Your hospital knows it has an operational problem but are they getting to the root cause?
Every week, hospitals struggle with the same question: why aren't we hitting our targets?
A Trust misses the 4-hour A&E target and diagnoses the problem as "staffing shortage". They increase clinical capacity. Performance doesn't improve.
Another Trust watches theatre utilisation drop and diagnoses the problem as "procedure inefficiency". They implement new processes. Nothing changes.
A third Trust sees long admission wait times and diagnoses the problem as"bed availability". They open more beds. The bottleneck persists elsewhere.
What if the real constraint isn't what you're tracking?
What if the first hospital's main issue isn't clinical capacity—it's porters? What if the second hospital's low theatre utilisation isn't the process—it's team cohesion? What if the third hospital's bottleneck isn't beds—it's lack of car park spaces?
These aren't theoretical problems. They're discoveries I've seen emerge from process mining analysis of hospital operations. They reveal something critical: hospitals are operating on assumptions about their own bottlenecks that don't reflect reality.[¹]
Why Hospitals Can't See Their Own Constraints
Here's the structural reason this happens.
Most hospitals operate across three tiers, each making decisions within their own part of the system. Many of these decisions happen in real time, and in practice, it is rarely possible to constantly check and coordinate across every level. As a result, decisions are often made in isolation.
Tier 3 (Frontline): Senior registrars, nurses, physiotherapists and other frontline staff see only parts of the wider operation. They experience the pressures and delays directly in front of them and make decisions based on the problems they are trying to solve in that moment.
Tier 2 (Management): Bed managers, operations coordinators and shift supervisors work between these moving parts. They deal with escalations, resource pressures and competing operational demands in real time, often with limited visibility of historical patterns or strategic context.
Tier 1 (Board): The COO, CEO and Medical Director see the organisation through aggregate metrics: bed occupancy, 4-hour performance, staffing ratios. They see the outputs of the system, but not the thousands of operational decisions that created those outcomes.
And here is what's critical:
Each tier interprets what is being presented to them in the moment.
Tier 3 sees workload increase and assumes the problem is staffing. Tier 2 sees escalation requests and assumes the problem is process and protocols. Tier 1 sees missed targets and assumes the problem is productivity.
They're all looking at the same hospital, but from where they each stand, it's difficult to separate the signals from the noise.
What Process Mining Reveals That Your Dashboard Cannot
You probably have a command centre. You probably have dashboards showing real-time bed occupancy, queue lengths, staffing levels. You can see what is happening right now.
However, those dashboards cannot show: the real constraints (signals) that will cause a chain reaction (noise) for linked events.[²]
A dashboard says: "A&E has 47 patients waiting."
Process mining asks: "Which of the 47 patients experienced the expected paths and which ones experienced unwarranted variations? What is causing those variations?"
The answer is never just "we need more staff".
The answer is usually something like: "Patients arriving at 11 am are being triaged by one nurse instead of two (decision: staffing allocation).That one nurse can only assess 5 patients/hour (constraint). The clinical lead implemented an enhanced process for senior clinicians to sign-off on blood requests to prevent duplicate requests (decision: capacity management). New discharge best practice is being implemented to complete discharges before lunch, which coincides with main visiting hours and peak outpatient appointments, and there is a long wait for car park spaces (constraint)."
All these results in longer wait times for A&E patients (consequence), outpatient clinics running late (consequence), and poor patient and visitor experience (consequence).
A dashboard shows the symptom. Process mining reveals the cause and effect of variations in a sequence of events.