3. Feature-by-Feature Scrutiny (Your Stated Features)
Categories:
- Table stakes (TS): customers expect it. It earns nothing, but its absence costs you.
- Differentiator (D): a real reason to choose you over the alternatives.
- Distraction (X): built but unlikely to move revenue or retention.
- Liability (L): creates support, compliance, legal or trust risk.
Some features sit in two buckets. The primary bucket comes first.
3.1 Classification table
| # | Your feature | Category | Reasoning |
|---|---|---|---|
| 1 | PDF bank statements to CSV or Excel | TS | This was the whole product for the comparable in 2021. In 2026 it's free in QuickBooks (English), one prompt in any LLM chat, and $20 a month at DocuClipper. You need it, and it earns nothing on its own. |
| 2 | Credit card statements (implied) | TS | Same as bank statements. Every competitor includes them. |
| 3 | Brokerage statements | D (for the wedge) | Multi-section documents (holdings, activity, dividends, realized gains) that break "one table per page" converters. DocuClipper claims brokerage support, so the edge comes from splitting sections into proper sheets and proving the totals, not just from having it. Estate and divorce work needs it. P1, top 10 US brokerages. |
| 4 | Payslips | X at launch | The main buyer for payslips is lenders (income verification), and they buy through enterprise deals. Divorce income schedules need them occasionally. P2. |
| 5 | Invoices | X / L | A different job (accounts payable), a different buyer, and the most crowded category in accounting software: Dext, Hubdoc, QuickBooks, Xero, Bill and every AP tool. It also adds VAT and tax-field liability. Cut. |
| 6 | CSV and Excel output | TS | Expected. The differentiator is what's in the workbook (proof column, source citations, one sheet per account, a summary sheet), not the file type. |
| 7 | Import files for accounting packages (QBO, IIF, OFX, QFX, Xero CSV) | TS | DocuClipper exports 7 formats, and ProperSoft built a business on QBO files. QuickBooks now reads PDFs directly. Needed for the catch-up buyer, and irrelevant to litigation. P1, QBO + Xero CSV only. |
| 8 | Wedge: "a region's banks" | Depends on the region | India/Tally: X. Vyapar TaxOne sells unlimited statements for ₹10,000 a year (verified). Non-English markets: possible D. QuickBooks' native import is English only (verified), but local competitors are unverified and EU feeds are strong. Pivot B. US long tail (credit unions, small banks): D inside the recommended wedge, because that's where general tools fail most. |
| 9 | Wedge: "document types the leader handles poorly" | D, if it means multi-account cases and brokerage | The leader (Bank Statement Converter) is built around one statement becoming one table. The job it handles least well is many statements becoming one proven dataset. That's the wedge. |
| 10 | AI extraction (implied) | TS | Everyone uses it. A model alone is a liability (silent dropped rows). Wrapped in deterministic checks, it becomes part of a differentiator. |
| 11 | "For all accountants and bookkeepers" positioning | L | It makes you a generic converter in a price war with QuickBooks. It also makes every search page compete with the comparable's 5-year head start. |
3.2 Scoreboard
| Category | Count | Features |
|---|---|---|
| Table stakes | 5 | Bank statements, credit card statements, CSV/Excel, accounting import files, AI extraction |
| Differentiator | 2–3 | Brokerage statements (sectioned and proven), multi-statement cases (the reframed wedge), US long-tail layouts |
| Distraction | 3 | Payslips at launch, invoices, India/Tally |
| Liability | 1–2 | "All accountants" positioning, invoice tax fields |
About 20% of what was described is a reason to buy. The rest is table stakes or a fight with someone bigger. The real differentiators, proof, provenance and case workflow, weren't in the original description at all. They come from the expanded catalog in 04.
3.3 Feature bloat and "building to avoid selling": explicit call-outs
- More document types. Every new type (invoices, payslips, tax forms, receipts) feels like widening the market. Each one is actually a new competitor set and a new accuracy problem. Add a type only when 5 paying accounts ask for it.
- More output formats. IIF, QFX, QIF, QBJ, Sage, Zoho, Tally XML... Each is a day to build and a lifetime of support tickets. Ship Excel, CSV, QBO and Xero CSV. Add others only on paid demand.
- Analytics dashboards. Charts of spending, "lifestyle analysis", fraud scores. They demo well, the buyer's own expertise already covers them, and Valid8 and DocuClipper Business market them. Export clean data and let the expert analyze it.
- A "universal AI parser" instead of the test corpus. Tuning prompts feels like progress. The corpus plus balance gate is progress.
- Many countries at once. Each country is new layouts, a new language for search pages, and new accounting formats.
3.4 What a disciplined v1 looks like
Keep:
- a case folder: bulk upload, auto-sort by account and period, duplicate and gap detection;
- bank and credit card statements for the top ~50 US layouts, plus an LLM fallback;
- a balance proof for every statement;
- row-to-page citations;
- an Excel workbook (one sheet per account, plus a summary);
- a CSV export.
Next: brokerage statements, QBO and Xero import files, cross-account transfer matching, and a Firm plan with seats.
Cut: invoices, payslips (until paid demand), India/Tally, dashboards, a mobile app, and non-US countries (until Pivot B is tested).