On this page
- Why do specialty groups need a baseline first?
- Which revenue leakage metrics should you track?
- How should you read referral and scheduling metrics together?
- What do procedure conversion and order aging reveal?
- How should recovered value be reported?
- When are location and provider comparisons fair?
- What data-quality caveats should every report carry?
- Who should own the metrics?
Key takeaways
- Measure leakage as conversion between steps of care, not as total volume.
- Build a baseline from about two years of history so each month compares with the same month a year earlier.
- Report recovered value as care completed after detection versus baseline: correlation, not proof of cause.
- Compare locations and providers only after definitions and data quality are consistent.
- Mark every rule or definition change on trend charts so it is not mistaken for progress.
The revenue leakage metrics that matter most for specialty medical groups measure conversion between steps of care: referral to visit, visit to procedure, order to completion, cancellation to rebooking, and recall to completed care, plus how long each step takes and how old the open items are. Track them monthly against a baseline, by location and provider, and report recovered value as care completed after detection, not as proven revenue.
Good revenue leakage KPIs answer three questions: what was expected to happen, what actually happened, and how big is the gap. Each metric below follows that pattern, which we explain in more depth in expected vs. actual events in healthcare.
Why do specialty groups need a baseline first?
A leakage program needs a starting point, or a good month and a definition change look the same. Use about two years of history so each month can be compared with the same month a year earlier, which removes much of the seasonal swing from holidays, deductible resets and provider vacations.
For each month and location, record volume (completed visits and procedures by service), conversion at each step, open items at month end and their estimated value. Freeze the baseline once leadership approves it. If rules, definitions or service values change later, rebaseline and keep the old version so trends before and after the change can still be compared. If you have not yet sized the problem, start with a revenue leakage assessment.
Which revenue leakage metrics should you track?
The table below covers the core set. Rates should always appear next to their counts; a rate on a small denominator moves a lot for reasons that mean little.
| Metric | Definition or formula | What it shows | Caution |
|---|---|---|---|
| Referral conversion | Referrals with a completed visit within the window / referrals received | How well inbound demand becomes care | Exclude duplicates and referrals the practice declined; see how to calculate referral leakage |
| Time to first contact | Days from referral received to first documented contact | Speed of the referral team | Depends on contact attempts being logged in structured fields |
| Time to schedule | Days from referral or order to a booked appointment | Friction in scheduling and authorization | Long waits may reflect capacity, not effort |
| Visit completion | Completed visits / scheduled visits | No-show and late-cancel burden | Rescheduled visits can be double counted |
| Procedure conversion | Completed procedures / procedure orders in the period | How many ordered procedures happen | Clinicians decide what is ordered; orders canceled for clinical reasons belong out of the denominator |
| Cancellation recovery | Canceled cases rebooked within the window / canceled cases | Whether canceled patients get a new date | Waitlist refills keep utilization high while this stays low |
| Recall conversion | Recalls completed by due date plus grace period / recalls due | Whether planned follow-up happens | Due dates must come from clinician decisions; check data quality first |
| Unresolved-order aging | Count of open orders by age band, such as 0-30, 31-90, 90+ days | Whether the backlog is growing or getting older | An order can be open because the EHR status was never updated |
| Verified completed-service value | Completed care after detection x the practice's average value | Estimated value of care completed after a patient was flagged | Estimated, not collected; correlation, not cause |
Specialty groups often add one or two service-line metrics: surveillance procedures overdue by count and age in GI, imaging orders open past the expected window in cardiology, imaging results never reviewed in orthopedics, or cataract consults with no surgery date in ophthalmology.
How should you read referral and scheduling metrics together?
Referral conversion tells you how much demand turns into care. Time to first contact and time to schedule tell you why. A site with low conversion and long time to first contact usually has a staffing or inbox problem. A site with fast contact but long time to schedule usually has a capacity, authorization or appointment-type problem. Looking at one number alone sends the fix to the wrong team.
What do procedure conversion and order aging reveal?
