ITR Data Extraction for Underwriting: Form-by-Form Guide
Last reviewed: November 2026
ITR data extraction for underwriting is the process of pulling income, turnover, profit and tax figures from a borrower's income tax returns (ITRs) into the lender's credit file without manual entry. When analysts retype these figures, they pick the wrong schedule, miss a revised year and carry typing errors into eligibility.
This guide is for chief credit officers (CCOs) and heads of credit operations at non-banking financial companies (NBFCs), banks and fintech lenders. It shows where each income figure sits, what manual entry costs, which checks make extracted income safe, and how Digiqt's DocuEye removes the retyping.
Key Takeaways
- The ITR form decides where income lives, so extraction rules must be form-specific.
- Underwrite on income before deductions, not taxable income. Tax-saving deductions reduce tax, not cash.
- Capture how and when each return was filed. A last-minute updated return is weaker evidence than an on-time original.
- Retyping is avoidable. Extraction tools read ITR PDFs and scans and leave analysts only the exceptions.
What Is ITR Data Extraction for Underwriting?
ITR data extraction for underwriting turns income tax returns into structured, checked data for the credit appraisal memo (CAM). It is a data-capture job, not a fraud check, which is covered in our guide to fake ITR detection for self-employed loans.
Which ITR Forms Do Lenders Receive?
Lenders receive six ITR forms in practice: ITR-1, ITR-2, ITR-3, ITR-4, ITR-5 and ITR-6. Each one gives credit a different view of income.
| ITR Form | Who Files It | What Credit Gets |
|---|---|---|
| ITR-1 (Sahaj) | Resident individuals, income up to ₹50 lakh from salary, house property, other sources | Salary, rent, interest |
| ITR-2 | Individuals and Hindu Undivided Families (HUFs) without business income | Salary, capital gains |
| ITR-3 | Individuals and HUFs with business or professional income | Profit and loss account, balance sheet |
| ITR-4 (Sugam) | Resident individuals, HUFs and firms (not limited liability partnerships, or LLPs) on presumptive taxation, income up to ₹50 lakh | Turnover, deemed income |
| ITR-5 | Partnership firms and LLPs | Firm profit, partner pay |
| ITR-6 | Companies | Company financials |
The Central Board of Direct Taxes (CBDT) notified ITR-1 and ITR-4 for assessment year (AY) 2026-27 on 30 March 2026, so extraction templates need a yearly review.
What Does Manual ITR Entry Cost a Lender?
Manual ITR entry costs analyst hours on every self-employed file, plus the risk that one wrong figure changes a credit decision. In this illustrative scenario, a lender handles 1,200 self-employed files a month, each with two years of ITR-3.
At 25 minutes to key each return, that is 1,000 analyst hours a month. Assume ITR data extraction with review of flagged fields cuts this to 8 minutes.
| Measure | Manual Entry | Extraction With Exception Review |
|---|---|---|
| Minutes per return | 25 | 8 |
| Analyst hours a month | 1,000 | 320 |
| Hours released | – | 680, about four full-time analysts |
Which Fields Should ITR Data Extraction Capture?
ITR data extraction should capture seven fields: Permanent Account Number (PAN) and name, filing section and date, tax audit flag, gross salary, net profit with depreciation, gross total income and taxes paid.
| ITR Section | Field to Extract | Credit Use |
|---|---|---|
| Part A-General | PAN, name, AY | Identity match |
| Part A-General | Section filed under, filing date | Original, belated, revised or updated return |
| Part A-General | Tax audit flag | Whether audited financials should exist |
| Salary schedule | Gross salary, employer name | Salaried income |
| Part A-Profit and Loss | Gross receipts, net profit, depreciation | Business income and cash profit |
| Part B-Total Income | Gross total income, deductions | Income before and after deductions |
| Tax schedules | Taxes paid, tax deducted at source (TDS) | Tax footprint behind the income |
Our income verification AI agent page shows how these figures can be checked against other income evidence.
How Do Lenders Calculate Income From an ITR for a Loan?
Lenders calculate income from an ITR by starting from the figure that matches the borrower type, such as gross salary, net profit or presumptive income. They then add back the non-cash charges their credit policy allows.
| Borrower Type | Usual Form | Starting Figure | Common Add-Back |
|---|---|---|---|
| Salaried | ITR-1 or ITR-2 | Gross salary | None |
| Proprietor or professional with books | ITR-3 | Net profit | Depreciation |
| Presumptive business or professional | ITR-4 | Presumptive income | None; validate with bank and Goods and Services Tax (GST) data |
| Partner in a firm | ITR-3 plus firm's ITR-5 | Partner salary and interest | Profit share, per policy |
For business borrowers, our guide to financial statement spreading for lenders shows how profit and loss figures move into a standard spread.
Should Lenders Use Taxable Income or Gross Total Income?
Lenders should use gross total income, or the salary or business figure, before Chapter VI-A deductions, not taxable income. Chapter VI-A allows deductions for items such as insurance premiums and tax-saving investments, which reduce tax but not the cash available to repay a loan.
How Should Lenders Read ITR-4 Presumptive Income?
Lenders should read ITR-4 presumptive income as a floor, not real profit. Under section 44AD a business declares at least 6% of turnover received through banking channels and 8% of the rest, and under section 44ADA a professional declares at least 50% of receipts.
Many returns declare exactly the minimum, so ITR-4 shows turnover but little about margin. According to CBDT Notification No. 45/2026 (30 March 2026), ITR-4 now makes "balance with banks" mandatory, giving lenders a direct cross-check against the bank statement.
Where returns are weak or missing, see our guide to income estimation for thin file borrowers.
How Does DocuEye Automate ITR Data Extraction?
