How Many Times Did This Patient Have to Chase You?

A patient is referred to you in early February. She calls the following week, reaches a voicemail, and leaves a message. Nobody calls back, so she calls again and books for late March. In March she was asked for insurance details she already gave the referring office. Her plan changed in January, so the coverage on file is wrong, which nobody notices until she is at the counter. She is seen. Two weeks later a bill arrives that she does not understand, and she spends forty minutes on the phone establishing that it is correct. Her follow-up is never scheduled because the instruction went into a portal she has not activated.

Count the contacts from her side. Six, maybe seven, most of them initiated by her.

Now ask who in your organization saw that whole sequence. Almost certainly nobody. The scheduler saw one call. Billing saw one call. Each interaction was handled competently and closed. The pattern only exists from where the patient was standing, and there is usually no report that shows it.

This is what administrative friction looks like in practice, and it is the reason patient journey automation is a different subject from front desk automation. The unit of analysis is not the call or the task. It is the whole path a person takes through your organization over months, and the number of times that path requires them to do the work. That is the question patient journey automation has to answer.

Friction Is Three Different Costs, Not One

The most useful frame here comes from health services research rather than from healthcare technology. Researchers studying patient administrative burden use a model that splits it into three distinct costs, and the distinction matters because they call for different fixes.

Learning costs are the work of figuring out how the system operates: which provider is in network, what a deductible means for this visit, whether a referral is required, what the practice actually needs from you before an appointment.

Compliance costs are the work of doing the required tasks: making the calls, completing the forms, producing the documentation, showing up with the right card.

Psychological costs are the stress, anxiety, and loss of control that accumulate through the process, which researchers note recur consistently in qualitative studies but are rarely measured.

That framework is the organizing structure of a November 2025 scoping review in Health Affairs Scholar, which synthesized 63 studies of patient administrative burden published between 2002 and 2024. The same review cites national survey evidence that nearly a quarter of insured US adults report delaying or forgoing care because of administrative tasks, with an impact the authors describe as comparable to financial barriers.

Read that last point carefully. Administrative friction removes people from care at roughly the rate cost does, which is why patient journey automation deserves strategic attention rather than operational attention. Most organizations treat the first as an efficiency issue and the second as a strategic one.

Your team is not the source of this. Friction of this kind is structural, produced by how coverage, referrals, and documentation are arranged rather than by anyone's effort at the desk. It is also, in meaningful part, addressable.

The Three Phases Where Friction Lives

The same review organizes the evidence chronologically across the patient journey, in three phases. The structure is useful because friction behaves differently in each, and patient journey automation has to be designed for each phase separately.

Seeking Care

Everything before any clinical contact. Working out coverage, finding an in-network provider who is accepting patients, understanding whether a referral is needed, and getting through to somebody.

Learning costs dominate here, and the evidence base is the most mature of the three phases. The review also notes secret shopper research in which more than seven in ten calls failed to secure a new patient or urgent care appointment, which is a striking finding for anyone who assumes the phone is a solved problem.

This is the phase where patients are lost before they exist in your system. They are invisible in your reporting precisely because they never became a record, and most patient journey automation never reaches them. Our piece on why patients cannot get through covers what that costs.

Receiving Care

Arranging and attending the visit: scheduling, intake, eligibility, prior authorization, pre-visit instructions, and the visit itself.

Compliance costs dominate. This is also the phase most healthcare technology addresses, and unsurprisingly the phase where organizations report the most progress. Two-thirds of the studies reviewed touched this phase, though the review judged the evidence less mature than for care seeking, since most of it comes from broad patient experience surveys rather than studies designed to isolate administrative tasks.

Resolving Care Issues

Everything afterward: understanding and disputing bills, appealing denials, chasing test results, requesting records, and coordinating between providers.

The review identifies this as the least developed evidence base of the three, with little systematic research on what patients do when something goes wrong. That gap is mirrored operationally. Most practices have some visibility into scheduling and almost none into how often patients are quietly repairing the organization's errors on its behalf.

All three cost types converge here, which is why this phase produces the most damaging experiences. A billing dispute involves learning what the charge is, complying with an appeals process, and absorbing the anxiety of an unresolved obligation. Prior authorization sits here too, and the AMA reports that 95 percent of physicians say it delays access to necessary care. Our overview of AI in revenue cycle management covers the operational side of that phase.

