Patient Journey Automation: Map the Pipeline From Call to Payment

A patient's visit does not end when they leave the exam room. It ends when the claim gets paid, and between those two points sit at least 5 handoffs: intake, insurance verification, visit documentation, coding and claim submission, and finally payment or collections. Each handoff is a place a single wrong field, a skipped check, or a slow follow-up can turn a completed visit into unpaid revenue. Experian Health's 2025 State of Claims survey found 41% of providers now report claim denial rates above 10%, up from 30% just 3 years earlier, and that trend describes what happens when the handoffs break, not when the visit itself goes wrong.

Most practice-automation content treats those handoffs as separate problems, because separate vendors sell separate tools for each one. An eligibility platform sells the insurance check. A claims-scrubbing tool sells a cleaner submission. A collections service sells the follow-up call. Nobody selling one piece of the pipeline has a reason to show you the whole pipe, and a practice that fixes one stage while ignoring the rest still loses money, just 2 or 3 stages later than before.

This post walks the entire pipeline in order, from the first call to the final payment, prices what breaks at each stage using published numbers, and shows exactly where one stage's mistake becomes the next stage's cost. The front-desk slice, phones, intake forms, and scheduling, is only the first of 5 stages. This is the rest of the pipe, all the way to the money.

Why the patient revenue pipeline is one connected system, not five separate problems

A patient's journey runs through 5 connected stages: intake, insurance verification, the visit itself, coding and claim submission, and payment or collections. Each stage hands its output to the next, so a mistake in stage 1 does not stay in stage 1. It resurfaces as a denial, a delay, or a write-off 2 or 3 stages later.

Five connected stages carry a patient from first call to paid claim, in order.

5 stages carry a patient from first contact to a paid claim. Intake collects who the patient is and what insurance they carry. Eligibility and verification confirm that coverage is active before the visit happens. The visit itself generates the clinical documentation everything downstream depends on. Coding and claim submission turn that documentation into a billable claim. Payment and collections close the loop, whether that is an insurance remittance or a patient balance.

Treat any one of those stages as an island and you fix it in isolation while the other 4 keep leaking. The front-office half of this pipeline already maps the first 2 stages, phones, intake, and scheduling, in real depth. This post picks up from there and walks the rest of the pipe through to the money: documentation, coding, claim submission, and collections.

What breaks at each stage, and why the damage shows up two stages later

Most pipeline failures are invisible where they happen. An incomplete intake field does not bounce a claim on the spot, it bounces 3 weeks later when the payer rejects a mismatched member ID. A documentation gap does not cost money at the visit, it costs money when a coder cannot support the billed service.

The pattern repeats across the pipeline. Something breaks quietly, work continues as if nothing happened, and the cost shows up as a rejected claim, a stalled payment, or a write-off somewhere downstream.

Automating insurance eligibility verification covers the second row of that table in real depth, including the real-time check that catches most of it before the visit even starts. The rest of this post covers the other 4 rows.

Where claims and revenue actually leak

Not every claim makes it through clean, and not every denial gets fixed. Industry estimates put first-pass claim acceptance at roughly 83% to 87% industry-wide, which means 13% to 17% of claims need rework or an appeal before they get paid. A widely cited estimate suggests a large share of those denials are never resubmitted at all.

Most claims clear on the first pass. Many of the ones that do not are never resubmitted.

Model the math and the size of the permanent leak becomes clear. If roughly 15% of claims are denied and about 65% of those denials are never resubmitted, that works out to close to 10% of all claims a practice submits that never get paid at all, not delayed, not appealed, just gone. That figure is modeled from 2 directional ranges, so treat it as a reason to measure your own resubmission rate, not as your number.

Prior authorization is a growing share of the problem. Industry data suggests prior-auth denials now account for 34% of all first-pass denials, up from 22% in 2023, which means a check that used to catch mostly eligibility errors now has to catch authorization gaps too.

The real cost of a leaking pipeline, in numbers

Four numbers size the leak without any vendor spin. Providers report rising denial rates, practices commonly carry more days in accounts receivable than the target, reworking one denied claim costs real staff time, and a measurable share of net revenue gets written off before it ever reaches the bank. None of these figures is a Cart Gaze result.

