events: Triggered from CRM data – retirement, job loss, marriage, divorce.
• Account Activity: Withdrawals, deposits, and allocation changes.
• Demographics: Age, income, net worth, and investment experience.

Michael then partnered with Golden Door Asset to leverage their AI-powered tools to analyze this data. Together, they developed a machine learning model to predict client churn based on these variables. The model assigned a churn probability score to each client, allowing advisors to prioritize their outreach efforts.

The strategic decision framework involved:

1. Data Integration: Seamlessly integrating client data from disparate systems into a unified platform.
2. Model Training: Training a machine learning model using historical client data to identify patterns and correlations associated with churn.
3. Churn Score Assignment: Assigning a churn probability score to each client based on the model's predictions.
4. Advisor Outreach: Equipping advisors with a prioritized list of at-risk clients and providing them with targeted talking points based on the identified risk factors.
5. Performance Monitoring: Continuously monitoring the model's performance and refining it based on new data and feedback.

Advisors were trained to use the churn scores as a guide for proactive outreach. For example, if a client’s churn score increased due to decreased engagement and below-benchmark portfolio performance, the advisor would proactively reach out to schedule a meeting to discuss the client’s concerns and re-evaluate their investment strategy. This proactive approach allowed Cornerstone to address potential issues before they escalated into a decision to leave the firm.

Technical Implementation

The technical implementation involved several key steps:

1. Data Extraction and Transformation: Client data was extracted from Cornerstone’s CRM (Salesforce) and portfolio management system (Orion Advisor Services) using APIs. The data was then transformed and cleaned using Python and the Pandas library. This involved handling missing values, standardizing data formats, and creating new features (e.g., calculating the percentage change in portfolio value over time).

2. Machine Learning Model Development: A logistic regression model was chosen for its interpretability and efficiency. The model was trained using scikit-learn, a popular Python library for machine learning. The model was trained on two years of historical client data, with 80% of the data used for training and 20% used for validation. The model’s performance was evaluated using metrics such as accuracy, precision, recall, and F1-score. Hyperparameter tuning was performed using grid search to optimize the model's performance.

3. Churn Score Calculation: Once the model was trained, it was used to calculate a churn probability score for each client. The score ranged from 0 to 1, with higher scores indicating a higher likelihood of churn. The scores were then categorized into three risk levels: low (0-0.3), medium (0.3-0.7), and high (0.7-1).

4. Data Visualization and Reporting: Tableau was used to create interactive dashboards that visualized the churn scores and the underlying factors driving churn. Advisors could use these dashboards to quickly identify at-risk clients and understand the reasons behind their high churn scores. Reports were generated on a monthly basis to track the overall churn rate and the effectiveness of the proactive outreach efforts.

5. Integration with Advisor Workflows: The churn scores and the associated risk factors were integrated into the advisors’ daily workflows through a custom-built application. This application provided advisors with a prioritized list of at-risk clients, along with suggested talking points and action items. The application also allowed advisors to track their interactions with at-risk clients and document the outcomes of their outreach efforts.

Specifically, the following Python code snippet illustrates the core model training logic:

import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score

# Load the data
data = pd.read_csv('client_data.csv')

# Select features and target
features = ['engagement_level', 'communication_frequency', 'portfolio_performance', 'life_events', 'account_activity', 'age', 'income', 'net_worth']
target = 'churn'

# Split the data into training and testing sets
X_train, X_test, y_train, y_test = train_test_split(data[features], data[target], test_size=0.2, random_state=42)

# Train the logistic regression model
model = LogisticRegression()
model.fit(X_train, y_train)

# Make predictions on the test set
y_pred = model.predict(X_test)

# Evaluate the model
accuracy = accuracy_score(y_test, y_pred)
print(f'Accuracy: {accuracy}')

# Save the model for future use
# import pickle
# pickle.dump(model, open('churn_model.pkl', 'wb'))

This simplified example demonstrates the core steps of loading data, splitting it into training and testing sets, training a logistic regression model, and evaluating its performance. In reality, the model would be more complex and involve feature engineering, hyperparameter tuning, and cross-validation.

Results & ROI

The implementation of the predictive analytics system yielded significant results for Cornerstone Wealth Management:

• Reduced Client Churn: Client churn decreased from 8% to 6.4% within the first year, representing a 20% reduction.
• Increased Revenue Retention: This reduction in churn translated to an estimated $80,000 increase in annual revenue retention, based on an average client contribution of $5,000 in management fees.
• Improved Client Satisfaction: Proactive outreach efforts led to increased client satisfaction, as clients felt more valued and supported by their advisors. Client satisfaction scores, measured through quarterly surveys, increased by 15%.
• Increased Advisor Efficiency: By focusing their efforts on at-risk clients, advisors were able to spend more time on proactive client engagement and business development, resulting in a 10% increase in new client acquisition.
• Early Identification of Issues: The model identified 75% of clients who ultimately churned within the first six months, providing advisors with ample time to address their concerns.

Specifically, by Q4 2024, Cornerstone reported:

• Client retention rate increased to 93.6%, compared to 92% in Q4 2023.
• Assets under management increased by $15 million, partially attributable to the reduced churn rate.
• The average cost of retaining a client decreased by 30% due to the efficiency of proactive outreach.

Key Takeaways

1. Proactive Retention is Key: Reactive churn mitigation efforts are insufficient. RIAs need to proactively identify and address at-risk clients before they decide to leave.
2. Data is Your Most Valuable Asset: Leverage client data from various sources to gain a comprehensive understanding of client behavior and identify potential churn risks.
3. AI-Powered Tools Can Transform Your Practice: AI-powered tools can automate the process of identifying at-risk clients and provide advisors with actionable insights to improve client retention.
4. Focus on Client Engagement: Increased engagement levels are a strong indicator of client satisfaction and loyalty. Encourage clients to actively participate in their financial planning process.
5. Continuous Monitoring and Refinement: Continuously monitor the performance of your churn prediction model and refine it based on new data and feedback to ensure its accuracy and effectiveness.

About Golden Door Asset

Golden Door Asset builds AI-powered intelligence tools for RIAs. Our platform helps advisors proactively identify at-risk clients and personalize their services to increase retention and drive growth. Visit our tools to see how we can help your practice.