9. Assumptions and Open Inputs
The idea arrived as a one-paragraph concept from the 30 Sept 2026 micro-SaaS research, with evidence on the comparable (Bank Statement Converter). Team, hours per week, geography, stage, pricing and traction weren't supplied. Here is what I assumed, and how the verdict changes if each assumption is wrong.
9.1 Assumptions made
| Input | Assumption used | If actually different… |
|---|---|---|
| Hours per week available | Two tracks modelled: ~15 hrs/week and full-time | Below ~10 hrs/week, the search routine and the layout fixes can't both happen, so halve the 15-hrs/week odds. |
| Target customer | US reconstruction work (tax resolution, forensic, divorce, estates) plus catch-up bookkeeping | If you have a strong network in a non-English market, Pivot B (non-English statements) moves up. Check local competitors first. |
| Geography of sales | US first, priced in USD, sold online from anywhere | If you can't take USD payments or form a US-friendly entity, add 1–2 months of setup (Stripe Atlas or similar). |
| Stage | Idea. No code, no users, no revenue. | Any existing converter code or parsing experience shortens v1 by 4–6 weeks. |
| Team | Solo, technical, no stated accounting or litigation background | A founder who has worked in forensic accounting, tax resolution or bookkeeping is the single biggest positive change (+~10 points on the $1k–5k odds). |
| Pricing | Recommended: Solo $49, Firm $149, Case pass $99. Blended subscription ARPU ~$73. | If the market only accepts ~$30 (the comparable's entry price), you need ~40% more accounts at every level. |
| Churn | ≤5% monthly after month 6, with annual plans | At 10%, $10k is barely reachable and $20k isn't. See the churn check in WHAT-TO-DO-AND-AVOID.md. |
| Model costs | ~$0.003–0.007 per page with one retry (Gemini Flash-Lite prices, verified) | Even at 10x that, gross margin stays above 80%. It isn't a risk. |
| Platform moves | QuickBooks stays English-only and single-statement for 12+ months. Xero's PDF import is unknown. | If either ships bulk multi-statement import with proof, the catch-up segment disappears. The litigation and estate segments survive. |
| Safeguards rules | US tax professionals must oversee service providers handling client data | Not verified in this run. Confirm before writing the security page. |
9.2 Inputs that would change the verdict most (please supply)
- Hours per week available, and for how many months.
- Your background. Any accounting, bookkeeping, tax, legal or banking experience? Any PDF parsing or OCR work?
- Your network. Do you know any forensic accountant, divorce financial analyst, enrolled agent, CPA doing tax resolution, or catch-up bookkeeper?
- Country and entity. Can you invoice and collect in USD? Does US data handling create issues for you?
- Languages. Are you fluent in any non-English market's language, and do you know its accountants? This is for Pivot B.
- Budget. Can you pay a design partner $300–500 a month for 6 months?
- Risk appetite. Are you willing to handle litigation documents at all? If not, drop that segment and the odds fall by about a third.
9.3 Definitions used throughout
- Reconstruction work: rebuilding a transaction history from statements for a purpose other than ongoing bookkeeping. That covers tax resolution, litigation, divorce, estates and catch-up.
- Balance proof: a statement's opening balance plus extracted rows equals its closing balance, and every running balance matches.
- Case: a set of statements for one client or matter, often spanning several accounts and months.
- MRR: recurring subscription revenue only. Case passes are excluded.
- Tier A / Tier B: see 01-verdict.md.
- Churn benchmarks (analyst estimates): small-firm B2B SaaS 3–5% a month. Episodic professional tools 6–10% a month without annual plans.