InvoiceFit — NRS e-invoice prep for Nigerian SMEs

1. One-liner

Turns a Nigerian SME’s Excel and PDF invoices into NRS-valid e-invoices that clear validation on the first try.

2. Trend signal — why now?

Nigeria turned e-invoicing into law and gave SMEs no soft landing. Phase 1 went live for large taxpayers (₦5B+ turnover) on 1 November 2025. The phased rollout then pulls in medium businesses (₦1–5B) from July 2026 with enforcement January 2027, and small businesses below ₦1B following in 2027. The mechanics are unforgiving: every invoice must carry 55 mandatory fields across eight categories, be submitted as UBL-format XML or JSON, pass real-time validation on the Merchant-Buyer Solution (NRS MBS), and come back stamped with an IRN and a cryptographic seal. Get a field wrong and the invoice is rejected — which under real-time transmission means the sale can’t proceed clean.

Here’s the gap that makes this a business: “The vast majority of Nigerian SMEs and mid-market businesses use accounting tools, or no accounting tool at all, that cannot produce UBL-format XML invoices. Excel spreadsheets, Word templates, and even many legacy versions of accounting software simply cannot interface with the NRS MBS platform.” A Lagos survey found 68% of small businesses were unaware of the mandate at all. Penalties are real money: ₦200,000 plus 100% of the tax due plus interest for processing supplies outside the system, and ₦1,000,000 on day one (₦10,000/day after) for failing to let the tax authority deploy its tech.

QuickBooks and Xero are priced in dollars — punishing at current naira exchange rates — and aren’t NRS-native anyway. The accredited transmission layer (Interswitch, PwC, Duplo and other Access Point Providers) is being built for corporates, not for the bookkeeper running ten clients off spreadsheets. The hole is the prep layer: getting a messy invoice clean and schema-correct before it hits an APP.

Provenance:

• Signal 1 (Demand): “Vast majority of Nigerian SMEs use accounting tools, or none, that cannot produce UBL-format XML; Excel/Word/legacy software cannot interface with NRS MBS” + 68% of Lagos small businesses unaware of mandate — https://businessday.ng/business-economy/article/nigerias-e-invoicing-push-tests-smes-readiness-for-tax-automation/ + https://tryduplo.com/blog/nrs-e-invoicing-explained-irn-qr-ubl — observed 2026-06-01
• Signal 2 (Feasibility): NRS MBS exposes REST APIs for submission/IRN status/rejection handling; APPs and System Integrators are accredited intermediaries (SI license = ₦10M paid-up capital, not ₦100M APP tier), so a prep tool can launch unaccredited and partner an APP for transmission — https://nitda.gov.ng/wp-content/uploads/2025/08/NATIONAL-REGULATORY-GUIDELINE-FOR-ELECTRONIC-INVOICING-IN-NIGERIA-2025.pdf — observed 2026-06-01
• Signal 3 (Economic): Mandatory mandate = forced spend; penalties ₦200K + 100% of tax due + interest, ₦1M day-one; medium-business compliance July 2026 / enforcement Jan 2027; APP ecosystem (Interswitch, PwC, Duplo) already forming and funded — https://www.globalvatcompliance.com/globalvatnews/nigeria-e-invoicing-rollout-key-updates-2026/ + https://interswitchgroup.com/news/interswitch-named-firs-access-point-provider-for-mandatory-e-invoicing/ — observed 2026-06-01 Category: Geographic arbitrage

3. The opportunity

A government turned a clean-data problem into a legal deadline for ~hundreds of thousands of businesses that keep their books in Excel and WhatsApp. The incumbents racing in — the APPs — are solving the transmission problem (the pipe to NRS). Almost nobody is solving the upstream problem: an SME’s actual invoice data is dirty. TIN missing or wrong, VAT inconsistently applied or hand-overridden, line items with no tax category, customer details that don’t match what NRS expects across 55 fields. Hand that to an APP and it bounces.

InvoiceFit is the pre-flight check. Drop in your Excel sheet, your PDF, your photo of a handwritten invoice — it extracts, maps every value to the NRS UBL schema, flags exactly which fields will fail and why, fixes what it can, and hands back a validated payload ready for whichever APP the customer uses. The disruption isn’t against a software incumbent — it’s against the accountant doing this by hand and the APP that assumes clean input. Collapse a 30-minute-per-invoice reformat-and-pray into 30 seconds, and catch the rejection before it costs a sale.

4. Target market

• Primary customer: Nigerian bookkeepers and accounting firms serving 5–50 SME clients each, plus medium businesses (₦1–5B turnover) who handle their own books — the cohort hitting the July 2026 deadline. Lagos, Abuja, Port Harcourt first.
• Why they buy: “My clients send me invoices in Excel and screenshots. NRS wants 55 fields in XML or it rejects. I can’t hand-build that for every invoice for every client before the deadline.” The pain is per-invoice, recurring, and tied to a date they can’t move.
• Rough TAM reasoning: Nigeria has well over a million registered businesses; the VAT-registered SME/medium band running the mandate gauntlet is in the hundreds of thousands. Even capturing the bookkeeping firms — a few thousand active practices each managing dozens of clients — is a multi-thousand-account wedge.
• Why now for them: The deadline is a calendar event, not a preference. July 2026 medium-business compliance with January 2027 enforcement means the buying window opens this quarter. 68% didn’t know in the survey — awareness (and panic) is climbing right now.

