The Medical Billing Process, Step by Step

On the timings in this table. The per-step durations are operating targets, not measured industry figures — no free primary source publishes step-level cycle times for physician practices. Treat them as a starting shape and replace each with your own measured value. The one step with a published, enforceable timing is payer adjudication, and only for Medicare; see step 10.

End-to-end cycle time is two clocks, not one. The insurance clock runs from date of service to the payer's remittance. The patient clock runs from that remittance to a collected balance or a documented write-off. They are measured separately because different people work them with different levers, and a single blended "days to cash" figure hides which of the two is actually the problem.

Failure at any stage cascades through the rest. The process can only be as fast and accurate as its slowest, weakest stage — and the Owner column above is the one most often left blank in practice. The last section of this page covers what to ask for at each handoff when billing is outsourced.

When a patient calls or books online, the practice captures: name, date of birth, insurance information, reason for visit, and contact details. This is where most downstream billing problems are seeded.

Common failure: rushed front-desk intake that captures insurance card information incorrectly, leading to claim rejections downstream.

Best practice:

• Collect insurance card images (front and back) at scheduling, policy numbers verbally
• Verify spelling of patient name vs. Insurance card exactly
• Capture both primary and secondary insurance if applicable
• Capture employer information for COB analysis
• Flag self-pay / cash-pay / sliding-scale eligibility upfront

Time: 5-10 minutes per new patient, 2-3 minutes for established patients.

Cost of failure: a demographic or insurance error captured here is not detected until the claim rejects or denies, so it gets paid for twice — once in rework labour, once in delayed cash. We publish no per-denial rework figure: the ones in circulation are drawn from hospital and mixed populations, and the most-quoted of them carries no sample, denominator or year at all. Measure your own as touches per denial × loaded staff cost.

Output: Complete patient registration record with verified demographic and insurance information.

24-48 hours before the visit, run real-time eligibility verification on the patient's insurance.

What to verify:

• Active coverage on the date of service
• Copay amount
• Deductible status (year-to-date applied vs. Remaining)
• Whether prior authorization is needed for the planned service
• Out-of-network status
• Referral requirements
• Coordination of benefits (primary vs. Secondary)

Common failure: skipping verification for established patients (assuming nothing changed since last visit). Coverage changes mid-cycle, employer changes, plan changes within payers — all create denial risk if eligibility isn't checked every visit.

Best practice: verify every visit, every patient. Use real-time eligibility (RTE) integration in PM/EHR — most modern systems support this via 270/271 EDI transactions.

Time: 1-2 minutes per patient with electronic verification. Manual verification (phone call to payer) takes 10-20 minutes — only used as backup when RTE fails.

Cost of failure: eligibility is consistently among the largest single denial categories in every published breakdown, but those breakdowns are measured on hospital claim populations and do not transfer to an outpatient practice, so we publish no share. Measure your own: tag every denial with the stage that produced it, and the eligibility share becomes a number you can manage against.

Output: Documented eligibility response captured in PM system, surfaced to front desk and provider before visit.

If the planned service requires prior authorization, submit the auth request to the payer with clinical justification.

Services typically requiring auth:

• Most surgical procedures (orthopedic, cardiac, bariatric, vascular)
• High-cost imaging (MRI, CT, PET, nuclear medicine)
• Specialty consultations (depending on plan)
• Specialty medications (biologics, infusions, specialty pharmacy)
• DME items above thresholds
• Behavioral health services beyond initial assessment
• Physical therapy beyond initial visits

Wait for approval (typically 3-10 business days) before performing the service. Some emergency or urgent services can be performed before auth and authorized retrospectively, but this carries risk.

Common failure: performing the service before auth is received, then trying to obtain retroactive auth (rarely granted by commercial payers).

Best practice: build a hard stop in scheduling — services requiring auth cannot be performed until auth is documented in the system.

Time: 30 minutes to several hours per request depending on payer documentation requirements. Specialty cases can take half a day.

Cost of failure: an authorization denial is not a rework cost, it is the whole allowed amount of the service at risk, and commercial payers rarely grant authorization retroactively. The exposure is your contracted rate on PA-required procedures times the number that go out unauthorised — a figure only your own fee schedule can produce. Worth knowing before writing one off: HHS OIG reviewed a stratified random sample of 250 prior authorization denials issued by 15 of the largest Medicare Advantage organisations in June 2019 and found that 13 percent of them met Medicare coverage rules — a Medicare Advantage finding, not a commercial one, but evidence that a PA denial is not self-evidently correct.

Output: Auth approval number with effective dates and authorized service codes, captured in PM system.

On the day of service, verify patient identity, confirm insurance hasn't changed, collect any updated insurance cards, collect copay (if applicable), and have patient sign required forms (HIPAA acknowledgment, financial responsibility, consent).

