Drop in any pay stub PDF — ADP, Gusto, Paychex, QuickBooks Payroll, OnPay, Rippling, Justworks — and get a clean Excel file with employee name, pay period, gross, net, taxes withheld, deductions, and year-to-date totals.
ADP stubs put gross and net side-by-side, Gusto stacks them vertically, Paychex breaks deductions across two pages, and QuickBooks Payroll embeds the YTD column inside the current-period table. Generic PDF tools that assume a consistent layout fall apart on the first stub from a new provider.
Lenders verifying income for mortgage or auto loans need 2-4 stubs per applicant, often from multiple jobs at different providers. HR ops doing wage audits or comp benchmarking might pull hundreds of stubs across an employee population. Manual transcription is error-prone — and a single transposed YTD figure can mess up an underwriting decision.
PDFExcel reads pay stubs by structure. The model knows what gross pay looks like vs. net, where YTD totals live on each provider's layout, and how to handle multi-page stubs where deductions wrap to a second page. ADP, Gusto, Paychex, QuickBooks Payroll, OnPay, Rippling, Justworks, Square Payroll, and most regional providers all extract the same way.
Each stub becomes one row with employee name, pay period start/end, pay date, gross, net, federal withholding, state withholding, FICA, Medicare, and any deductions you specify (401k, health, dental, garnishments). YTD column is preserved alongside current-period values so income-verification math is instant.
The model knows which provider's stub it's looking at and adapts to each layout — no need to tell it 'this is an ADP stub' or 'this is a Gusto stub'. Trained on real stubs from every major U.S. payroll provider.
Most pay-stub tools focus on creating stubs, not extracting data from them. PDFExcel does the inverse — converts stubs you already have into clean spreadsheet data, free to start.
One row per stub with current-period and YTD values. Mix providers in the same upload — the export looks identical.
| # | Employee | Period End | Pay Date | Gross | Net | Fed Tax | 401(k) | YTD Gross |
|---|---|---|---|---|---|---|---|---|
| 1 | J. Patel | 03/15/2025 | 03/20/2025 | $3,461.54 | $2,488.21 | $432.69 | $346.15 | $22,500.01 |
| 2 | J. Patel | 03/29/2025 | 04/03/2025 | $3,461.54 | $2,488.21 | $432.69 | $346.15 | $25,961.55 |
| 3 | L. Morrison | 03/15/2025 | 03/20/2025 | $2,884.62 | $2,098.44 | $314.71 | $144.23 | $18,749.99 |
| 4 | L. Morrison | 03/29/2025 | 04/03/2025 | $2,884.62 | $2,098.44 | $314.71 | $144.23 | $21,634.61 |
| 5 | S. Wong | 03/29/2025 | 04/03/2025 | $4,615.38 | $3,124.19 | $646.15 | $461.54 | $32,307.66 |
Mortgage and auto lenders verifying income, HR ops running comp audits, payroll teams reconciling third-party reports, attorneys handling wage and divorce disputes.
Applicant submits 4 most-recent pay stubs from 2 jobs. Upload all 4, export to Excel, copy gross/net/YTD straight into the underwriting income model. Verification time drops from 20 minutes to under 5.
Pulls stubs for 200 employees to verify a payroll provider's reports against the actual stubs after a benefits change. Batch upload, export, pivot — done in 30 minutes.
Opposing party produces 18 months of stubs in discovery. Convert all to Excel to build an income exhibit showing gross, net, and discretionary deduction patterns over time.
10 documents per month, free, forever. Most lender income-verification workflows (2-4 stubs per applicant) fit in the free tier. Plans start at $69/month for 50 documents.
Yes — plus QuickBooks Payroll, OnPay, Rippling, Justworks, Square Payroll, and most regional providers. The model adapts to each layout automatically; you don't need to tell it which provider.
Yes. Each stub row gets both current-period gross/net/taxes AND the year-to-date column. Income-verification math is instant — no manual cross-referencing.
Yes. Add the specific deduction name as a custom field in step 2 and the model finds it on each stub regardless of which row it's printed on. Useful for retirement-plan audits and wage-garnishment compliance.
Yes. Files are encrypted in transit, processed in memory, and permanently deleted as soon as extraction completes. Never stored, never shared, never used to train AI. SOC 2 controls in progress.