Practices often start a digital project with a tool: a new portal, a scheduling add-on, an AI assistant. A few months later the tool works, but the problem it was bought for is still there, because the real issue was a handoff nobody owned or a step everyone did differently.
Mapping a workflow before changing it takes a few hours and often saves months. You do not need special software; a whiteboard, sticky notes or a shared spreadsheet is enough.
Step 1: Choose one workflow and set its boundaries
Pick a single workflow that causes visible pain: new-patient intake, prior authorizations, referral processing, patient statements, or claim follow-up. Then write down:
- where it starts (for example, "a referral arrives by fax or portal"),
- where it ends (for example, "the patient is scheduled and the authorization is on file").
A clear start and end stops the map from growing into "everything the office does".
Step 2: Walk through it with the people who do the work
Sit with the staff who actually do the work, not only the managers who describe it. Ask them to talk through the last few real cases. For each step, record:
- what triggers it,
- who does it,
- which system or paper form is used,
- what information is needed,
- how long it usually takes,
- what happens next.
You will usually find that two people do the same task in different ways. Write down both.
Step 3: Mark handoffs and waiting time
Most delays happen between steps, not during them. Mark every place where work passes from one person, team or system to another, and estimate how long it waits there. A task that takes five minutes but waits two days in an inbox is a handoff problem, not a speed problem.
Step 4: List the exceptions
Every workflow has a "normal" path and a set of exceptions: missing information, a payer with unusual rules, a patient who cannot be reached. Count roughly how often each one happens. Exceptions are where staff spend disproportionate time, and they are also where automation most often breaks.
Step 5: Count volume and time
Even rough numbers help:
- how many cases per week,
- average hands-on time per case,
- how many need rework or a second touch.
These numbers become your baseline. Without them you cannot tell whether any change worked.
Step 6: Find the root causes
Look at the map and ask why the slow or error-prone steps happen. Common answers:
- information captured late or incompletely at the front desk,
- duplicate data entry between two systems,
- no clear owner for a step,
- unclear rules, so each person decides differently,
- a step that no longer serves any purpose.
Many of these can be fixed with a checklist, a template, a changed form or a clear owner, with no new software at all.
Step 7: Only then decide what to automate
After the simple fixes, look for steps that are:
- repetitive (the same action many times a week),
- rule-based (clear inputs and decisions),
- digital (the information is already in a system, not on paper),
- easy to check (a person can quickly confirm the result).
Those are good automation candidates. Steps that depend on judgment, sensitive conversations or incomplete information are usually better left with people, perhaps with better tools to support them.
Step 8: Plan for data and privacy
If the workflow involves patient information, write down which systems hold it, who can access it, and what a new tool would need to see. Any vendor that would handle PHI on the practice's behalf generally needs a business associate agreement. Give any tool only the access the task requires.
Step 9: Pilot, measure and adjust
Change one workflow at a time. Run the new version for a few weeks, compare it with your baseline, and ask the staff what got easier and what got harder. Adjust before rolling out further.
A short checklist
- One workflow, with a clear start and end
- Steps recorded with the people who do the work
- Handoffs and waiting time marked
- Exceptions listed with rough frequency
- Baseline volume and time captured
- Root causes identified, simple fixes made first
- Automation candidates chosen by clear criteria
- Data access and PHI handling planned
- Pilot measured against the baseline
How FZ Digital Solutions can help
Our AI consultancy work starts with exactly this kind of workflow review, so that any automation recommendation is based on how your practice actually runs. Where a custom solution makes sense, custom AI agents and workflow automation are scoped as a separate, optional project, with human review set according to the workflow's risk. For billing workflows, you can also see how our managed billing service works.
Book a consultation to talk about a workflow you would like to improve. Please do not send patient information or PHI by email.