Skip to content
Idea Atlas
Sign in

StatementKit

As described:StopNarrowed:Continue

No as a generic PDF-to-Excel converter. Go as a reconstruction tool for litigation, tax-resolution and catch-up work.

forensic & catch-up accountants · ~$73 a month · last reviewed (next review by 29 Dec 2026) · 33 cited sources

Save
Share

← All research

One-Page Brief

Verdict

As described, a document-to-spreadsheet converter covering bank statements, payslips, brokerage statements and invoices, for all accountants, has about a 2% chance of reaching $10k MRR in 24 months and ~1% of reaching $20k (judgment estimates). The demand is proven: one solo founder turned the same idea into $318,112.61 of revenue in 2024 (10). But the plain job, "turn this PDF into Excel", became a commodity during 2025–26:

  • QuickBooks Online now extracts PDF statements with its own AI, for any bank, as long as the statement is in English;
  • a general model reads a page for well under 1 cent;
  • DocuClipper, Dext, ProperSoft, Nanonets, Parseur and dozens of clones already sell it.

It isn't fundable, and it doesn't need to be. Cut to one job, reconstructing financial history from statements that bank feeds can't reach, it's a credible small business: ~10% at ~15 hrs/week and ~18% full-time for $10k MRR by month 24.

The five things that matter

  1. The comparable is real, but it was early. Bank Statement Converter went from $3,181.72 (2021) to $49,427.67 (2022), $147,336.55 (2023) and $318,112.61 (2024). By mid-2025 it was making about $38–42k a month. It won with bank-specific parsing, obsessive accuracy fixes and long-tail SEO, and it started in 2021, before LLMs made extraction cheap. You're starting 5 years later.
  2. QuickBooks has already taken the easy job. As of 24 Aug 2026, QuickBooks Online lets users upload a PDF or image statement and "our AI extracts the info". The limit is English-language statements, up to 1,000 lines per upload (10). Single-statement catch-up for QuickBooks users is now a free feature.
  3. Bank feeds shrink the current-period job, not the historical one. Plaid returns up to 24 months of history, with the account holder's consent. UK open banking passed 1 billion payments in July 2026. The US open-banking rule is being reconsidered. None of this reaches the opposing party's statements in a divorce, a deceased person's accounts, closed accounts or 5-year-old records. That's the wedge.
  4. The moat is proof and bulk, not extraction. An LLM can read a statement. It can't prove that 300 statements across 6 accounts all balance, that no month is missing, and that every row traces back to a page. Forensic and tax-resolution professionals have to defend those numbers. That's what they'll pay for.
  5. The market is narrow and episodic. Case-based users churn between cases. The plan only works with annual plans, case passes for occasional users, and catch-up bookkeeping as the "between cases" use.

Feature sort (headline)

  • Table stakes: PDF bank and credit card statements to CSV and Excel, scanned PDF (OCR) support, QuickBooks and Xero import files.
  • Differentiators: per-statement balance proof, row-to-page citations, case folders with auto-sorting and gap detection, brokerage statements parsed into sections, cross-account transfer matching.
  • Distractions: invoices and receipts, payslips at launch, a mobile app, many countries at once.
  • Liabilities: "99.9% accuracy" claims, "court-ready" or "audit-ready" labels, keeping customer files, a pure LLM pipeline with no balance gate.

US accountants who rebuild history from PDFs: forensic accountants, divorce financial analysts, tax-resolution practitioners (unfiled years), catch-up bookkeepers and estate accountants. Rejected wedges:

  • India into Tally. Vyapar TaxOne sells unlimited statements for ₹10,000 a year.
  • Invoices. Dext and Hubdoc own them.
  • Lender APIs. Enterprise sales.
  • Non-English statements. A real gap, since QuickBooks is English-only, but unverified. It's kept as Pivot B.

See 01 and 02.

Most likely cause of death

Commoditisation before a search position is won. You build a good converter, Google and the LLM tools answer "convert my statement" directly, and QuickBooks and Xero absorb the bookkeeping use. You never get enough organic traffic for the niche. Early warning: at month 6, fewer than 30% of signups come from organic search, and most signups convert one statement and leave.

Next 90 days

Do: find a forensic or tax-resolution design partner, build a 300-statement test corpus, and sell 5–10 concierge cases before building the UI. Then ship the case workflow with the balance proof. Stop: thinking of this as "a better PDF to Excel". It's a reconstruction tool.