See how Willow can help you cut costs, save energy, and operate your buildings more efficiently and sustainably.
Gartner’s Digital Markets Buyer Trends Survey reveals that 25% of organizations increased spending on facility management SaaS by 16–30% year-over-year, but many failed to realize ROI due to poor implementation and lack of user adoption. Keith LaRose’s recent LinkedIn post further illustrates these perils, and it had many of us reflecting on our learnings at Willow. We credit our Growth Mindset and commitment to continuous learning as key factors in successfully navigating challenges and proactively avoiding common pitfalls.
Willow has evolved over the years as our industry has gone through a hype cycle with Operational AI and digital twins.
Normalization of data from multiple sources in the knowledge graph marked a pivotal turning point. Through deep engagement with our lighthouse customers, we uncovered meaningful value, transforming Data into synthesized Insights that drive automated Actions. Additionally, we’ve learned a lot of things along the way that we can summarize below.
Systemic bottlenecks in building management range from project management to last-mile logistics – not things traditionally solved with product features. These dependencies can slow down the process of onboarding from weeks to months. So how do we approach this? At Willow, The Goal by Eliyahu M. Goldratt is required reading. We apply the Theory of Constraints to our e2e onboarding flow which starts with collecting documents and data to generate a knowledge graph for the buildings, layering floors, rooms and zones with assets and capabilities to stream telemetry from various data sources.
By the end of onboarding, customers can see skills firing across all systems and insights being generated in real time. Across product and deployment, in each planning cycle we revisit these five focusing steps: Identify the constraint, exploit the constraint (maximize its use), subordinate everything else to the constraint, elevate the constraint (increase its capacity) and repeat the process.
We’ve created an Onboarding Efficiency team and routinely review which step in the process is the greatest time sink. Do we see a pattern across customers? As an example, we observed that a lot of manual effort goes into hydrating the knowledge graph. Additionally, we noticed Edge deployments taking longer to get to insight generation relative to deploying Cloud connectors.
Our team redefined onboarding as a two-phase process: first, establishing telemetry flow, followed by data processing to produce actionable insights. We subsequently refined both our product and processes to eliminate bottlenecks and implemented optimizations through automation and enhanced tooling, ensuring that each stage remains focused on the desired outcomes. Furthermore, we have made a significant investment by integrating GenAI for subgraph generation, which streamlines repetitive tasks and allows personnel to concentrate on reviewing and approving graph content.
A leading customer at Willow has 25 different connectors that we’re ingesting data from. Rationalizing across different schemas, units and naming conventions, varying time intervals and time zones across this many connectors can be complex. Willow’s knowledge graph eliminates data silos and provides a consistent structure across varied building systems.
We lead with digital twin definition language (DTDL) and a comprehensive ontology to describe twins and their relationships to other twins in the system. From the very beginning, we recognized the constraints of relational databases and the power of traversing edges and nodes in an underlying graph. This enables seamless data ingestion, generation of calculated points to power up skills and create insights within the graph, and egress to other systems.
Data normalization is key for interoperability as well as scale. A BMS system might emit thousands of points, but only a few hundred may be key to generating meaningful insights. Structured data in a knowledge graph makes it easy to focus on data that is relevant and actionable. This makes things easy to replicate across hundreds of buildings and multiple portfolios.
Understanding and optimizing customer workflows drives feature investments in Willow. As an example, let’s follow the workflow of the Operations team of one of our lighthouse customers.
First, the customer’s team is notified about an insight indicating AHU-01 in One Bradford Square is running 24/7 and has a significant avoidable cost. Previously, their manual workflow was to create a ticket for AHU-01, provide details on steps to correct the issue, and then assign it to a technician, and let the ticket sync from Willow to a work order in their CMMS. Then a technician picks it up, works on it, and closes it.
The Operations team noticed that this insight occurs frequently and is consistently actioned. Willow responded by introducing an automated work order generation feature. A simple configuration automates and speeds up the entire process from insight generation to work order creation in the customer’s CMMS. Visualization now augments the workflow, while a 3D viewer makes it easy to locate the asset with the open work order and lights up open work orders on assets nearby, so facility managers can group together tasks for a technician who will be in the vicinity.
Fundamentals come first at Willow, and it’s not unusual to pause new feature work in favor of investing in reliability and performance. If you were to join our weekly Live Site Shiproom, you might hear the words ‘partial data outage’. Our goal is to catch flat-lining points proactively. QAQC checks, instrumentation in code and observability dashboards allow us to determine quickly if the problem lies outside Willow, like a network outage at a customer site, or within a specific subsystem in Willow.
A key learning for us has been to watch for false positives. Are data quality alerts the result of an issue in an actual production environment, or are they due to an active deployment in progress? Working cross-group and integrating connector deployment status into observability tools now allows us to weed out issues that can be safely ignored.
As we evolved the product, we’ve learned the importance of engaging our customers on outcomes and impact. Our experience leads with presenting actionable insights and Avoidable Cost, accruing direct energy savings as well as less tangible costs like wear and tear. As insights are actioned and technicians are dispatched to fix issues, customers can track what portion of the avoidable cost now turns into realized savings.
Something else we learned is that technicians completing a work order doesn’t always equate to actual root cause resolution. Identifying which insights remain unresolved even after work orders are actioned has been key to improving operations. Giving technicians diagnostic results and clear instructions on what to fix has led to measurable, visible gains in efficiency.
To quote psychologist Carol Dweck, “In a growth mindset, challenges are exciting rather than threatening.” By reflecting on our own outcomes and studying the approaches of others, at Willow we strive for continuous improvement. This mindset fosters innovation and adaptability, enabling us to streamline operations and deliver greater efficiency and value to our customers. Ultimately, learning from both wins and failures helps us build more resilient, customer-centric solutions.