Procedure conversion shows how many ordered procedures are completed. Order aging shows what the open remainder looks like. Two practices can have the same conversion rate while one has a small, fresh backlog and the other has hundreds of orders more than 90 days old. The older the order, the less likely the patient completes it without help, so aging is often the better guide for where to start.
How should recovered value be reported?
Recovered value is care completed after a patient was flagged, compared with what the baseline predicts. It is useful, but it shows correlation with outreach, not proof that outreach caused the care. Some patients would have come back anyway.
Illustrative example. Suppose a group's baseline shows that, in a typical month, 80 patients with overdue recalls complete their procedure without targeted outreach. After the group starts working a ranked list, 110 complete in the same month. The gap of 30 is the recovered figure. At an assumed internal average of $900 per procedure, the estimated recovered value is 30 x $900 = $27,000. Report it as estimated value, labeled as correlation, and show it separately from collected revenue until claims and remittance data are connected.
When are location and provider comparisons fair?
Location and provider comparisons are fair only when definitions, data entry and case mix are reasonably consistent. Before ranking sites:
- Confirm every location uses the same order codes and appointment types, or map them.
- Check that recalls and contact outcomes are recorded the same way at each site.
- Show counts alongside rates, so a small site's swings are visible.
- Compare each site with its own baseline first, then with peers.
Averages hide the sites that need help. A group-wide referral conversion rate can look healthy while one location quietly loses a large share of its referrals. The same rules apply when comparing practices in a PE portfolio or across an MSO's affiliated groups, where EHRs and code sets often differ as well; MSO revenue leakage definitions covers how to align them, and revenue leakage intelligence for MSOs shows how Nuvory keeps them consistent across practices. For multi-site groups, see revenue leakage intelligence for specialty medical groups.
What data-quality caveats should every report carry?
- Status fields lie. Orders and referrals are often left open or closed incorrectly. Confirm with appointment and procedure evidence.
- External care is invisible without claims or HIE data. Some apparent leakage is care completed elsewhere; network leakage software shows where patients went when that data is connected.
- Contact is not completion. A call logged is not a visit completed.
- Definition changes look like trends. Mark every rule change on the chart.
- Estimated values are estimates. Use conservative internal averages and label them.
Who should own the metrics?
The COO or VP of operations owns the overall number. Service-line leaders own their slice. Scheduling and referral managers own the daily queue. A monthly review that walks from the summary to one location's actual work queue keeps the three levels aligned and keeps the numbers grounded in real patients.
A simple monthly pack covers it: one page of rates with counts, one page of open items by age and value, one page of changes to rules or definitions, and one location's queue as a spot check. Teams that cannot maintain the matching by hand often use healthcare revenue leakage software to keep the baseline and the monthly numbers on the same definitions. For a worked specialty example, see gastroenterology revenue leakage; other guides are in the revenue leakage category.
Leakage metrics setup checklist
- Write a one-line definition and formula for every metric before reporting it.
- Load about two years of history and freeze the approved baseline.
- Use the practice's own average reimbursement for estimated values, labeled as estimates.
- Exclude patients completed elsewhere or closed with a documented reason from denominators.
- Report by location, provider and service line, with counts next to every rate.
- Mark rule and definition changes on every trend chart and rebaseline when needed.
- Show estimated value separately from collected revenue until claims data is connected.
- Review metrics monthly with a walk from the summary to one location's work queue.
Frequently asked questions
Is recovered value the same as collected revenue?
No. Recovered value is an estimate of care completed after a patient was flagged, compared with a baseline and valued with the practice's own averages. It shows correlation with outreach, not proof of cause, and should be reported separately from collected revenue.
Which leakage metric should a specialty group start with?
Start where the volume and value are highest for your specialty. Procedure-heavy groups often begin with procedure conversion and unresolved-order aging; referral-dependent groups often begin with referral conversion and time to first contact. Add the rest once definitions are stable.
What happens to trends when a rule or definition changes?
Mark the change on every trend chart and rebaseline, keeping the old version. Otherwise a definition change can look like improvement or decline when nothing in the workflow actually changed.