Paper sign-in sheets can be a payroll bottleneck: carbon copies, varied handwriting, changing layouts and union agreements require more than transcription alone. myCrewIQ pairs multimodal document vision with deterministic pay-rule auditing.
How we tested
The benchmark used physical event paperwork across three trade shows and one concert, with four unions represented. Handwritten cells were audited against high-resolution scans of the source sheets. Production labels are anonymized; no crew names, signatures, contact details or other crew PII are published.
Swipe to see all columns
| Production | Local | Sheets | Crew shifts | Line-item accuracy |
|---|---|---|---|---|
| Production A · Touring / Concert | Local 314 | 5 | 64 | 91.2% |
| Production B · Convention Expo | Local 1340 | 8 | 63 | 93.3% |
| Production C · Commercial Exhibit | Local 1340 | 4 | 29 | 97.9% |
| Production D · Regional Event Staffing | Local 123 | 5 | 43 | 98.9% |
| Total | Multi-local | 22 | 199 | 90%+ baseline; 98%+ post-rule |
The supplied production records list three local numbers: Local 314, Local 1340 and Local 123. Local 1340 appears on two production records. The overall post-rule figure is not a mean of the four production percentages.
Flexible layouts, traceable review
The workflow handles both three-tier union sign-in sheets and punch-time grids. Extracted entries remain reviewable against the original sheet, and uncertain handwriting or discrepancies are surfaced for human confirmation.
Read the handwriting; let fixed rules do the math
The reading step transcribes entries; it does not calculate payroll. Straight time, overtime, double time and other configured pay outcomes are calculated by deterministic rules. In these audited runs, payroll arithmetic was handled by those fixed rules rather than generated by the reading step.
Automated CBA reconciliation checks recorded hours against applicable approved rules, including shift timing, lunch deductions, day-of-week conditions, meal penalties and configured parking or travel caps. Discrepancies are surfaced before export; qualified reviewers retain responsibility for interpreting the agreement and approving payroll.
Payroll rules vary by employer and agreement. These results describe the benchmark and are not a guarantee for every production.
Frequently asked questions
- How accurate is myCrewIQ at reading handwritten sign-in sheets?
- Across four productions, baseline line-item accuracy exceeded 90%, scaling past 98% after deterministic rule validation.
- How is payroll calculated?
- No. The reading step copies what was written. Pay rules are applied afterwards by fixed rules from the agreement, and anything that looks wrong is flagged for a person.
- Is crew information published?
- No. Productions are anonymized and this case study contains no crew PII.