Why Small Frictions Compound

Here is the part that survives contact with a spreadsheet.

Take a simple path: the patient finds you, gets through, books, receives intake forms, completes them, has coverage verified, attends, and schedules follow-up. Eight steps. Suppose each one works 90% of the time, which would be a strong result for most organizations.

Run 90 percent eight times over and about 43 percent remain. Fewer than half the patients who started make it through the whole path without something requiring rescue.

No individual step looks broken at 90%. Every department reports acceptable performance. The failure is only visible end to end, and almost nobody measures end to end. This is the single strongest argument for treating patient journey automation as a distinct program rather than a series of departmental improvements: the compounding happens between the departments that are each doing fine.

It also explains a pattern we see often. From our experience with practices across the country, organizations that automate one step well are frequently surprised that patient experience scores barely move. Lifting one step from ninety to ninety-eight percent changes the end-to-end number by a few points. Lifting all eight is what changes the experience, and that is the case for patient journey automation over point solutions.

Most Automation Only Addresses One of the Three Costs

Now put the two frames together, because this is where most patient journey automation programs go wrong.

Look at what healthcare AI typically does: it answers the call faster, completes the booking, sends the form, verifies the coverage, submits the authorization. Every one of those reduces compliance costs. The patient does less work.

Learning costs are mostly untouched. The patient still has to understand what their coverage means for this visit, whether they need a referral, what happens next, and why. An AI Agent that books an appointment without explaining what the patient should expect has moved a task without removing the confusion around it.

Psychological costs are barely addressed at all, and can get worse. Anxiety in healthcare admin comes largely from uncertainty: not knowing whether the request landed, whether the authorization went through, whether the bill is right. Automation that completes tasks silently can actually increase this, since the patient has less evidence that anything happened.

The practical implication is specific. Proactive communication is not a nice finishing touch on top of automation. It is the part that addresses the other two costs. Confirming in plain language what was done, what happens next, and when is what converts a completed task into a reduced burden. Outbound reminders, status updates, and pre-visit instructions do more for psychological cost than any inbound capability, and Voice AI that only answers is only doing part of the job.

The same logic applies to language. A patient who cannot conduct the interaction in their own language faces all three costs at once, which our piece on serving multilingual patient populations examines in more detail.

Where Friction Should Stay

Not all friction is waste. Some of it is doing a job, and removing it would be a mistake.

Clinical triage should stay deliberate. A moment of human judgment about whether symptoms need urgent attention is friction that protects people, and speed is the wrong optimization target.

Informed consent should stay effortful. The point of consent is comprehension, not throughput, and a frictionless consent process is usually a defective one.

Financial hardship conversations should stay human. A patient who cannot pay needs a person with discretion, not an efficient collections path.

Identity verification should stay rigorous. The friction of confirming who somebody is protects the record and the patient behind it.

The useful test is whether a given step exists to protect the patient or to protect a process. Protective friction stays and should be designed well. Process friction, which is most of it, is what patient journey automation should target.

How to Map Friction in Your Own Journey

Journey maps built from process documentation describe the intended path. Friction lives in the gap between that and the real one, so patient journey automation should start from observation rather than from a diagram. A few practical steps close the gap.

‍Follow real patients, not diagrams. Take twenty recent patients across different entry routes and reconstruct every contact from first attempt to resolution, including the ones they initiated. Count contacts, elapsed days, and how many times they supplied the same information.

‍Count patient-initiated contacts specifically. This is the clearest single measure of friction. A patient chasing you is doing your work. Track it as a rate.

‍Look hardest at the handoffs. Front desk to clinical, clinical to billing, site to site, referral in and referral out. Friction concentrates where accountability changes hands.

‍Ask which cost each friction point creates. A confusing benefits explanation is a learning cost and needs clearer communication. A duplicate form is a compliance cost and needs integration. Silence after a request is a psychological cost and needs proactive updates. The remedy follows from the classification.

‍Instrument the resolving phase. Since it is the least measured and the most damaging, start by counting billing disputes, records requests, and results inquiries. Our guide to front desk KPIs is a reasonable baseline for the rest.