5. Product sketch (MVP)

• Import an invoice from Excel/CSV, PDF, or photo (camera/WhatsApp-forwarded image)
• AI extraction + mapping to the full NRS UBL 55-field schema across the eight required categories
• Pre-flight validator: red/green field-by-field report — “TIN format invalid,” “VAT rate inconsistent with line category,” “missing buyer field X”
• One-click fixes and smart defaults for recurring fields (your business header, common buyers, standard tax categories)
• Outputs a clean, schema-valid XML/JSON payload — download or push to a connected APP
• Reusable customer + product master list so repeat invoices auto-fill correctly
• Naira-priced, mobile-first (most SME bookkeeping happens on a phone)
• Bulk mode: batch-validate a month of invoices an accountant has been sitting on

6. AI angle — what’s load-bearing

AI is the product, not a garnish. The core job is turning unstructured, inconsistent human-entered invoice data — Excel with merged cells, a PDF scan, a phone photo of a paper invoice — into a precise 55-field structured schema with the right tax categories and validated TIN/VAT logic. That’s exactly what modern vision + LLM extraction is good at and what rules-based parsers fail at, because every SME’s invoice looks different. Remove the AI and you’re left with a manual data-entry form that’s no faster than what NRS already forces — the product collapses. The AI is what compresses 30 minutes of human reformatting into 30 seconds.

7. Localization angle (if any)

This is localization as the entire moat. The 55-field NRS UBL schema, Nigerian TIN/VAT validation rules, naira pricing (QuickBooks’s USD pricing is a non-starter at current FX), and WhatsApp-forwarded invoice ingestion are all Nigeria-specific. A generic global e-invoicing tool cannot serve this — it doesn’t know the schema or the local input habits. The same playbook later ports to other African mandates (the pattern is spreading), but Nigeria is the wedge.

8. Business model — path to $1M–$5M ARR

• Pricing: SME tier ₦15,000/mo ($10) for self-serve businesses; Accountant/firm tier ₦60,000/mo ($40) for multi-client bookkeepers managing dozens of clients. Usage cap on invoices/month with overage.
• ACV: Blended ~$180–280/customer/year (skewed by firm tier carrying most of the volume).
• Rough math to $1M ARR: ~4,500 accounts at a $220 blended ACV. Realistically: ~2,000 accountant firms ($40/mo) + ~6,000 SMEs ($10/mo) ≈ $960K + $720K… trim to ~1,500 firms + 4,000 SMEs ≈ $1.2M ARR. The firm tier does the heavy lifting.
• Rough math to $5M ARR: Requires expansion beyond prep into transmission (become an accredited System Integrator, capture per-invoice transaction value) and/or porting to a second African mandate. Roughly 8,000 firm-tier accounts plus a transaction-fee layer.
• Expansion path: Prep → transmission (SI accreditation, ~$6.5K capital, captures the send) → light bookkeeping/VAT-filing add-ons → second-country mandate. ACV grows as you move from “fix my invoice” to “run my whole NRS workflow.”

9. Go-to-market wedge — first 100 customers

• Bookkeeper/accountant associations and WhatsApp groups: Nigerian accounting practice is organized into dense professional WhatsApp/Telegram groups and ICAN/ANAN member networks. Post a free “Is your invoice NRS-ready?” checker, seed it in 20–30 of these groups, convert the firms that realize they have 40 clients and a deadline.
• The free pre-flight checker as top-of-funnel: Single-invoice validation free, no signup — “upload your invoice, see what NRS will reject.” Shareable result. Firms hit the paywall when they need bulk/repeat.
• Direct outreach to the July-2026 cohort: The medium-business band (₦1–5B) is identifiable through business directories and FIRS/CAC registration data. Targeted outreach with a Loom showing their own dirty invoice getting cleaned — “here’s exactly which 6 fields would have bounced.”
• APP partnership referrals: APPs (Interswitch, Duplo, PwC) want clean input — dirty invoices are their support burden. Co-market: they send their fumbling SME signups to InvoiceFit for prep, you hand back validated payloads to their pipe.

10. Build complexity — justification

Medium. The web/mobile front end and the validation-rules engine for 55 fields are standard work. The hard parts are (a) robust extraction across genuinely messy inputs — phone photos, merged-cell Excel, scanned PDFs — which leans on off-the-shelf vision+LLM APIs but needs real tuning and a feedback loop, and (b) getting the NRS UBL schema and TIN/VAT validation rules exactly right (domain work, not research). No new model training, no hardware. A 2–3 person team ships a credible v1 in ~12–16 weeks; the schema-correctness work is where the weeks go.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Prep/validation layer needs no accreditation; transmission (later) needs SI license (~₦10M capital).
Ethical — no harm / dark patterns	✅	Helps SMEs comply with the law correctly.
Market exists (evidence above)	✅	Mandatory mandate, 55-field schema, 68% unaware, tools can’t produce UBL.
1–5 person team can build this	✅	2–3 people, ~12–16 weeks.
Launchable with <$50K / ₹40L	✅	Pure software; prep layer dodges the accreditation capital requirement at launch.

