Case Studies

Client

The client is a real estate investment firm specializing in property acquisitions in Houston and Dallas-Fort Worth (DFW).

Challenge

The client sought to improve the efficiency of its lead conversion process by:

• Predicting lead closure probability based on seller, property, and lead characteristics.
• Estimating expected revenue for each lead.
• Enhancing operational decision-making to improve lead prioritization and resource allocation.

Solution

Hexadec Analytics developed a data-driven lead scoring model that analyzed thousands of leads between 2021 and 2024. Key steps included:

1. Data Cleaning & Preparation

2. Predicting Lead Conversion

• Built a machine learning model to predict the probability of a closed deal.
• Used property, location, and lead-related features.
• Achieved 80% predictive performance using an ROC-AUC model evaluation.

3. Predicting Revenue for Closed Leads

• Created a revenue estimation model to forecast expected net revenue per lead.
• Found that mortgage rates and tax rates significantly impacted revenue.
• Model performance: $10,244 Mean Absolute Error (MAE).

4. Operational Insights & Recommendations

• Lead Prioritization: The firm could focus on high-probability, high-value leads based on predicted closure rates and expected revenue.
• Strategic Marketing Adjustments: Certain property types and seller motivations correlated with higher closure odds, informing better targeting strategies.
• Process Automation: Suggested integrating the model into the CRM system for real-time lead scoring and monthly recalibration.

Results

✅ Higher efficiency in lead targeting, reducing time wasted on low-potential leads.
✅ More accurate revenue projections, aiding financial planning.
✅ Data-backed operational strategies, leading to smarter decision-making.

Key Takeaways

📌 Lead scoring can significantly improve conversion rates when backed by robust data analytics.
📌 Integrating predictive models into CRM systems enables real-time decision-making.
📌 Data-driven insights on seller behavior and property characteristics drive better targeting and higher ROI.

Case Study: Enhancing Medical Device Demand Forecasting with Time Series Modeling

Client

A leading healthcare company seeking to improve the accuracy of demand forecasting for orthopedic implants (Total Knee Replacement, IM Nail-Hip, Plates Tibia).

Challenge

The client faced significant forecasting challenges due to:

• Disruptions from COVID-19 in 2020, making past data unreliable.
• Data inconsistencies due to missing values and irregular reporting.
• The need for a scalable model that could be used at a national level.

Solution

Hexadec Analytics developed a univariate time series forecasting model to predict monthly demand for medical devices while addressing data quality and scalability issues.

1. Data Preprocessing & Cleaning

Before building the forecast model, the team:
✅ Standardized historical data, ensuring at least six months of reports per year.
✅ Imputed missing values, treating April-May 2020 as zero due to pandemic disruptions and using a neighboring month approach for other gaps.
✅ Scaled forecasts to a national level, determining appropriate adjustment ratios for accurate projections.

2. Time Series Forecasting Model

• Built a univariate time series model capable of producing monthly forecasts for each medical device category.
• Analyzed the impact of 2020 data distortions and tested model robustness against pandemic-related shifts.
• Identified seasonal trends and demand fluctuations, allowing for better inventory planning and production scheduling.

3. Potential Model Enhancements

Future refinements include:

• Incorporating COVID-19 case loads as an explanatory variable to adjust for pandemic-induced demand fluctuations.
• Expanding the model to include external factors such as economic trends, hospital capacity, and surgical backlog data.

Results

✅ Improved forecast accuracy, ensuring better inventory management and reducing supply shortages.
✅ A scalable model adaptable to national demand, helping the company optimize its logistics and distribution strategies.
✅ Insights into demand volatility, enabling proactive planning for future disruptions.

Key Takeaways

📌 Data preprocessing is critical for accurate time series forecasting.
📌 COVID-19 disruptions require adaptive forecasting strategies.
📌 Scalable models help optimize inventory planning and reduce supply chain risks.

Case Study: Understanding Real Estate Investment Preferences Through Conjoint Analysis

Client

A real estate investment firm looking to refine their sales targeting and property competitiveness strategies using data-driven insights.

Challenge

The client needed a better understanding of investor preferences to:

• Identify key property features that drive investment decisions.
• Segment investors into meaningful groups based on their priorities.
• Improve sales and marketing strategies to match buyer demand.

Solution

Hexadec Analytics conducted a conjoint study to analyze how investors value different property attributes. The study surveyed hundreds of respondents, categorized into:

• Flip/Wholesale Investors – Those looking to buy, renovate, and resell properties.
• Rent-Oriented Investors – Those seeking long-term rental income.

1. Key Findings from Conjoint Analysis

The study examined nine investment-related real estate features, including price, square footage, bedroom/bathroom configuration, and school rating.

Key insights:
✅ School ratings have the highest impact – Investors strongly prefer properties near highly rated schools.
✅ Square footage is not a major factor – Investors prioritize other attributes over sheer size.
✅ Bathroom count matters more than bedrooms – A mismatch (e.g., too few bathrooms for the number of bedrooms) significantly lowers property appeal.
✅ Garage conversions are a deal-breaker – Properties with garages converted to living space see a sharp drop in attractiveness.

2. Investor Segmentation

Using hierarchical Bayesian clustering, the study identified three distinct investor segments in each group.

Flip/Wholesale Investor Segments

📌 Quality Seekers – Investors willing to pay more for high-quality, well-located properties.
📌 Bargain Hunters – Those prioritizing low-cost properties regardless of other features.
📌 Middle-Ground Investors – A mix of both, preferring good deals but considering property condition.

Rental Investor Segments

📌 Rental ROI Maximizers – Investors focusing on rental income potential over property price.
📌 Neighborhood-Sensitive Investors – Those prioritizing school ratings and location.
📌 Future Flippers – Investors who initially rent but seek properties with strong future resale value.

3. Applications for Sales Targeting & Property Evaluation

The insights from the study were integrated into the company’s sales and marketing strategy:

• Sales teams can now predict which properties appeal most to each investor segment and target them accordingly.
• Properties with high predicted demand can be marketed aggressively, while niche properties require a more focused approach.
• Competitive pricing and feature prioritization strategies were refined based on investor preferences.

Results

✅ Improved investor targeting, leading to higher engagement and faster deal closures.
✅ More effective marketing messaging, tailored to different investor priorities.
✅ A data-driven approach to property valuation, reducing guesswork in pricing strategies.

Key Takeaways

📌 Understanding investor preferences leads to better-targeted sales strategies.
📌 Data-driven insights can improve property pricing and marketing efficiency.
📌 Predictive modeling enhances decision-making in real estate investments.