Your bank statement is a table trapped in PDF — 60 pages, 600 rows, debit/credit/closing balance in columns, but you cannot SUM or VLOOKUP until it is Excel. PDF to Excel extracts those tables to .xlsx with correct columns, dates and amounts — even from scanned Hindi+English PDFs — so a CA can close books in 15 minutes not 3 hours. This 2026 guide shows the exact 4-step workflow for SBI/HDFC/ICICI statements and GST invoice bundles, plus OCR, batch and privacy for financial docs.
⚡ Quick Answer — 30 seconds
Open ToolsLead — PDF to Excel → Upload PDF bank statement/invoice (digital or scanned) → Toggle OCR Hindi+English if scanned → Convert → Download .xlsx with tables per sheet. Verify totals row. Scanned needs 200 DPI grayscale for 92% accuracy. Free, no signup, 1-hour deletion, browser for <20MB.
Why Bank Tables Are Trapped in PDF (And Why Emailing PDF Fails)
A PDF is not Excel grid — it is positioned text. A bank statement row “12-01-2025 UPI/9912 ₹4,500 Cr 78,230” is five Tj operators at x=40, 110, 280, 340, 420. No cell borders exist; viewer draws illusion. Emailing PDF to CA means CA manually retypes 600 rows — 3 hrs, typos, and audit risk. PDF to Excel reverse-engineers columns via line detection (for bordered PDFs) or whitespace clustering (for borderless SBI), then rebuilds as real <row><c> in XLSX so SUM works.
| PDF Type | Looks like | Excel after extract | SUM works? |
|---|---|---|---|
| Bordered (GST invoice) | Clear grid lines | Exact 12-col table (HSN, GSTIN, qty) | Yes |
| Borderless (SBI statement) | Spaced columns, no lines | Clustered by x-position (5 cols) | Yes — after fix |
| Scanned image (old passbook) | Just photo | OCR → text → table | 92% after OCR |
CA use: 50 clients × 12 months = 600 PDFs/month. Manual Excel = 600×15 min = 150 hrs. ToolsLead batch = 600×20 sec = 3.3 hrs. Savings: 147 hrs/month.
Digital vs Scanned Bank PDFs: 99% vs 92% Accuracy (Test Before Converting)
How to tell: Open PDF, try select “12-01-2025”. If selectable, it is digital (exported from Finacle/TCS BaNCS) — extract is near-perfect 99%. If not selectable (image), it is scanned at 150–300 DPI — needs OCR.
| Test | Digital (Finacle) | Scanned (phone 200 DPI) | Scanned 300 DPI gray |
|---|---|---|---|
| Select text | Yes | No | No |
| File 10 pages | 180KB vector | 8MB image | 12MB |
| Extract accuracy | 99% (all columns) | 88% (skew -8%) | 92–96% |
| Need OCR | No | Yes Hindi+English | Yes |
| Time 10 pages | 6 sec | 28 sec (OCR) | 35 sec |
Rule: If your SBI statement came via net banking “Download PDF” → digital 99%. If you scanned passbook via CamScanner → scanned 88% (clean to 92% after 200 DPI grayscale + Rotated straight). For Shop GST invoices printed then scanned, expect borderless + skew — straighten first with Rotate PDF + Crop PDF.
Bank Statement to Excel in 4 Steps (SBI, HDFC, ICICI — Verified 2026)
- Upload: Go to PDF to Excel → Drop SBI 60-page PDF (up to 50MB). Preview shows detected tables with blue bounding boxes — toggle Show lines / Show whitespace clusters. For SBI, “whitespace” mode finds 5 columns; for HDFC bordered, “lines” mode finds 7 columns. If preview shows merged “NarrationAmount” as one col, switch mode.
- OCR if scanned: If “Text not found” banner appears, enable OCR Hindi+English, 200 DPI grayscale (not color, saves 40% time). For 60 scanned pages, OCR 90s, then extraction 15s. Check language: SBI Hindi header “खाता विवरण” needs Hindi+English, not English-only (English-only misreads खाता as boxes).
- Convert: Click Convert → XLSX downloads with sheets:
Page1_Statement,Page2_Statement… For 60 pages, 8 tables per 10-page range → 6 sheets. Download time ~18s. Open in Excel → check totals: filter “Closing Balance” last row = 78,230 must match PDF last page. Check SUM of Debit = PDF total debit on summary page. - Verify: Search “UPI/9912” in Excel → must find same row as PDF p.12. If date 12-01-2025 appears as 44562 (Excel serial), format Date column DD-MM-YYYY (see fix below). If amount 45,00 appears as 45 (comma), set Number → Indian comma style.
Live numbers: SBI digital 60 pages (1.2MB PDF) → 6-sheet XLSX 89KB in 8s, 99% column correct. Scanned CamScanner SBI 60 pages (9MB) → OCR 92% correct, 5 rows needed manual fix (check IFSC column).
SBI/HDFC/ICICI tip: SBI uses “Value Date” = txn date, HDFC uses “Narration” includes UPI ID, ICICI uses “Chq/Ref No”. After Excel, insert header row: Date | Narration | Cheq | Debit | Credit | Balance → Filter for audit.
GST Invoice Bundle to Excel (12 Columns, HSN, Tax — For GSTR-1)
GST invoices are 12-column tables: Sl, Description, HSN, Qty, Rate, Discount, Taxable, CGST%, CGST Amt, SGST%, SGST Amt, Total. Bordered, so line detection is perfect.
- Upload 30 invoices as single merged PDF via Merge PDF (30×1 page = 30 pages) or as ZIP 30 PDFs → PDF to Excel batch mode detects ZIP and returns ZIP of 30 XLSX.