Best-practice check-in workflow:

1. Verify patient identity (photo ID match)
2. Confirm insurance — has anything changed since scheduling?
3. Re-scan insurance cards if changed
4. Collect copay — point-of-service collection is far more recoverable than later billing
5. Collect prior balance if any
6. Have patient sign required forms (HIPAA acknowledgment annually, financial responsibility, treatment consent, advance beneficiary notice if applicable)
7. Communicate expected out-of-pocket if known (deductible status from eligibility)

Common failure: skipping insurance re-verification at check-in. Patients sometimes have insurance changes between scheduling and visit, especially during open enrollment periods (October-December for commercial, year-round for Medicaid).

Time: 5-10 minutes per patient.

Cost of failure: a coverage change between scheduling and visit produces a denial that looks like a front-desk error but originated with the patient's employer or plan. It is catchable only at check-in, which is why re-verification is a separate step from step 2 rather than a repetition of it.

Output: Completed registration with current insurance info, signed forms, copay collected, patient-portal account verified.

The provider sees the patient and documents the encounter — chief complaint, history, exam findings, assessment, plan, and any procedures performed.

Documentation must support:

• Medical necessity (the diagnosis justifies the service)
• Level of E/M service — medical decision making (problems addressed, data reviewed and analysed, risk) or total time on the date of the encounter
• Specific procedures performed (with detail required for procedural CPT codes)
• Time if billing time-based codes (mental health, prolonged services, critical care)
• Modifier use (e.g., modifier 25 documentation, modifier 59 distinct procedural service)

Common failure: documentation that supports a lower level of service than was actually performed, leading to systematic under-coding. The CPT E/M revisions dropped history and exam as elements of code selection for office visits in 2021, and the 2023 code set extended the same structure to the remaining E/M categories. Providers who never recalibrated are still writing notes for a framework that no longer selects the code.

Best practice:

• EHR templates that prompt for required MDM elements
• Provider training on documentation requirements
• Periodic chart audits to identify documentation gaps
• Real-time prompts when required documentation appears missing

Time: Varies by encounter complexity — typically 5-15 minutes documentation per visit; longer for complex surgical or procedural encounters.

Cost of failure: under-coding is invisible in denial reporting, because an under-coded claim pays. It surfaces only in a chart audit, and its size is the gap between the level billed and the level the documentation supports, across the encounters that pattern affects — specific to your providers and not estimable from outside. Note that the direction runs both ways: over-coding found in the same audit is a repayment exposure, not a revenue finding.

Output: Complete clinical note ready for coding.

Coders review the documentation and assign CPT codes (procedures), ICD-10 codes (diagnoses), HCPCS codes (supplies/equipment), and modifiers.

Coding accuracy directly determines claim acceptance and reimbursement.

Common failure: coding from a 'super-bill' the provider ticked, rather than from the documentation, leading to coded services that aren't supported. This is the difference between code-first practices and documentation-driven practices.

Best practice:

• Certified coders (CPC, CCS) review documentation and assign codes
• Provider sign-off confirms coding accuracy before claim submission
• Specialty-trained coders for specialty practices (orthopedic surgical coding requires different expertise than primary care E/M coding)
• Periodic coding audits to verify accuracy

Time: 5-15 minutes per encounter for standard E/M; 15-30 minutes for complex surgical or multi-procedure cases; 1-3 minutes for follow-up visits.

Cost of failure: coding denials are the most expensive category to rework, because resolving one needs a coder rather than an A/R clerk, and a share of them are appeals rather than corrections. Track coding-attributed denials as their own line in the root-cause split instead of inside a blended denial rate.

Output: Coded claim with all CPTs, ICD-10s, modifiers, ready for charge entry.

Coded charges are entered into the practice management system. The system applies fee schedules, generates the claim with patient and payer information, and stages it for submission.

Common failure: charge lag (entering charges 3-7 days after the encounter), which delays cash flow and creates timely-filing risk.

Best practice: same-business-day charge entry.

The math of charge lag:

• 1 day lag: 1 day extra in A/R
• 3 day lag: 3 days extra in A/R, slight timely-filing risk on tight payer windows
• 5 day lag: 5 days extra in A/R, increasing timely-filing risk, documentation-recall problems
• 7+ day lag: 7+ days extra in A/R, real timely-filing risk, systematic problem

Time: 1-3 minutes per claim with efficient workflows.

Cost of failure — an illustration you can reproduce. Charge lag converts directly into days in A/R, so the arithmetic is unusually clean. Take a practice posting $1,000,000 of charges a year: average daily charges are $1,000,000 ÷ 365 = $2,740. Moving from a 6-day lag to a 2-day lag removes 4 days of A/R, or 4 × $2,740 ≈ $10,960 of working capital released — once, not annually, because this is a timing difference and not new revenue. Substitute your own charge volume. The second effect is not arithmetic at all: every day of lag is a day off the shortest filing window in your payer mix.

