At a glance
- Role
- Product Manager, part-time contract. Hospital-side product owner for the EMR
- Product
- The hospital’s electronic medical record (EMR) system. It is a vendor product, so I do not name it here
- Period
- 2025 – 2026
- Setting
- Private multi-specialty hospital, inpatient and outpatient
- Modules
- Ward rounds, discharge, billing, pharmacy and medication orders, reporting
- Team
- In-house developers, with the vendor for the core system
- Worked with
- Physicians, nurses, pharmacy, administration, procurement and finance
- Tools
- SQL, Excel, Power BI, Microsoft Copilot Studio, n8n, Zapier, Make
Context
Duchess International Hospital runs on a vendor-supplied EMR. I joined on a fixed-term, part-time contract as the hospital-side product owner for it. Some modules were being rolled out for the first time. Others were already live and were not working for the people using them.
I decided how the system was configured and rolled out, wrote the requirements, and prioritised one backlog for the in-house developers. They built the configuration and forms, the integrations, the reports and dashboards, and the automations that sit on top of EMR data.
The work crossed department boundaries. A change to discharge in the EMR could affect pharmacy, billing and nursing, so the teams needed a shared view of the problem and the proposed change.
The problem
Patient flow and billing were breaking down, and each department had a different explanation. I was asked to find the causes and decide what the EMR, and the workflows around it, should do differently.
In discharge, billing and medication use, operational data challenged the initial explanation.
Discharge: the wait came after the decision
People blamed the doctor's sign-off. The wait was in pharmacy and billing clearance after the decision to discharge had been made.
Billing: the errors started on the ward
People blamed the billing desk. Charges were being entered late or incompletely at the point of care, so the invoice was wrong before billing saw it.
Medication: stock, not pharmacy
People blamed pharmacy. The causes were stock gaps and slow procurement, one step behind the dispensary.
Five groups, five sets of priorities
Clinical, pharmacy, administration, procurement and finance each had urgent needs, in a heavily regulated setting where safety and compliance come first.
Finding the cause
Workshops and journey mapping
I ran workshops and journey-mapping sessions with physicians, nurses, pharmacy, administration and procurement to find where workflows broke down.
Current-state maps
I mapped current-state patient flow and billing across clinical, pharmacy, administrative, procurement and finance teams, and reviewed workflows across inpatient and outpatient departments.
What people said, checked against the data
I combined what staff described with operational data, and used the data to find where delays really occurred, which was not where people assumed.
Product decisions
Redesign the workflow before configuring the system
I redesigned the ward-round, discharge, billing and medication-use workflows first, then configured the EMR to match them.
The evidence pointed to problems in the sequence of work. Building the existing sequence into the EMR would have preserved the delays.
Start discharge clearance early
In the EMR, pharmacy, billing and admin clearance now start as soon as a discharge is planned, not after it is signed.
The wait sat after the decision, so the fix was to move that work earlier and run it alongside the clinical steps.
Fix billing errors at the source
Charges are captured in the EMR at the point of service under agreed billing rules, and checked automatically before the invoice is raised.
The errors came from what was entered on the ward. Checking invoices at the end would not have removed them.
Build agreement with billing and finance
Billing and finance were wary of automation. I showed them the error data before proposing any change.
The data showed that errors began upstream of billing. Establishing that shared understanding helped the team support the change.
One EMR backlog for every department
I turned competing departmental requests into a single prioritised backlog for the developers, judged against patient and operational impact, and ran regular demos.
Department heads were protecting their own priorities. One visible list and regular demos showed each of them where their request stood and why.
Check AI outputs before staff rely on them
I wrote the system prompts and instructions for the agents and automations, and evaluated their outputs for accuracy before staff relied on them.
In a clinical setting a confident wrong answer is a safety risk. Compliance, clinical safety and usability had to be balanced in every release.
Discharge, before and after
Before
- Doctor decides to discharge
- Discharge is signed
- Pharmacy, billing and admin clearance begin
- Patient waits for clearance
- Patient leaves
Clearance began only after sign-off. This is where the wait was.
After
- Discharge is planned
- Pharmacy, billing and admin clearance begin
- Discharge is signed
- Patient leaves
Clearance starts when discharge is planned and runs alongside the clinical steps.
What shipped
Ward rounds module
Decisions are recorded in the EMR during the round, so pharmacy, billing and nursing can act straight away.
Discharge module
Pharmacy, billing and admin clearance start in the system when discharge is planned.
Billing module
Charges are captured at the point of service under configured billing rules and checked automatically before the invoice.
Pharmacy and medication orders
Orders are checked for risk before dispensing, and stock levels are visible before a medicine runs out.
Integrations
In-house developers connected the EMR to the hospital’s other systems.
Reporting and dashboards
KPI dashboards covering patient, medication and inventory measures, built from EMR data in SQL, Excel and Power BI.
Guides for staff
FAQs, user documentation and training for each module.
To protect patient confidentiality, this case study uses aggregate results and simplified diagrams. It includes no patient records or identifiable clinical information. The EMR is a vendor product and is not named.
Delivery
I wrote the requirements and acceptance criteria for each change and prioritised the work for the in-house developers.
Staff tested each change and signed it off before go-live, with a follow-up review afterwards.
A champion network of department superusers drove adoption, gathered feedback and kept changes visible across teams.
Readiness assessments, stakeholder engagement plans and training came before each module rollout.
I led the cross-department rollout, balancing compliance, clinical safety and usability.
Measuring success
Wait times were measured at four points: discharge decision to the patient leaving, outpatient registration to seeing a clinician, prescription to dispensing, and service delivery to a final invoice.
Billing errors were counted from invoices corrected or reissued, patient disputes, missed charges found on review, and sample audits of invoices.
Compliance meant every department meeting the hospital's medication-safety policy and its external regulatory standards.
Dashboards tracked patient, medication and inventory KPIs, and adoption was monitored after each deployment.
Outcomes
Process and patient wait times cut by 50%.
Billing errors reduced by 70%.
100% compliance with safety and regulatory standards across departments.
The results came from changes in the EMR and from the workflow changes around it. I do not separate the two.
Lessons learned
Observe the work as it happens
Time with clinical, pharmacy, administration and procurement staff revealed handoffs the process documents did not capture. Those observations made the workflow maps useful.
The team that gets blamed is rarely the cause
Discharge, billing and medication delays were each pinned on the last team to touch them. In all three the cause sat one or two steps earlier.
Set the baseline before the change
Baselines against the legacy manual processes were what let me show the impact of each change, and what kept leadership behind the next one.
Go-live is the midpoint
Adoption problems show up after launch. Spotting them early and following up with staff after go-live mattered as much as the rollout itself.
Explore the product
Agents and automations
Built with Microsoft Copilot Studio, n8n, Zapier and Make on top of EMR data, to take repetitive work off staff.
Request routing+
Requests from departments are routed to the right owner automatically.
Stock alerts+
An alert fires when inventory falls below a set level.
Automatic reporting+
Reports and dashboards refresh without manual work.
Billing rule check+
Billing entries that break a rule are flagged before the invoice is raised.