Extract data from W-4 forms
Unstract reads each Form W-4, older allowance-era W-4s, state withholding certificates and direct deposit forms, scanned or e-signed, and sends the employee's elections and bank details to your HRIS or payroll system.
spouse ticked 2(c) on her job too, per MD
Form W-4 (2026) · give to your employer, keep on file
{ "tax_year": "2026", "employee": "Marisol A. Delacroix", "ssn": "***-**-3906", "filing_status": "Married filing jointly", "step2_two_jobs": "checked", "step3_dependents_total": 4900, "step4a_other_income": 1800, "step4b_deductions": 6500, "step4c_extra_withholding": 40, "first_date_of_employment": "11/09/2026"}
Trusted by HR and payroll teams around the globe
New hires fill in a W-4 on whatever is in front of them
Some W-4s come through the onboarding portal as clean PDFs. The rest arrive as photos, scans and packets, often with the deposit form stapled behind.
Phone photos taken in a break room on the first day
Allowance-era W-4s still sit in the files of long-tenured staff
Hand-filled forms with an amount crossed out and rewritten
W-4, state form, deposit form and voided check in a single PDF
Mid-year changes after a marriage, a baby or a second job
Payroll staff key filing status, dependents and bank numbers from each form by hand, and one transposed digit means a wrong first paycheck or a deposit sent to the wrong account.
Unstract reads each W-4 and deposit form, checks the amounts and the bank details, and sends the employee's elections to your HRIS or payroll system.
Read the W-4, return the elections, set up the new hire
LLMWhisperer reads the form, Document Studio returns the employee's elections as fields, and Flow runs onboarding around them. A payroll team can start with W-4s alone and add the other products later.
Read W-4 forms with each step, checkbox and dollar amount kept apart
A W-4 packs five steps onto one page: three filing status boxes, a single checkbox in Step 2, two multiplied lines in Step 3 and three dollar amounts in Step 4. Flattened to plain text, the Step 4(a) income slides next to the Step 3 total. LLMWhisperer keeps every line in its step, shows which boxes are ticked and which are empty, and reads handwritten amounts and dates on scans and phone photos.
Pass the text to your own LLM, or make it the first step of a full onboarding run.
Learn more about LLMWhispererTurn every W-4 form into one record of the employee's elections
Name the fields once: filing status, the Step 2 checkbox, the Step 3 lines and total, the three Step 4 amounts, the signature date and the employer section. W-4 form data extraction then returns the same JSON for a current W-4, an allowance-era form kept on file and a state certificate. The checks are yours to set: Step 3 lines add up to the total, Step 5 is signed, the SSN matches the offer letter and the routing number on the deposit form matches the voided check. Unstract records what the employee elected; it does not compute withholding or give tax advice.
Deploy it as an API behind your onboarding portal, or as an ETL pipeline over a folder of scanned W-4s from past years.
Learn more about Document StudioRun employee onboarding from the signed W-4 to the first paycheck
Describe how your team handles W-4s and deposit forms in plain language, and Flow's agents build the onboarding workflow from it. You approve the spec, and it runs the same way for every hire. The W-4, state form, direct deposit form and the rest of the packet come in from the onboarding portal or the HR inbox, are read and compared, and the checked fields go to your HRIS or payroll system, such as Workday or ADP, with benefits elections passed to benefits administration.
A missing W-4, a name that differs from the offer letter, deposit details that disagree with the voided check or a start date at risk stops for an HR coordinator. The employee's elections pass through as written; withholding is calculated by payroll, never by Unstract.
Learn more about FlowRead your W-4s today. Plan onboarding with us next.
Start for free and pull the elections from a folder of W-4s. A full onboarding run tied to your payroll system begins with a consultation.
Form W-4, state certificates and direct deposit forms
Documents
- Form W-4 Employee's withholding certificate
- W-4s from 2019 and earlier, with allowances
- State withholding certificates
- Direct deposit authorization forms
- Voided checks and bank letters
- Mid-year W-4 changes from current employees
- E-signed W-4s from onboarding portals
- Scanned and photographed paper W-4s
- Form I-9 Employment eligibility verification
- Offer letters
Fields captured
For the teams that set up a new hire's pay
Every W-4 in the new-hire packet checked for a signature and a complete Step 1 before the start date, with gaps sent back to the employee.
Filing status, Step 3 and Step 4 amounts and deposit splits keyed into payroll by the extraction, and mid-year W-4 changes applied from the form itself.
W-4s and state certificates for every client company read the same way, each record filed under the right client and its EIN.
Placed workers' W-4s and deposit forms read on the day they sign, so a temp starting Monday is paid on the first payroll.
Fewer first paychecks to reissue, fewer forms keyed by hand
SSNs and bank account numbers stay protected
A W-4 carries an employee's SSN and home address. The direct deposit form clipped behind it carries the routing and account number that payroll pays into. Paycheck diversion fraud starts with exactly that pair: a forged deposit change sends wages to an account the employee never opened.
- W-4s, deposit forms and voided checks are encrypted in transit and at rest.
- No employee's form is used to train any model.
- SSNs and account numbers can come back masked, and access is set by role, client company and location.
- You set how long forms are kept, and they are deleted when that period ends for each employee.
- Payroll providers that keep employee bank data on their own servers can run Unstract self-hosted.
What payroll and HR teams ask about W-4s
Yes. Unstract reads federal W-4s, state withholding certificates and the direct deposit forms that come with them, and returns each as a structured record. Run it on every new-hire packet as it arrives, or over a folder of past forms during a payroll system change.
One JSON record per form: name, masked SSN, address, filing status, the Step 2(c) checkbox, each Step 3 line and the total, the three Step 4 amounts, the signature date, and the employer name, EIN and first date of employment. An older W-4 returns its allowances and any additional amount instead of Steps 2 to 4.
Call the Document Studio API from your onboarding portal with each W-4, or point an ETL pipeline at the inbox or folder where packets land. Either way you get the same fields, ready for your HRIS or payroll system to import.
Yes. Direct deposit form OCR reads the bank name, account type, routing and account numbers and any split between accounts by amount or percent. When a voided check is attached, its numbers are read too and compared with the form, and a mismatch is flagged before payroll runs.
No. Unstract records what the employee wrote and ticked on the form, and checks that the lines add up and the form is signed. Your payroll system calculates withholding, and questions about what to claim stay between the employee and their tax adviser.
Yes. Forms from 2019 and earlier return marital status, the number of allowances, any additional amount withheld and an exempt claim, kept in their own fields so they never mix with the newer Step 3 and Step 4 amounts.
Yes. State forms differ in layout and in what they ask for, so you list the fields you need per state form, such as allowances, exemptions or an extra amount, and each comes back tagged with its state.
The record comes back flagged with the reason: no Step 5 signature, Step 3 lines that differ from the total, or a box that could not be read. A blank or unreadable value stays empty rather than guessed, with the spot on the page it came from.
Yes. Each new W-4 is matched to the employee by name and masked SSN, and the record shows which elections changed from the form on file, so payroll updates only what the employee changed.
Unstract is SOC 2, ISO 27001, GDPR and HIPAA compliant. Forms are encrypted, never used for training and deleted on your schedule, and self-hosting is available. The Trust Center sets out each control for your security team.
Bring a stack of new-hire W-4s and deposit forms
A current W-4, an allowance-era form from the file, a state certificate and a deposit form with its voided check. We read them in the session and show the record each one returns, flags included.