Output: Claim staged in PM system ready for scrubbing.

Before submission, claims pass through a scrubbing engine that checks for:

Pass 1 — Structural validation:

• Required fields present and correctly formatted
• Code formats valid (CPT, ICD-10, HCPCS)
• Modifier validity (correct set, valid combinations)
• NCCI edits (mutually exclusive code combinations)
• Place-of-service consistency

Pass 2 — Payer-specific validation:

• Payer-specific rule library
• LCD/NCD requirements for Medicare
• Plan-specific edits
• Authorization verification

Pass 3 — Contextual validation:

• Eligibility verification on file
• Prior denial pattern check
• Timely-filing window check
• Duplicate-claim prevention

Scrubbing is where a preventable error is cheapest to fix — before the claim leaves, rather than after a payer has adjudicated it. We publish no catch rate, because it depends entirely on how current the rule library is, which is the point of the next paragraph.

Common failure: relying on the PM system's basic edits without payer-specific scrubbing rules. Static scrubbing rules degrade in effectiveness over 6-12 months as payer rules evolve.

Best practice: weekly scrubbing rule updates fed by recent rejection patterns. The scrubbing engine should improve continuously rather than be rebuilt periodically.

Time: Automated, seconds per claim.

Cost of failure: a scrubber that is not maintained fails silently. It keeps passing claims, and the errors it no longer catches reappear weeks later as denials attributed to coding or eligibility rather than to the scrubber. The measurable signal is the trend in denials whose reason code matches an edit the scrubber is supposed to be enforcing.

Output: Scrubbed claim ready for submission.

Clean claims are submitted electronically to payers via clearinghouse (most common) or directly. The clearinghouse forwards claims to the appropriate payer and returns acknowledgment receipts.

Major clearinghouses:

• Availity
• Change Healthcare (Optum)
• Waystar
• Trizetto
• Office Ally
• Inovalon

Common failure: not actively monitoring clearinghouse rejection reports. Claims rejected at the clearinghouse never reach the payer — practices accumulate weeks or months of backlog without realizing it.

Best practice: daily review of clearinghouse rejection reports with same-day resolution. SLA: 1 business day to resolve clearinghouse rejections.

Time: Automated submission; rejection follow-up varies (typically 5-10 minutes per rejection).

Cost of failure: rejected claims never reached a payer, so they appear in no denial report and age silently until the filing window closes. That makes clearinghouse rejections the one failure mode that can be large and completely invisible at the same time — and it is why the rejection rate belongs on its own line rather than blended into the denial rate.

Output: Claim accepted by payer (or rejected at clearinghouse and reworked).

The payer processes the claim through its adjudication system, applying contract terms, coverage rules, medical necessity edits, and payment policies.

The result is one of:

• Paid (full or partial)
• Denied (with CARC reason code)
• Pended (additional information required)
• Adjusted (paid but at reduced rate due to contractual terms or modifier)

Adjudication time. Medicare is the one payer whose timing is published and enforceable. Under the Medicare Claims Processing Manual, Chapter 1, a HIPAA-compliant electronic claim sits behind a "payment floor" and cannot be paid earlier than the 14th day after the date of receipt (paper claims received from 1 January 2006 onward: the 29th day), while a clean claim must be paid or denied within the 30-day "payment ceiling" or interest is owed. Source: CMS, Medicare Claims Processing Manual, Chapter 1, §§80.2.1.1 and 80.2.1.2, retrieved 17 September 2026.

Commercial and Medicaid timing is set by contract and by state prompt-pay law rather than by any national standard, so there is no figure we can responsibly publish for them. Pull your own median days from submission to remittance per payer — that is the number a contract conversation actually needs.

Output: ERA (electronic remittance advice / 835 transaction) with adjudication details for each claim line, returned to the practice via clearinghouse.

This is the only stage outside the practice's control. Every other stage is internal operational discipline.

ERAs are reviewed daily. Each claim line falls into one of four categories:

1. Paid as expected. Posted to the patient's account. No further action.

2. Paid less than contracted. Investigated as potential underpayment. Often appealed.

3. Denied. Categorized by root cause and routed to:

• Correct-and-resubmit (if fixable)
• Appeal (if denial is incorrect)
• Write-off-with-documentation (if denial is correct and not appealable)

4. Pended. Information request answered with required documentation.

Common failure: weekly or monthly ERA review, allowing denials to age into appeal-window problems.

