A washroom in a busy lobby and one in a quiet back-of-house area rarely experience the same level of usage, yet both have traditionally been cleaned according to the same schedule.
Justin Chay, Managing Director of CSP Maintenance, explains how the company is using sensor data and real-time monitoring to move towards demand-based cleaning operations, allowing teams to respond to actual building conditions rather than fixed cleaning cycles.
Why did CSP move to demand-based cleaning?
The traditional frequency-based cleaning model no longer reflected how modern buildings operate. A washroom in a busy commercial lobby and one in a quieter back-of-house area were often cleaned on the same schedule regardless of how many people used them. This meant some areas were cleaned more often than necessary, while busier locations could experience service gaps during peak periods.
Fixed cleaning schedules also made it harder to respond when conditions changed. Supervisors and cleaners may have been aware that certain locations were experiencing unusually high usage, sudden spikes in foot traffic, or cleanliness issues, but they had limited flexibility to redeploy staff. As a result, cleaners were not always deployed where they were needed most.
At the same time, Singapore’s cleaning industry continues to face labour constraints and rising manpower costs under the Progressive Wage Model. Supervisors often had limited visibility into where resources were most urgently needed, while performance was still largely measured by hours worked rather than cleanliness outcomes.
These challenges led CSP to adopt a more demand-responsive operating model. Instead of relying on fixed schedules, cleaning activities are increasingly guided by building conditions, occupancy patterns, utilisation data, and feedback from staff on the ground. This allows resources to be deployed more efficiently and helps maintain more consistent cleanliness standards across sites.
What technology powers CSP’s smart-cleaning operations?
Our operating model integrates multiple layers of smart-cleaning infrastructure, including people-counting systems, bin fill-level sensors, connected washroom monitoring devices, and digitally enabled cleaning equipment that transmits operational status and usage data in real time.
Across our portfolio, different sites may operate on different technology ecosystems depending on client requirements and deployment environments. The greater challenge is not collecting the data, but turning multiple live data streams into a consistent operational workflow across all sites.
Our operations teams work through integrated monitoring platforms that consolidate site data into actionable information. These systems generate alerts, task triggers, escalation workflows, and performance visibility for supervisors and cleaners on the ground.
Live site data allows supervisors to prioritise deployment based on actual demand rather than predetermined cleaning cycles. Tasks are routed directly to ground teams with location-specific instructions and priority levels, enabling faster responses and more efficient use of manpower.
How does site data influence day-to-day cleaning decisions?
Together with our technology partners, we configure site-specific thresholds based on each environment’s traffic profile and operational requirements. For example, washrooms may trigger service alerts after reaching predefined occupancy thresholds, while waste management systems generate notifications once bins approach capacity levels.
Once those thresholds are reached, the system automatically generates work instructions with location, urgency, and task details, which are routed directly to operational teams. This removes reliance on fixed cleaning intervals and allows response times to align more closely with actual site conditions.
Supervisors monitor live dashboards throughout the day and redeploy manpower based on changing demand. A cleaner assigned to a lower-utilisation zone may be redirected immediately to a high-traffic lobby or washroom experiencing a surge in usage.
As a result, cleaning activities are guided by live building conditions rather than fixed schedules, helping teams respond more quickly to changing demands across a site.
What improvements has CSP seen since implementing the model?
Since implementing this operating model, we have seen improvements in operational efficiency, service responsiveness, and performance visibility.
From an operational standpoint, demand-driven deployment has reduced unnecessary cleaning cycles while allowing manpower to be concentrated in higher-utilisation areas. At more mature deployments, this has improved productivity and enabled resources to be redirected towards activities such as deep cleaning, preventive maintenance, and targeted disinfection.
Service responsiveness has also improved, particularly in high-traffic environments where conventional scheduled cleaning models often struggled to respond quickly enough between cleaning rounds. We have observed more consistent cleanliness audit performance and fewer reactive service complaints due to faster intervention.
The technology has also improved visibility for both our teams and clients. Task completion, response times, audit outcomes, and deployment histories are digitally logged and traceable through operational platforms. This allows service performance to be assessed using operational data rather than manual reporting alone.
The visibility created by these systems also allows clients and service providers to review operational data together and refine deployment strategies, response priorities, and service expectations over time.
How is the system adapted for different sites and environments?
We have progressively deployed this operating model across a range of environments, including commercial office developments and institutional campuses. These environments differ significantly in occupancy patterns, traffic levels, and service expectations.
The deployment typically covers common-area cleaning, washroom operations, waste management, and high-touchpoint maintenance, supported by IoT-enabled monitoring and live operational visibility.
Each deployment is configured differently based on how a building is used. For example, a high-footfall public environment with fluctuating traffic patterns requires different alert thresholds and escalation procedures from a corporate office floor with relatively stable occupancy levels.
At the start of each deployment, workflows, alert thresholds, escalation paths, and manpower response structures are configured for the site. This ensures the technology supports the operating requirements of each environment rather than relying on a one-size-fits-all approach.
Not every site requires a full-scale IoT deployment. Smaller projects with limited manpower requirements, such as sites operating with fewer than 10 cleaning staff, may not generate sufficient productivity gains to justify a fully integrated deployment.
In these cases, the focus may shift towards selective technology adoption, process improvements, and stronger supervisory workflows rather than extensive sensor deployment. The objective is to match the approach to the scale, complexity, and operational requirements of the site.
What challenges did CSP face when scaling smart-cleaning operations?
The technology deployment itself was relatively manageable. The more complex challenge was operational transformation and workforce adaptation.
One technical challenge was interoperability across different technology ecosystems. Different sites operated on different platforms, each with its own dashboard structure, alert logic, and reporting format. Ensuring consistency across these environments required us to standardise the operational data points and response protocols most relevant to our workflows.
Connectivity conditions also vary significantly across building environments in Singapore, particularly in basements, older facilities, and back-of-house areas. These factors must be considered during deployment planning to ensure system reliability and operational continuity.
However, the largest challenge was change management on the ground.
Many operational personnel had spent years working within fixed-route cleaning structures. Moving to a responsive deployment model driven by live data required a significant shift in mindset and workflow discipline. Initial resistance was less about the technology itself and more about workforce confidence and familiarity with new ways of working.
We addressed this through phased deployments, multilingual training programs, structured onboarding, and supervisor-led pilot implementations. Once teams experienced how the systems reduced unnecessary movement, improved task prioritisation, and provided clearer operational direction, adoption improved significantly.
A critical part of this transition has been continuous upskilling and retraining to help personnel use IoT-enabled workflows and digital tools effectively. The success of any smart-cleaning deployment depends heavily on the competency and confidence of the people operating it.
Another important lesson was the need for careful calibration. Alert thresholds must be tuned to operational realities. Excessively sensitive triggers can create alert fatigue, while thresholds that are too loose reduce responsiveness and affect service quality. Refining these settings required ongoing adjustments based on historical data and feedback from supervisors and ground teams.
Operational transformation also requires adaptation on the client side. Transitioning from fixed manpower visibility to performance-based service models requires trust, transparency, and close coordination between both parties. The most successful implementations are those where clients and service providers work together to refine workflows, response expectations, and operational priorities over time.
