I spent over ten years working with hospitals on operational improvement, and this stayed with me – A team would go into an organisation, identify opportunities, implement changes to reduce pressure and then return, sometimes within the same year, to do it all over again.
It wasn’t a one-off. It happened repeatedly. Most of my colleagues saw the same thing. I found that really unsettling. The valuable work that the NHS and management consulting teams put in just didn’t stick. I didn’t understand why but I knew I had to do something differently to change that pattern.
I decided to leave management consulting and took on a secondment to help design the target operating model of a digital control centre for an NHS Trust considered one of the most digitally mature.
To design something that works, I needed to understand how hospitals actually operate — hour by hour.
So I observed and got stuck in.
I sat with operational teams managing patient flow in real time. I spoke to clinicians and managers about the challenges they face. And I got involved directly — supporting discharge, admissions and transfers.
What I saw was very different from how we often describe healthcare systems. We often say healthcare is “complex”. And it is, but not in the way we usually mean.
At a high level, demand can look predictable. You can model how many patients arrive at the front door each day with a fair amount of accuracy but the reality is that every patient is different.
Different conditions. Different social needs. Different demographics.
And to meet that demand, hospitals must coordinate staff, beds, diagnostics, equipment, and supporting services (in and out of hospital) — all at the same time. This regularly happens under pressure with limited resources and almost always with incomplete information.
And that coordination doesn’t happen once a year. It happens every hour.
But most hospitals are trying to manage this with operating models that are, in comparison, relatively static. Resource planning is rigid with limited flexibility within shifts. In reality, teams on the ground struggle to adapt to what is actually happening in real time. So teams do what they have to do, firefighting more often than they would like.
Over time, workarounds are introduced. People adjust the way they operate, consciously and sub-consciously, to make things function for them and their teams. They do it with the best intentions — to keep patients moving, to reduce pressure, to help colleagues.
But these adjustments introduce friction elsewhere. Processes drift and dependencies break. The system is no longer working as planned, without a way to analyse variations, it is difficult to avoid inefficiencies.
I remember seeing managers trying to expedite care by booking patients into the earliest available appointment — only to find those patients needed to return because diagnostic results weren’t ready.
It wasn’t a failure of effort. It was a failure of coordination. That was a turning point for me. Up until then, much of the work I had been involved in, like much of the industry, was focused on modelling demand and capacity based on volumes. How many patients. How many attendances.
But that view doesn’t reflect how care is actually delivered. It doesn’t reflect the flow of patients through the system, or what resources are needed at each step.
That led me to explore different ways of understanding healthcare operations. Eventually, I found process mining. For the first time, clinicians could see a representation of their system that matched their lived experience.
I remember working with A&E teams who immediately recognised what they were seeing:
- where patients were experiencing unwarranted variation
- how much delay was being introduced at each step
- which parts of the process were creating downstream pressure
They didn’t need convincing. It wasn’t abstract. It wasn’t theoretical. It reflected reality and that made it engaging.
The next challenge was more difficult.
How do I help teams act on that understanding? Many of my NHS clients tell me that most of the tools they use in healthcare today don’t answer the most important question:
What should I do right now?
Dashboards show what has already happened. Command centres show where pressure is building. Neither tells an individual — whether that’s a nurse, an operational manager, or an executive — what action to take in a specific situation, at a specific moment in time.
So decisions are still made based on experience. Sometimes that works very well. But it isn’t consistent. It isn’t measurable. And crucially, we can’t always learn from it.
In many cases, pressure builds before it is fully understood. The levers available to teams on the ground are not up-to-date. Escalation happens later than it should and the system is already harder to manage leaving teams with limited and costly options.
I’ve come to believe that one of the missing pieces in healthcare is surprisingly simple. We have never really treated decisions as something we can structure, measure, and improve.
Hospitals run on thousands of decisions. Every minute. Every day.
These decisions are made by highly skilled people, doing their best in complex conditions with years of directly relevant experience in their local environment. We don’t capture those decisions properly today. We don’t connect them to what happens next. Therefore, we miss the opportunity to systematically learn from them.
That’s what I’m increasingly focused on now.
How we can take everyday operational decisions, the ones that quietly shape patient outcomes, and turn them into something structured, measurable, and improvable. And ultimately, to make healthcare systems work more fairly for the patients who rely on them.