How do you calculate referral leakage?

Divide referrals that went outside the preferred network by total referrals over the same period, then cut that rate by specialty, payer, and referring provider. Value it by multiplying leaked referrals by the average downstream margin, or under risk contracts by the average cost differential between preferred and non-preferred specialists.

What is a normal referral leakage rate?

Organizations that have never instrumented the point of order commonly find a third or more of referrals leaving the preferred network. Vanguard Medical Group measured 33% before deployment and 8% after. Treat any rate you have not measured by specialty as unknown rather than acceptable.

Is referral leakage always bad?

No. Some leakage is clinically correct — subspecialty capability you do not have, patient choice, geographic access, or a genuine wait-time problem. The goal is eliminating unintended leakage, not forcing every referral inside the network.

How do you reduce referral leakage?

Validate coverage and present the best in-network specialist at the point of order inside the EHR, automate prior authorization so preferred specialists are not the slower path, engage the patient directly to get the appointment booked, and confirm attendance with a returned consult note.

What does referral leakage cost a health system?

It depends entirely on specialty mix and contract type, so model it from your own data rather than an industry average. Customers who closed leakage documented $8M in first-year referral cost savings in a single market and up to a 45% reduction in referral cost.

Get a Demo

Guide

Referral Leakage: Definition, Causes, and Cost

How to measure leakage honestly, separate intended from unintended leakage, put a defensible dollar figure on it, and close it at the point of order.

Where the referral actually goesFour ways a referral leaks

01Out-of-network specialist
Selected by habit or convenience, without plan-level network status at the point of order.
02Never scheduled
The order exists, the patient was asked to call, and the call never happened.
03Lost while pending
Authorization delay stretched the wait until the patient went elsewhere or gave up.
04No consult note back
The visit happened somewhere; the referring clinician never learned the outcome.

Claims

the only data that shows where leaked referrals actually landed

Top 3

specialties usually hold most of the leaked dollars

Keepage

the paired metric to report — leakage alone hides the win

Size leakage in dollars from claims before evaluating any product to fix it.

Short answer

What is referral leakage?

Referral leakage is the share of patient referrals that leave your preferred network — going to an out-of-network or non-preferred specialist — or that never result in a completed specialist visit at all. It is measured as a percentage of total referrals and valued in forgone downstream revenue or excess cost of care.

Key takeaways

Leakage has two forms: referrals that go elsewhere, and referrals that go nowhere. Most programs only count the first.
An organization-wide leakage percentage is not actionable. Cut it by specialty, payer, and referring provider.
Separate intended leakage (capability, geography, patient choice) from unintended leakage (friction, stale directories, slow authorization).
Leakage is a point-of-order problem. Retrospective reports name it; only in-EHR steerage prevents it.
Value leakage against your own contract mix — a single industry dollar figure will not survive a CFO review.

Defining leakage precisely enough to act on

Two distinct events are usually lumped together. Outbound leakage is a referral that reaches a specialist outside the preferred network. Abandonment is a referral that never becomes an attended visit — no one scheduled it, or the patient did not show. Both represent lost care coordination; only one shows up in out-of-network claims.

A working definition therefore has to include completion. If a third of retained referrals never turn into a visit, high keepage is masking a care gap and an unbilled encounter at the same time.

How to measure referral leakage

Run four calculations over the same 12-month window:

Leakage rate. Referrals to non-preferred specialists divided by total referrals.
Abandonment rate. Referrals with no attended visit within the clinically appropriate window divided by total referrals.
Leakage concentration. The same rates cut by specialty, payer, and referring provider. Leakage is almost never evenly spread — a handful of specialty-and-payer combinations usually carry most of it.
Dollar value. Leaked volume multiplied by the relevant financial impact for each contract type (see below).

Claims data is required for an honest number. EHR referral orders tell you what was intended; claims tell you what happened. Programs that measure only from the order side systematically understate leakage.

Root causes, in order of how much they usually explain

Friction at the point of order. The in-network option takes minutes to confirm; a remembered name takes seconds.
Directory decay. Panel status, accepted plans, and subspecialty focus go stale, so staff stop trusting the list.
Access failure. The preferred specialist cannot see the patient inside the clinically appropriate window, so the coordinator goes outside.
Authorization drag. When the preferred path takes three extra days of portal work, the process routes around it.
No patient follow-through. Referrals handed to the patient to schedule abandon at a predictable rate.
Unmeasured accountability. Without per-provider reporting, no behavior changes.

