The card expired at midnight. Nobody found out until 8:15.

A privately held healthcare staffing firm runs roughly $165M in gross revenue with about 1,100 clinicians on assignment in any given week, across 13-week travel contracts, allied specialties, and local contracts. Two delivery hubs, 130 internal staff, and vendor status with four health-system managed service programs. Bullhorn was the applicant tracking system, Dynamics GP carried the ledger, credential expiries lived in a spreadsheet one coordinator rebuilt every Monday, and everything between was shared drives and email.

• A traveler in week three of a 13-week contract was turned away at a badge office because her BLS card had expired at midnight, though she had recertified two weeks earlier
• The reminder had fired 11 days before to a colleague who had left, and the coordinator learned of it at 8:15 from the hospital
• The 7 a.m. shift ran a nurse short and the house supervisor logged it as an agency cancellation, which nobody saw until the quarterly review 11 weeks later
• Recruiters retyped every submission into every client portal by hand, about 40 minutes each, so a recruiter with 14 openings and a four-hour window submitted nine
• Finance reconciled approved hours to placements across four portal layouts in spreadsheets, and wrote off every variance under $250

We ran a phased modernization, sequenced so the credentialing failures stopped first, while building toward one picture where credentials, submissions, and billed hours agree. Where documents are read to extract credential data, values are proposed rather than posted: a specialist confirms every field before it becomes a record.

Credential requirements as data

Roughly 260 requirement sets were captured from contracts, vendor manuals, and the coordinator's notes, then modeled at the level they actually vary: client, facility, unit, specialty, and state, with effective dating, so a change carries a start date rather than an email date.

The pre-start readiness gate

Every assignment carries a readiness state computed from those requirements, the clinician's credential record, and monthly exclusion re-screens. It evaluates forward: a credential expiring in week seven of a 13-week contract fails in week one. Alerts escalate at 60, 30, 14, and 7 days, and the gate acts before a contract starts, never after.

Submission assembly and portal delivery

Submission packets assemble from the clinician's structured record instead of Word files on a shared drive. Where a program licenses a supplier interface the submission goes out through it; where none exists, the packet renders into that program's format and a coordinator files it in one pass, with the portal ID recorded against the placement.

Hours-to-placement reconciliation

Each program's approved-hours export is restated against one set of fields, with week-ending conventions and rounding applied per program, then matched to its placement. The output is not a total but a variance register typed by cause, with a dollar amount and an owner on every line.

Pay packages and stipend compliance

The pay-package workbook became a service that prices each package against the applicable federal locality ceiling as it is quoted, and requires current tax-home documentation before non-taxable components are approved. Recruiters see the ceiling while they are still negotiating, the only moment the check is worth anything.

Analytics, expiry forecasting, and redeployment

Readiness, submission throughput, variance recovery, and scorecard position report weekly rather than quarterly. A forecasting model projects credential expiries against the assignment calendar, flagging clinicians who will fall out of readiness mid-contract and those finishing assignments who could be redeployed. Every recommendation arrives as a ranked list for a person to act on.