About Client:
A large-scale real estate property management firm overseeing thousands of residential units across its portfolio, where maintaining strong occupancy rates is critical to revenue performance and long-term business growth.
With properties distributed across a large geographic footprint, the organization needed an efficient way to monitor occupancy conditions, identify potential risks, and reduce reliance on manual property inspections.
Background:
Before adopting smart technology and data-driven monitoring, the client faced difficulty identifying two critical occupancy scenarios:
Skippers: Tenants who vacate a property without providing notice.
Squatters: Unauthorized occupants residing in properties without permission.
Traditionally, these situations were discovered only after missed rent payments, neighbor complaints, maintenance visits, or physical property inspections.
This reactive approach meant potential vacancies or unauthorized occupancy could remain undetected for weeks or even months, resulting in lost rental income, unnecessary inspection costs, and increased security and liability risks.
Although the client had already deployed smart devices across properties, the data generated by smart locks, thermostats, and motion sensors remained largely disconnected and underutilized.
The organization needed a way to transform this IoT data into actionable intelligence that could identify occupancy risks earlier and help property managers respond proactively.
Challenge:
Inefficient & Costly Inspections: Physically inspecting thousands of properties required significant time and resources, making portfolio-wide monitoring expensive and difficult to scale.
Delayed Vacancy Detection: Properties abandoned without notice could remain unidentified for extended periods, increasing vacancy-related revenue losses.
Underutilized IoT Data: Smart devices generated valuable behavioral and occupancy signals, but siloed data prevented the organization from using those insights effectively.
Unauthorized Occupancy Risk: Delayed identification of potential squatters increased security concerns, property risks, and potential liability.
Reactive Property Management: Teams often responded only after a missed payment, complaint, or inspection revealed an issue rather than identifying potential risks earlier.
The challenge was to transform existing IoT data into predictive intelligence that could help property managers identify potential vacancy and unauthorized occupancy risks before they resulted in significant operational or financial impact.
Solution:
The client partnered with Supply Medium to build an IoT and AI-powered property intelligence ecosystem on Microsoft Azure, leveraging the organization’s existing smart device infrastructure.
The solution was built around three core pillars:
1. IoT Data Ingestion & Storage
Azure IoT Hub collected continuous streams of data from connected devices, including smart locks, thermostats, and motion sensors deployed across residential properties.
Azure Data Lake Storage Gen2 provided a centralized and scalable repository for real-time and historical IoT data.
Bringing these previously disconnected data streams together created a unified foundation for occupancy monitoring, behavioral analysis, and machine learning.
2. Machine Learning Intelligence
At the center of the solution was a custom machine learning model developed using Azure Machine Learning and XGBoost.
The model analyzed smart device activity to identify behavioral patterns associated with potential vacancy or unauthorized occupancy.
Potential Skippers: Extended periods of inactivity—such as no lock or unlock activity, thermostat adjustments, or motion events for seven or more days—could indicate that a tenant had unexpectedly vacated the property.
Potential Squatters: Unusual combinations of activity—such as motion events without corresponding authorized lock activity or behavioral patterns inconsistent with expected tenant usage—could indicate potential unauthorized occupancy.
To improve predictive accuracy, the model incorporated engineered features such as:
Time-Based Metrics: Time elapsed since the most recent lock, thermostat, or motion activity.
Frequency Patterns: Number and frequency of device events across daily and weekly periods.
Cross-Device Correlation: Relationships between signals, such as door unlock events followed by motion activity within a defined timeframe.
Combining multiple IoT signals enabled the system to assess occupancy risk more intelligently than relying on individual device events alone.
3. Actionable Insights & Alerts
Properties identified as potentially high risk triggered automated alerts through Azure Functions, notifying property managers through dashboards and mobile channels.
Power BI dashboards provided a portfolio-wide view of property health, highlighting properties with elevated vacancy or unauthorized occupancy risk scores.
Instead of manually inspecting every property, managers could prioritize locations requiring attention based on data-driven risk indicators.
This shifted property monitoring from a reactive inspection model toward a more proactive and targeted approach.
Outcome:
90% Faster Vacancy Detection: Potential vacancies could be identified in days rather than weeks, enabling property teams to investigate sooner and reduce avoidable rental income loss.
50% Lower Inspection Costs: Risk-based monitoring allowed managers to prioritize inspections for properties showing unusual activity, significantly reducing unnecessary site visits.
Earlier Unauthorized Occupancy Detection: IoT signals and predictive analytics helped identify potential unauthorized occupancy sooner, supporting faster investigation while reducing security and liability exposure.
Improved Property Management Efficiency: Automated monitoring reduced the need for repetitive manual checks, allowing property managers to dedicate more time to tenant relationships, leasing activities, and portfolio operations.
Smarter Portfolio Decisions: IoT-driven insights provided greater visibility into occupancy patterns, property activity, and portfolio-level trends, supporting more informed operational decisions.
The solution transformed existing smart device data into a proactive property intelligence capability, enabling the client to identify occupancy risks earlier, reduce inspection costs, protect rental revenue, and manage thousands of residential units more efficiently.