100%
Worked
- Of the rural hospital book worked at least once, August 2026, platform in place
Home / Case studies / Work Genie
Case study · Platform rules, AR follow-up
Was the claim worked, or just tagged? The audit that made a touch a submitted record
Short answer: in June 2026 we audited our own first month of AR follow-up on a health center book, tracked in spreadsheets. 2,747 touches collected $736K; the same touches could have reached $1.56M. The findings became the rules Work Genie enforces for every client. Load-test figures are synthetic.
Capture efficiency · 47%
1,059 touches · lines under $100
$138,668 · parked, no owner
A touch = one submitted record
Edition
Case study · Oct 2026
Archetype
Health center book
Metric
Capture efficiency
Horizon
June 2026
Framework
Touch = submitted record
Length
5 min read
47%
Capture efficiency: $736K collected against $1.56M reachable with the same 2,747 touches, June 2026
ASP-RCM engagement records1,059touches
Spent on lines under $100 while only 29 of 124 lines of $1,000 or more were touched
ASP-RCM engagement records$138,668parked
In internal hand-offs between coding, billing and posting with no owner
ASP-RCM engagement records24hours
QC turnaround every submitted record now gets in Work Genie, across 42 QC users in 22 organizations
ASP-RCM engagement recordsSame 2,747 touches: collected versus reachable
June 2026, health center book, spreadsheet era
Diagnostic of our own first month, before Work Genie enforced targeting. Not a client result.
$1.56Mreachable, 100%Reachablesame touches, re-ranked by collectible balance
$736K47% capturedCollectedas worked
Reachable with the same touchesActually collected
Where the touches went
Every touch was logged; the problem was targeting and ownership.
Touches on lines under $100of 2,747
1,05938.6% of all touches
Lines of $1,000 or more touchedof 124 such lines
2923.4%
"Adjusted" tags that truly cleared ARwhat held up
99.7%99.7%
Targeting failureHeld up
Takeaway
A spreadsheet can prove a claim was tagged. It cannot prove it was worked, or that the next touch went to the dollar that mattered. Our own numbers showed both failures.
Two books, two eras of tracking
4.3
Touches
- Touches per claim, average
$490K
Surfaced
- Older than 60 days and untouched; the number a spreadsheet never shows
| Measure | Health center, spreadsheet era (June 2026) | Hospital, Work Genie (August 2026) |
|---|---|---|
| Touches on sub-$100 lines | 1,059 | Blocked by targeting rules |
| Parked in hand-offs, no owner | $138,668 | Hand-off requires a named next owner |
| Untouched over 60 days | Not measurable | $490K, surfaced and queued |
| Touches per claim | Not measurable | 4.3 |
Diagnostic on the left, achieved on the right; the books differ in size and payer mix, so this compares what each method could see, not performance.
A touch is a submitted record: how a claim moves through Work Genie
StepWhat the platform enforcesWhat it stops
01
Submit or it did not happenAll claim work goes through the workflow on the server: the claim, the action, the payer response, the next owner
StopsA note typed elsewhere counting as a touch
02
Twelve checks per recordEach submitted record is audited for claim, action, payer response, follow-up date and next owner
StopsTags with nothing behind them
03
QC inside 24 hoursSampled records go to one of 42 QC users across 22 organizations; "completed" means a second person agreed
StopsThe caller asserting quality alone
04
Hand-offs carry an ownerA record cannot be parked between coding, billing and posting
StopsThe $138,668 problem; it is now a validation error
05
Every CARC has a categoryCoverage went from 27 codes to all 297 CARCs
StopsDenials landing in "other"
About 86,000 rows in 1.67 seconds, and 297 of 297 CARCs categorized
Reference masters and denial-category coverage
Achieved, ASP-RCM engagement records.
ICD-10 codesof about 86,000 rows
74,71986.9% of rows
HCPCS codes
8,62310.0%
RARC codes
1,1371.3%
CARC codes
2970.3%
CARCs with a denial category, beforeof 297
279.1% coverage
CARCs with a denial category, afterof 297
297100% coverage
Rows loaded, share of 86KCoverage beforeCoverage after
Footprint is real; load figures are synthetic
| Measure | Value | Label |
|---|---|---|
| Client organizations live, reporting on, 2 servers | About 29 to 30 | Achieved |
| Onboarding wave, 29 to 30 September | 19 orgs / 6,966 claims to 24 orgs / 18,463 claims in a day | Achieved |
| QC users | 42 across 22 organizations | Achieved |
| Prior-authorization form library | 952 rows, about 302 payers; 582 + 144 official PDFs; 98 payers portal-only | Achieved 4 of a team’s 5 in-house forms were out of date |
| Release gate | 704 of 704 pages across 11 roles; 881 of 881 checks; about 1,412 wrongly closed claims restored | Internal |
100-user load test, synthetic data
30.4 claims per second; at 90,000 claims every endpoint answered under 60 ms. Generated data: what the platform can carry, not what a client collected.
Claims worked by 100 simulated callers
45,900100%
QC verdicts by 25 simulated auditors
36,70080% of claims worked
Failures
00%
Synthetic loadFailures
What the platform enforces, not yet a client result it alone produced
- ToolNo tool-attributed before and afterThe health center and hospital gains on this site were delivered by the service team; the platform share is not isolated.
- LoadLoad figures are synthetic45,900 claims at 30.4 a second were generated data. Live latency after the latest release has not been read back from production.
- 47%Capture efficiency after the rulesThe 47 percent is from before enforcement. The audit has not been re-run on a book worked entirely inside Work Genie.
- 1,412The restored 1,412Claims reopened by the integrity sweep are back in the worklist; what they collect is not yet known.
Frequently asked questions
What does "a touch is a submitted record" mean?
A claim counts as worked only when a caller submits a record through the Work Genie workflow, on the server, with the claim, the action, the payer response and the next owner. Notes kept anywhere else are not touches, and every record is subject to a 12-check audit and 24-hour QC.
What was capture efficiency?
In our own first month on a health center book in June 2026, 2,747 touches collected $736K; the same number of touches aimed at the most collectible balances could have reached $1.56M. That ratio, 47 percent, is the capture efficiency figure on this page.
Are the load-test numbers real client results?
No. The 100-user test, 45,900 claims at 30.4 a second with 0 failures, ran on synthetic data. It shows what the platform can carry. Production figures on this page are limited to the footprint: about 29 to 30 organizations on 2 servers.
Did Work Genie produce the hospital and health center results?
The service team did, using the platform. We have not isolated the contribution of the software, so this page describes the rules it enforces and labels every figure as diagnostic, achieved or test.
Checked October 25, 2026. Engagement figures are from ASP-RCM records for the periods stated; rules change, confirm against the source before relying on them.
TechnologyAR workflow engine: triage by recoverable dollars
ServiceRevenue cycle management services
ToolFree denial audit
Case studyDenial yield prioritization: work the money, not the oldest
Case studyWorking the timely-filing clock on aged AR
Case studyHospital: submissions up 94%, AR held flat
Case studyHealth center four-month AR turnaround
Audit your own touches before you buy a worklist
Send us a month of follow-up notes and the AR file behind them. We will compute capture efficiency, touches under $100 and dollars parked in hand-offs, the way we audited ourselves.