‍Check where patients drop out entirely. The people who gave up are absent from your data by definition, which makes them easy to ignore and expensive to lose.

Here's How Confido Health Can Help

This article mapped friction across three phases and three kinds of cost, and argued that most automation only addresses one of them. Confido Health's AI Agents are built to work across the whole journey rather than a single step, handling inbound requests and proactive outbound communication so patients do less work and know more about what is happening.

Here is what Confido Health delivers:

• Coverage across all three phases, from the first call a prospective patient makes, through scheduling, intake, eligibility, and prior authorization, to post-visit follow-up, refills, billing questions, recalls, and reactivation
• Proactive outbound communication, including confirmations, reminders, pre-visit instructions, and status updates, which is what reduces the learning and psychological costs that inbound-only tools leave untouched
• Integration-first approach with 40+ EHR and PMS systems including Epic, Athenahealth, and eClinicalWorks, so information a patient already provided is never requested twice
• Empathetic, natural conversations with 97 percent patient satisfaction, in more than 20 languages, with clinical triage, consent, and hardship conversations routed to your team with full context
• Proven ROI, with up to 70 percent reduction in staff call burden, 60 percent reduction in cancellations, 80 percent reduction in manual administrative work, 75 percent faster prior authorization processing, and a 15 to 20 percent increase in revenue collections
• Live in under 30 days using expert-approved templates, so you can address a full phase of the journey rather than a single step

Confido Health is more than a tool. It is the layer that carries a patient through the whole journey without asking them to carry it themselves.

Want to see how many contacts your patients currently need to get one thing done? Let's get started today.

Still in research mode? Start with our explainer on what an AI voice agent is, then read about reducing patient wait times.

Frequently Asked Questions

What is administrative friction in healthcare?

Administrative friction is the nonclinical work patients and staff must do to make care happen: understanding coverage, making calls, completing forms, chasing results, and disputing bills. Researchers classify it as learning, compliance, and psychological costs, and it measurably delays and prevents care.

What is patient journey automation?

Patient journey automation is the use of AI Agents to handle administrative work across a patient's entire path, from first contact through scheduling, intake, the visit, and post-visit resolution. It differs from single-step automation by treating the whole journey as the unit of measurement.

How much care do patients skip because of administrative burden?

Research summarized in a 2025 Health Affairs Scholar review found that nearly a quarter of insured US adults report delaying or forgoing care due to administrative tasks. The authors characterize the effect as comparable in magnitude to financial barriers, which receive far more attention.

What are the three phases of the patient journey?

Health services researchers describe seeking care, receiving care, and resolving care issues afterward. Seeking care is dominated by learning costs, receiving care by compliance costs, and resolving issues by all three at once. Evidence and operational measurement are both weakest for the third phase.

Why do small frictions matter so much?

They multiply rather than add. An eight-step journey where every step succeeds 90% of the time delivers around forty-three percent end-to-end. Each department reports acceptable performance while fewer than half of patients complete the path without needing rescue.

Which parts of the patient journey should not be automated?

Clinical triage, informed consent, financial hardship conversations, and identity verification. These frictions protect patients rather than processes. The useful test is whether a step exists to safeguard the person or simply to satisfy an internal procedure, and only the latter should be removed.

Does automation improve patient experience on its own?

Not reliably. Most automation reduces compliance costs while leaving learning and psychological costs untouched. Completing tasks silently can even raise anxiety, since patients get less evidence that anything happened. Proactive confirmation and status updates are what convert completed tasks into reduced burden.

How do you measure administrative friction?

Track patient-initiated contacts per resolved request, elapsed days from first attempt to resolution, how often patients supply the same information twice, and volumes of billing disputes and records requests. Baseline these before any patient journey automation work, since the two rarely match.

Where does friction concentrate in a healthcare organization?

At handoffs rather than within departments. Front desk to clinical, clinical to billing, and site to site are where accountability changes hands and requests stall. Each team can report good performance while the transitions between them quietly generate most of the friction.

Can AI address learning and psychological costs, not just compliance?

Yes, through proactive communication. Explaining coverage in plain language, confirming what was done, stating what happens next and when, and providing unprompted status updates all reduce uncertainty. Patient journey automation that only completes tasks silently addresses one cost of three.