All five pass.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	16/20	Hair-on-fire near the deadline: rejected invoice = blocked sale + penalty exposure. Felt per-invoice. Slightly below max only because the sharpest pain clusters around the compliance date rather than every day for every business yet.
Demand evidence	15	12/15	Multiple hard signals: legal mandate, explicit “tools cannot produce UBL” statements, 68% unaware survey, penalty schedule. Docked because direct verbatim SME complaints are thin this early — the mandate just reached SMEs.
Build feasibility	15	11/15	Standard stack + off-the-shelf vision/LLM extraction. Messy-input extraction and exact schema correctness are the real work. ~12–16 weeks.
Distribution clarity	15	11/15	Named channels: accountant WhatsApp groups, ICAN/ANAN networks, free checker funnel, APP referrals. Conversion math is still estimated, not proven.
Revenue mechanics	15	12/15	Naira-native pricing fits wallets; firm tier gives real ACV; path to $1M is a few thousand accounts. $5M needs the transmission expansion.
Time to first revenue	10	7/10	Free checker can convert paying firms within weeks of launch given the deadline pressure; not instant because trust + onboarding a firm’s clients takes a beat.
Defensibility	10	6/10	Soft moat: NRS schema/rules expertise, accountant relationships, accumulating customer/product master data, and an APP partnership. Copyable, but local execution + the SI-accreditation step ahead create a head start.
Total	100	75/100

13. Qualitative modifiers

Founder-fit tags

technical-heavy · domain-expertise-required — needs solid extraction engineering plus someone who genuinely understands Nigerian VAT/TIN rules and the NRS schema (a local accountant co-founder or advisor is close to mandatory).

Key assumptions to validate (3–5)

1. Assumption: Bookkeeping firms will pay ~₦60K/mo to avoid manual reformatting across many clients. How to test: 30 interviews with Lagos/Abuja accounting practices; show the cleaned-invoice demo, ask for a pre-order/LOI.
2. Assumption: AI extraction is accurate enough on real Nigerian invoices (phone photos, Excel) that fixes are minor, not a full re-entry. How to test: Run 100 real sample invoices collected from 5–10 firms through a prototype; measure field-level accuracy.
3. Assumption: APPs see clean input as their problem and will refer/partner rather than build prep themselves. How to test: Direct conversations with 2–3 accredited APPs about a referral arrangement.
4. Assumption: The prep layer truly needs no accreditation to launch. How to test: Confirm with a Nigerian tax/tech lawyer that producing validated payloads (without transmitting) falls outside the SI/APP licensing trigger.

Risk flags

1. Regulatory/platform risk: NRS could change the schema, or push APPs to bundle prep for free, collapsing the wedge. Mitigate by moving toward SI accreditation and stickier master-data lock-in.
2. Market timing: Slightly early for the smallest SMEs (their deadline is 2027). The July-2026 medium-business cohort is the live market; mis-targeting micro-businesses now wastes effort.
3. APP-eats-the-layer risk: A well-funded APP could add a good-enough extraction front end. Speed, accountant relationships, and naira pricing are the defense.
4. Extraction-quality risk: If AI accuracy on messy local invoices is poor, the value prop (faster than manual) evaporates. This is the single most important thing to test first.

14. Structured verdict

Score:                  75/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Technical founder + Nigerian accountant/tax co-founder or advisor
Time to revenue:        8–14 weeks (deadline pressure compresses the funnel)
Capital to launch:      ₦6–10M / $4–7K (pure software; accreditation deferred)
Top 3 assumptions to validate first:
  1. AI extraction accuracy on real Nigerian Excel/PDF/photo invoices — run 100 real samples, measure field-level accuracy
  2. Accountant firms will pay ~₦60K/mo — 30 interviews + LOIs/pre-orders
  3. Prep layer needs no accreditation to launch — confirm with Nigerian tax/tech lawyer
Kill criteria:
  - Abandon if field-level extraction accuracy on real samples is below ~90% (manual fix-up kills the time savings)
  - Abandon if <5 of 30 interviewed firms show concrete buying intent (LOI/pre-order)
  - Abandon if a legal check shows the prep layer itself triggers SI/APP accreditation before launch

15. Next step — 1-week validation sprint

• Day 1–2: Collect 100 real invoices from 5–10 friendly Lagos bookkeeping firms (Excel, PDF, phone photos). Pull the NRS UBL 55-field schema and the TIN/VAT validation rules.
• Day 3–4: Wire a throwaway prototype: vision/LLM extraction → schema mapping → field-level pass/fail report. Run all 100 invoices. Score field-level accuracy. In parallel, 30 short interviews with accountants — show the cleaned-invoice demo, ask the ₦60K/mo question, collect LOIs.
• Day 5: Go / no-go. Go only if extraction field-accuracy ≥90% AND ≥5 of 30 firms give concrete buying intent. Both numbers are falsifiable — no vibes.