Best practice: daily ERA review with same-week denial action. SLA: 5 business days to work each denial; 1-2 days for high-priority (high-dollar, late in appeal window) cases.

Time: 1-3 minutes per claim for standard payment posting; 15-45 minutes per worked denial depending on complexity.

Cost of failure: the binding constraint on an aged denial is not probability, it is the calendar. Every payer sets an appeal deadline, and a denial worked after it is unrecoverable regardless of merit. We publish no recovery-by-age curve: none exists in any free primary source, and the versions in circulation are invented. What you can measure is the share of your denials worked inside their own appeal window — a process metric you control, rather than an outcome you forecast.

Output: Posted payments, worked denials, identified underpayments routed for appeal.

After insurance adjudication, any patient responsibility (copays not collected upfront, deductibles, coinsurance, non-covered services) is billed to the patient.

Best-practice patient billing workflow:

1. Statement within 7 days of payer payment
2. Online payment portal with mobile support and Apple Pay / Google Pay
3. Reminder statements at 30 and 60 days (varying intensity)
4. Soft-touch reminder calls at 60-90 days
5. Payment plans offered for balances over $100
6. Bad debt threshold at 120 days with documented decision: third-party collection or write-off

Common failure: passive patient billing — sending statements but not following up with calls or offering payment plans.

Best practice: structured patient billing workflow with online payment portal, payment plans for balances over $100, personal follow-up before bad debt referral.

Time: ongoing throughout the patient A/R cycle. We publish no share-of-effort figure for it; measure your own, because it scales with your deductible exposure and self-pay mix rather than with claim volume.

Cost of failure: patient responsibility is the only receivable with no contract behind it and no appeal process in front of it. Once a balance ages past the point where the patient still associates it with the visit, it gets materially harder to collect. We publish no recovery percentages for structured versus unstructured patient billing, because no free primary source measures it — track your own self-pay collection rate by statement cycle, which reflects your patient mix and deductible exposure.

Output: Collected patient revenue or documented write-off / collection referral.

If you're improving the medical billing process and can only fix a few things, this is the order we would work them. It is an editorial judgement about effort against effect, not a measured ranking — no free primary source publishes one — so check it against where your own denials and A/R actually sit before committing to it.

1. Eligibility verification (Step 2). Lowest implementation cost of the five, and it removes a cause rather than reworking its output. Effect shows in the next submission cycle, because it changes behaviour at the point of the transaction.

2. Charge lag reduction (Step 7). Workflow discipline, no system purchase, and it converts directly into days in A/R — the arithmetic is worked in step 7 above. A one-time working-capital release rather than new revenue.

3. Daily ERA review and denial work (Step 11). Compresses denial-to-action time from weeks to days, which is what keeps denials inside their appeal windows. Medium cost, because it needs staffed capacity rather than a rule change.

4. Patient billing workflow (Step 12). Higher cost, usually a portal plus a defined cycle, but it addresses the receivable with no contract and no appeal behind it.

5. Documentation training for providers (Step 5). The highest-variance item: the size of the opportunity depends entirely on what a chart audit finds, and the gap runs in both directions — an audit that surfaces over-coding has found a repayment exposure, not a revenue opportunity. Slowest to take effect, because it depends on clinician behaviour.

We publish no combined uplift figure for working all five. Any such number is a forecast of recovery, and recovery depends on payer response, appeal outcome and filing runway.

Lower-priority fixes (still valuable but lower impact-per-effort):

• Step 3 (prior auth tracking) — important for procedure-heavy practices
• Step 8 (scrubbing rule maintenance) — important for high-volume practices
• Step 6 (coding audit and training) — important when audit data shows specific gaps

The order matters. Fixing patient billing before fixing eligibility leaves the underlying revenue leak. Fix the largest leaks first — and "largest" means largest in your own root-cause split, not in this list.

Knowing the twelve steps is one thing; knowing who is accountable for each is what determines whether the process actually runs. When billing is outsourced, the useful question stops being "what happens at this step" and becomes "who owns it, how often does it run, and what do I receive as evidence it ran."

A step with no named owner and no artifact is a promise rather than a process. Three specific handoffs are where outsourced arrangements most often fail:

1. Clearinghouse rejections. These never reach the payer, so they do not appear in denial reporting and can age invisibly until timely filing expires. Ask what the turnaround commitment is and what report shows it.
2. Payment variance. A claim can pay, post and close while paying less than the contract requires. Detecting it is a payment posting function, and it only happens if expected reimbursement is loaded and current.
3. Escalation. If escalation triggers are not defined in advance, escalation becomes a judgement call that gets deferred.

We publish our own stage-by-stage ownership, cadence and deliverable for all ten operating stages, plus the escalation triggers, in the operating framework on our about page — along with worked specimens of the reports themselves.