Four sourced numbers on denials, AR days, rework cost, and write-offs. None is your number yet.

Every range above comes from a different source and a different slice of the industry, so treat each one as a benchmark to compare your own numbers against, not a universal outcome. The accounts-receivable and write-off figures in particular skew toward larger, hospital-weighted datasets. A single-provider practice's actual dollars will look different, even if the percentages rhyme.

What the 4 numbers agree on is the shape of the problem: it is spread across the pipeline, not concentrated in one stage, which is exactly why a single point tool rarely closes the whole gap by itself.

Where an AI receptionist and intake automation fit, and where they stop

Automation earns its place on the mechanical parts of the pipeline: answering calls, collecting intake data, running an eligibility check, and flagging anything that looks incomplete. It should never make a clinical judgment, decide medical necessity, or resolve a coordination-of-benefits question. The safe rule is the same one that governs the phones: recognition and redirection, not diagnosis.

If you have never evaluated one of these systems, start with what an AI receptionist does and where it stops before you wire anything to your schedule or your intake forms. The same boundary that keeps a phone system safe keeps an intake and eligibility system safe: it gathers and checks, and a person decides anything ambiguous.

That boundary matters just as much once patient data starts moving through a clearinghouse. Any vendor in this pipeline that touches protected health information, the eligibility platform, the coding tool, the collections texting service, is a business associate under HIPAA and needs a signed agreement before it sees a single record. Compliance is a configuration and a contract, not a badge, and no page that calls itself HIPAA certified is telling you the truth, because no such certification exists.

Automating coding accuracy and claim submission before the denial happens

Claim scrubbing, prior-authorization checks, and documentation prompts catch the mechanical causes of a denial before a claim ever reaches the payer: an inactive plan, a mismatched code, a missing authorization, a detail a coder needs that never made it into the note. None of that guarantees payment. It just removes the errors a computer can actually see.

A clean claim still depends on a human judgment call underneath it: whether a service was medically necessary, whether 2 payers coordinate correctly, whether an unusual code combination is actually right. Automation should flag anything outside its confidence, not auto-submit around it.

When a denial does happen, the appeal itself is its own specialized job: reading the remark codes correctly, building the right documentation packet, and tracking the clock on a timely-filing deadline. Automating claim denial triage and appeals walks through that stage specifically, including which denials are worth appealing and which are not.

Closing the loop: AR follow-up and patient collections automation

Once a claim pays or a balance moves to the patient, speed decides whether that money gets collected or written off. A worklist that flags an unpaid claim at 30 days behaves completely differently than one nobody checks until 90 days have passed. That single habit separates a healthy accounts-receivable number from a written-off one.

Automated AR follow-up does the same job a good biller does every morning, just without skipping the unglamorous accounts: check what is unpaid, flag what has aged past a threshold, and start the next contact, whether that is a call to the payer or a text to the patient. Automating AR follow-up and patient collections covers the worklist logic and the patient-facing side of this stage in detail.

Patient balances need their own care. A text or a portal reminder recovers more than a monthly paper statement, but tone matters as much as timing when the message is about money owed for a medical visit. The goal is a clear, respectful nudge, not a collections script.

Map your own pipeline, then start where your numbers say to

The choice of what to automate first comes down to your own numbers, not a vendor's demo. Incomplete intake or eligibility data points upstream. Frequent coding or documentation denials point to the claims stage. Accounts receivable aging past your norm points to collections. Measuring your own pipeline for 30 days beats guessing from someone else's benchmark.

Pull a month of claims data before you buy anything: your first-pass acceptance rate, your top 3 denial reasons, and your actual days in accounts receivable. That single pull tells you which stage of the pipeline is costing you the most right now.

Route by which stage is actually failing, not by which tool has the best demo.

Walk the flow once. Incomplete data upstream gets fixed at intake and eligibility. Frequent coding or documentation denials get fixed at the claims stage. Accounts receivable aging past your norm gets fixed at collections. If none of those describes your practice, your pipeline is likely healthy, and the highest-value move is incremental tuning, not a new tool.

Whatever you pick, pilot it on one stage, measure it against your own claims and AR numbers for a real stretch of time, and only then move to the next stage. If you would rather have your own leak sized first, the free Growth Leak Audit does that from your own numbers before anyone talks tools.