Modeling what leakage costs you

Use two separate models and add them:

Fee-for-service exposure. Leaked referrals in a specialty, multiplied by the average downstream contribution margin for that specialty's typical care path — consult, imaging, procedure, facility. This is forgone revenue.

Risk-contract exposure. Leaked referrals multiplied by the cost differential between preferred and non-preferred specialists for the same episode. Under shared savings this is money you pay for care you did not route.

Add abandonment separately: the cost of the care gap, avoidable downstream acuity, and the quality-measure impact.

$8M / $9M

Year 1 / Year 2 referral cost savings

VillageMD Houston market, 2026

75%

Leakage reduction

Regional health system, 2026

97%

Keepage increase

Health system partner, 2026

Seeing leakage

What leakage looks like when it is finally measured

Most organizations cannot answer where their referrals went. Once the destination of every referral is recorded, the leak splits into four named causes — and three of them are fixable with workflow, not contracts.

Referral Leakage Control TowerRolling 6 months · 12,400 referrals

In-network rate

91%

+29pp

Leakage

-29pp

Referrals with no outcome

-31pp

Days to appointment

1.2d

-64%

In-network rate by month

62%

Feb

66%

Mar

71%

Apr

78%

May

85%

Jun

91%

Jul

Where the other referrals went

In-network specialist, appointment kept62%
Out-of-network by directory error14%
Never scheduled (patient never called)13%
Scheduled, then no-showed11%

Illustrative view: in-network rate climbing month over month, with the remaining volume attributed to directory error, referrals never scheduled, and no-shows.

Sizing the loss

The arithmetic that turns leakage into a budget line

Three inputs your finance team already has produce the number that funds the program.

Leakage Recovery ModelIllustrative — replace with your own figures

Inputs

Annual outbound referrals12,400
Share leaving the network38%
Downstream margin per retained referral$1,850

Result

Referrals leaving the network4,712
12,400 × 38%
Recoverable with in-network steering2,592
55% of leaked volume
Annual margin recovered$4.79M
2,592 × $1,850

Swap in your own referral volume, leak rate, and downstream margin — the structure of the calculation does not change.

The leakage math, on a worked example

Referrals ordered in a year40,000

Illustrative volume for a mid-size medical group. Start from your own EHR order counts.

Stayed in the preferred network (keepage)26,000 · 65%

Typical manual-process keepage before point-of-order steerage.

Left the preferred network (leakage)14,000 · 35%

The routable share — most of it chose an out-of-network specialist by default, not by preference.

Never became a completed visit (abandonment)4,000 · 10%

Counted separately from leakage: a care gap, not a routing loss.

Recoverable at a 75% leakage reduction10,500 referrals

Multiply by your specialty-level downstream margin or risk-contract cost differential.

Volumes above are illustrative arithmetic, not a customer result. Verified customer outcomes — including the 75% leakage reduction — are sourced on the facts page.

How to reduce referral leakage

Sequenced by impact per unit of effort:

Make the right answer the default. Surface the best-scoring in-network specialist at the point of order in the EHR, with coverage already validated.
Score on access, not just cost. A preferred panel that cannot see patients promptly will be bypassed regardless of policy.
Automate prior authorization. Removing days of portal work removes the main reason staff route around the preferred path.
Own the scheduling. Contact the patient directly, in their language and channel, and book the appointment rather than delegating it to them.
Close the loop and publish the numbers. Retrieve consult notes and report keepage and completion per referring provider monthly.

See how ReferralPoint implements each step, or the buyer's guide for evaluating vendors.

What good looks like

Mature programs hold keepage in the high eighties to low nineties for specialties where the network has genuine capability, keep time-to-attended-visit inside the clinical standard for urgent categories, and can explain every remaining point of leakage as an intentional clinical or access decision.

The tell for an immature program is not a high leakage rate — it is being unable to produce leakage by specialty and payer at all.

Stopping the leak