DocuEye automates ITR data extraction by reading ITR PDFs and scans with form-specific templates, detecting fields, applying validation rules and exporting structured data to your systems. Analysts then review only the flagged exceptions.
DocuEye is Digiqt's AI document extraction product for PDFs, images and scans. It works in four stages: upload, choose a template or custom fields, process with an optical character recognition (OCR) and logic-based engine, and export editable output.
| Step | Manual Today | With DocuEye |
|---|---|---|
| Collect | Analyst chases PDFs over email | PDFs, images and scans uploaded once |
| Classify | Analyst opens each file to find form and year | Extraction template set for each ITR form |
| Extract | Figures retyped schedule by schedule | Smart field detection reads PAN, AY, net profit |
| Check status | Filing section often skipped | Filing section and dates extracted as fields |
| Validate | Depends on the analyst | Built-in validation rules flag mismatches |
| Publish | Copied into the CAM | Integration sends data to your loan origination system |
Eligible-income calculation stays in your own credit rules. Our guide to OCR accuracy for Indian loan documents explains how to measure field-level accuracy.
Should Lenders Ask for the ITR PDF, JSON or ITR-V?
Lenders should ask for the complete ITR form in PDF or the JavaScript Object Notation (JSON) data file, because the ITR-V, the one-page acknowledgement issued after filing, carries only summary totals. According to the Income Tax Department's e-filing help page, taxpayers can download all three from the View Filed Returns screen.
Unlike GST returns, which lenders can fetch through the account aggregator framework, ITRs still usually arrive as borrower-shared PDFs and scans. That is where extraction saves the most time.
How Should Lenders Treat Revised, Belated and Updated Returns?
Lenders should treat revised, belated and updated returns differently, by extracting the section each return was filed under and applying a matching review rule.
| Filed Under (Income-tax Act, 1961) | Return Type | Underwriting Treatment |
|---|---|---|
| Section 139(1) | Original, on time | Standard evidence |
| Section 139(4) | Belated | Accept; note the delay |
| Section 139(5) | Revised | Use the latest; compare with the original |
| Section 139(8A) | Updated (ITR-U) | Review if filed close to the application |
According to the Finance Act, 2025, an updated return can be filed up to 48 months after the end of the AY. Under CBDT Notification No. 5/2022 (29 July 2022), a return not verified within 30 days takes the verification date as its filing date, so extract both dates.
Which Validation Checks Should Follow ITR Data Extraction?
Eight validation checks should follow ITR data extraction: PAN match, name match, year coverage, filing status, arithmetic, Schedule BP tie-out, audit flag and cross-document consistency.
| Validation Check | What It Catches |
|---|---|
| PAN matches the applicant | Returns from another person |
| Name and status match | A firm's return filed for a proprietor |
| Every required AY present once | Gaps or duplicates |
| Filing section and dates captured | Updated returns treated as original |
| Heads of income add up | Extraction errors, edited figures |
| Net profit ties to Schedule BP (business income) | Wrong figure picked |
| Audit flag fits turnover | Missing audited financials |
| ITR vs GST turnover and bank credits | Inflated or hidden income |
Our guides to GST return extraction for loan underwriting and fake bank statement detection cover the last check.
Is your team retyping two years of ITRs for every self-employed file?
See How DocuEye Extracts ITR Data
Explore DocuEye's templates, field detection and validation rules, or talk to our document automation team.
Worked Example: Two ITR-3 Returns, One Eligible Income
In this illustrative example, two original ITR-3 returns from a self-employed chartered accountant give eligible income of ₹19.65 lakh a year once the lender's averaging rule is applied.
| Extracted Field | AY 2025-26 | AY 2026-27 |
|---|---|---|
| Net profit | ₹14.5 lakh | ₹21.5 lakh |
| Depreciation | ₹1.5 lakh | ₹1.8 lakh |
| Eligible income | ₹16.0 lakh | ₹23.3 lakh |
Income grew 46%, and the policy averages two years whenever growth exceeds 25%, giving about ₹1.64 lakh a month. Using only the latest year would have overstated eligible income by 19%, and extraction makes the same rule run on every file.
How Will the Income-tax Act, 2025 Change ITR Data Extraction?
The Income-tax Act, 2025 changes ITR data extraction in three ways: new section numbers, split due dates and new form fields. According to a CBDT press release, the Act came into force on 1 April 2026 and replaces assessment year with tax year.
- New section numbers. Section 58 merges presumptive schemes 44AD, 44ADA and 44AE from tax year 2026-27.
- Split due dates. Under the Finance Act, 2026, non-audit business cases file by 31 August, and ITR-1 and ITR-2 filers by 31 July.
- New fields. AY 2026-27 forms add the ITR-4 bank balance and futures and options (F&O) turnover columns.
Lenders will hold returns under both Acts for years, so templates should map old and new section numbers to the same field.
Want income figures your credit committee does not have to re-check?
Explore DocuEye for Lending Documents
Learn how DocuEye turns ITRs into validated, credit-ready data, or visit Digiqt to book a demo.
Sources
- CBDT Press Release: Income-tax Act, 2025 Comes Into Force From 1 April 2026
- Income Tax Department: How to Know the ITR Status
- SCC Online: Key Changes in ITR Forms for AY 2026-27
- Taxscan: Bank Balances Mandatory in ITR-4 for AY 2026-27
- Taxmann: ITR Forms AY 2026-27
- Taxmann: ITR-U 48-Month Window
- Taxmann: ITR Verification Time Limit Cut to 30 Days
- TaxGuru: Presumptive Taxation Under Section 58
- Angel One: ITR Deadlines Changed in Budget 2026