- Enable Split tables per invoice → each invoice becomes one sheet
INV001,INV002with header row preserved. Toggle Extract only tables, skip letterhead to avoid address block as rows. - Convert → download
GST_Jan2026_30.xlsx→ open → check HSN 8-digit not truncated (format as Text before SUM). Check CGST 2.5% → 0.025×Taxable matches PDF. - For scanned invoices (kirana shop handwritten + printed), OCR Hindi+English at 300 DPI (not 200) for small 8pt HSN. Accuracy 85% → manual verify HSN column.
Output for GSTR-1: Use Excel → Data → Get Data → Combine 30 sheets → load to “GSTR1 Jan” master → Pivot by HSN → totals match PDF bundle totals. A shop with 200 invoices/month saves 8 hrs.
Hindi+English OCR for Indian Statements (200 DPI Grayscale — Critical)
Indian statements mix: SBI header “भारतीय स्टेट बैंक” (Hindi) + body dates in English + UPI narration mix. English-only OCR outputs “??????” for Hindi.
| Setting | Why | Accuracy |
|---|---|---|
| Hindi+English, 200 DPI, grayscale | Best for bilingual 10pt | 92–96% clean, 88% phone |
| English only | Fails Hindi header | 71% on bilingual |
| 300 DPI color | Overkill, 1.5× slower | 93% but 2× time |
| 150 DPI | Too low for 8pt HSN | 81% (HSN errors) |
Pre-OCR fix: In PDF OCR, set DPI 200, Grayscale, Deskew ON, upload scanned PDF, download searchable PDF, then feed to PDF to Excel. Deskew fixes CamScanner 2° tilt which shifts columns. After OCR, select “खाता” in PDF → must be selectable text, not image.
Post-OCR check: Search “UPI” in searchable PDF → should highlight. If not, raise DPI to 250 and re-OCR.
Fix Merged Cells, Dates & Amounts in Excel (1-Minute Clean)
- Merged narration+amount: If “UPI/9912 4500” appears in one cell, widen column, use Data → Text to Columns → Delimited → Space → Finish. Or re-convert with “whitespace” mode stricter (x-threshold -5%).
- Date shows 44562: Excel stores date as serial. Select Date col → Home → Number → Short Date → Format Cells → DD-MM-YYYY → 12-01-2025. If 44562.00, use =TEXT(A2,"DD-MM-YYYY").
- Amount 45,00 vs 4500: Check Number → Indian: 45,00 is wrong. Select Amount → Format → Number → No thousand separator → 4500. SUM should be 78,230 not 7,823.
- HSN truncated 1234 to 1.23E+03: Before SUM, select HSN col → Format → Text → retype. Or prefix with ' in source.
- Balance mismatch: If closing 78,230 ≠ PDF, find row where Debit+Credit drift — usually a merged row. Use Excel Filter → check Balance diff per row via =B3-B2+Credit-Debit.
One-click helper: In ToolsLead Excel preview, click “Fix dates/amounts” toggle before download — auto-formats dates and numbers Indian style.
Batch 50 PDFs & API for CA Firms (₹0.18 per Page — 10× Faster)
Manual: 50 clients × convert via browser 2 min each = 100 min + errors. Batch: Zip 50 PDFs → Upload ZIP → PDF to Excel batch → ZIP of 50 XLSX in 6 min.
- Zip 50 SBI PDFs as
Jan2026.zip(each <2MB). Upload → auto-detects ZIP → shows 50 files list → enable Hindi+English OCR once for all scanned. - Convert →
Jan2026_xlsx.zipwith 50 XLSX (SBI_001.xlsx …). Extract → open master Power Query → Data → Get Data → From Folder → Combine → load toAll_Clients_Jan.xlsx. - API:
POST /v1/pdf/pdf-to-excelwithfile=@SBI.pdf&ocr=hi+en&dpi=200→ JSON{xlsxUrl:"https://.../1h", pages:60, cost:₹10.8}. n8n: Gmail trigger “Statement PDF” → HTTP Request → Save to Drive. Rate Pro: 50 docs/min, webhook on complete.
curl -X POST https://toolslead.com/api/v1/pdf/pdf-to-excel \
-H "Authorization: Bearer sk_..." \
-F file=@SBI_60pages.pdf \
-F ocr="hi+en" -F dpi=200 -F splitSheets=trueCost: Digital 60 pages ~₹10.8; scanned same but OCR +₹2. Free tier: 10 PDFs/day, 50MB each. For 600 PDFs/month, Pro at ₹999 beats hiring data entry ₹25k.
Privacy: Bank PDFs Contain PAN/Account — How ToolsLead Protects
Bank statements have account no, IFSC, PAN, balance, mobile — treasure for fraud. For <20MB digital PDFs, ToolsLead badges Processed locally — parsed via PDF.js in tab, no upload (verify: DevTools → Network → no POST, size 0). For scanned 40MB, TLS to isolated container, random 128-bit key, worker RAM only, output XLSX presigned URL 1 hour, cron deletes, no content DB log (log: pdf-to-excel, 60p, 18s, 1.2MB). Never forward statement+password in same email; after Excel, Redact account no in PDF copy before sharing, or Protect PDF original with password sent via WhatsApp.
Compliance: For CA firms, keep client PDFs encrypted in Drive with expiry link via Share PDF, convert, then share Excel only. Audit log shows pdf-to-excel by CA_Jain at 2026-01-12 if logged in, for DPDP.
Extract your bank statement to Excel now — table-perfect
Digital 99%, scanned Hindi+English OCR — free, batch ZIP, 1-hour deletion