OYN: Market Position & Value Thesis¶
Founders Report — Steve Greszczyszyn & Ciaran O'Donnell¶
February 17, 2026¶
What This Is¶
This report is based on data we've systematically collected and analyzed over the past 48 hours:
- 432 Xero App Store apps scraped with full metadata
- 16,000+ individual reviews with ratings, dates, and developer responses
- 4,000 product ideas from Xero's feedback forums with vote counts and statuses
- 651 GitHub issues from Xero's developer repos
- 9 AI-powered research queries via Perplexity and Grok (pulling Reddit, forums, news)
- 4,086 extracted signals (pain points, feature requests, disruption opportunities)
- 8 independent research sources consolidated into a master strategy
This isn't opinion. It's market intelligence.
The One-Line Thesis¶
There is no AI financial analyst in the Xero ecosystem. The market is begging for one. OYN can be first.
The Market We're Entering¶
Xero by the Numbers¶
| Global subscribers | 4.6 million |
| Annual revenue | NZ$2.1B (+23% YoY) |
| UK growth | +25% YoY (fastest major market) |
| Monthly churn | 1.03% (best-in-class for SaaS) |
| Ecosystem apps | ~1,000 |
| Platform fee | 0.3-0.6% (vs Apple/Google 15-30%) |
The platform economics are absurdly favorable. A 0.3% platform fee vs 30% on consumer app stores means 50-100x better unit economics. And the buyers are accountants and business owners who pay because they have to — compliance software isn't discretionary.
The AI Gap¶
We scraped every app in the Xero App Store. Here's what the AI landscape actually looks like:
| App | Rating | Reviews | What It Does |
|---|---|---|---|
| Booke AI | 4.57★ | 124 | AI bookkeeping, transaction categorization |
| Tofu | 4.96★ | 52 | AI bookkeeping, 200+ languages |
| Vinyl | 5★ | 35 | AI notetaker for accountants |
| fileAI | 4.76★ | 37 | AI OCR for receipts/invoices |
| Simple AI Accounting | 5★ | 12 | AI agent for bookkeeping |
| FINNEX AI | 5★ | 3 | ChatGPT in Excel for Xero data |
| Clockwork AI | 5★ | 1 | FP&A platform |
There is no AI financial analyst. Booke AI does bookkeeping. Tofu does bookkeeping. FINNEX is an Excel plugin. Nobody is doing what OYN does: taking Xero data and producing actionable financial analysis with AI reasoning.
The closest thing to a competitor is a Reddit post from August 2025 — a solo developer building an "AI Financial Health Analyzer" in early alpha. That's it. One person, alpha stage.
What Users Actually Hate¶
Our signal analysis flagged 70 pain point signals and 12 disruption targets. The most vulnerable apps in the ecosystem:
| App | Rating | Reviews | What's Wrong |
|---|---|---|---|
| Stripe integration | 2.6★ | 351 | "Fees insane," "blocked account," zero support |
| Hubdoc | 3.34★ | 225 | "Stuck in the past," clunky, sync failures |
| WorkflowMAX | 3.69★ | 192 | Xero's own product — users furious about retirement |
| Shopify by Xero | 2.96★ | 139 | Xero built this themselves and it barely works |
| PayPal | 3.49★ | 135 | Reconciliation nightmare |
The pattern: Xero's own integrations are among the worst-rated in the store. Users are desperate for third-party solutions that actually work.
What Users Actually Want¶
Top product ideas by vote count (these are features Xero doesn't have and won't build):
| Idea | Votes | Status |
|---|---|---|
| Employee file uploads | 976 | Accepted (slow) |
| Schedule email sends | 683 | In development |
| Bulk file export | 663 | Accepted |
| Auto-email statements | 662 | In discovery |
| Custom invoice fields | 655 | Not in pipeline |
| Microsoft SSO | 641 | In development |
| Bank account access roles | 633 | Accepted |
| Item/description-only invoices | 550 | Accepted |
| Homepage simplification | 543 | In discovery |
| Invoice file attachment | 516 | Not in pipeline |
The "not in pipeline" items with 500+ votes = guaranteed demand that Xero has explicitly declined to solve. Third-party opportunity.
OYN's Competitive Position¶
The Reporting & Forecasting Category¶
This is our home category. 86 apps. Here's who matters:
| App | Rating | Reviews | Price Range | AI? |
|---|---|---|---|---|
| Fathom | 4.95★ | 557 | $39-389/mo | ❌ No |
| Syft (Xero-owned) | 4.83★ | 472 | Free-paid | ❌ No |
| joiin | 4.91★ | 424 | £16-130/mo | ❌ No |
| Float | 4.83★ | 341 | $59-199/mo | ❌ No |
| Spotlight | 4.85★ | 335 | $49-749/mo | ❌ No |
| Mayday | 4.98★ | 308 | Unknown | ❌ No |
| Futrli | 4.74★ | 140 | Unknown | ❌ No |
| Calxa | 4.98★ | 134 | A$59-299/mo | ❌ No |
Not a single one uses AI. They're all dashboards and template engines. They pull data, format it, and display it. That's it.
The user does the thinking. The user interprets the ratios. The user writes the commentary. The user decides what's important.
OYN does the thinking.
What Users Say About the Incumbents¶
From our review analysis (1-2 star reviews of direct competitors):
Fathom (557 reviews, 4.95★):
"Very bad experience, no consultant available, and reporting modules are very bad"
Syft (472 reviews, 4.83★):
"Unable to handle large datasets and the post sale client support team was unable to fix this" "The Trial period is limited to basic functions. To get access to all functions you have to pay for a month"
Futrli (140 reviews, 4.74★):
"Spammers. Repeated spam calls" "Support is not that responsive. Some emails are not returned"
joiin (424 reviews, 4.91★):
"AR AP and cash data showing wrong figures, support can't work it out. Have requested a refund!"
Spotlight (335 reviews, 4.85★):
"It works... when it works. And lately it has just stopped" "You are limited in the way you are able to present your reports"
These apps have high average ratings but real cracks. The core complaint pattern: they're rigid, complex, and when something goes wrong, support is absent. They've optimized for getting 5-star reviews on easy setups, not for serving power users.
Xero's Own AI Play: JAX¶
Xero launched JAX — their "AI superagent." The market reaction:
"The new AI is useless" "Makes the same errors when reconciling" "Tends to complicate things more often than they clarify" "Stop shilling paid add-ons and useless AI and fix the basics"
JAX is focused on conversational automation of existing Xero workflows (categorization, reconciliation). It is not doing financial analysis. And users already distrust it. This means:
- Xero won't compete with OYN directly — JAX is a different product
- The "AI in Xero" brand is currently tainted — OYN can differentiate by actually delivering results
- Xero's approach (embedded AI on core) leaves the analysis layer completely open for partners
The Timing Advantage¶
The March 2 Disruption Window¶
On March 2, 2026 — two weeks from now — Xero's new pricing model goes live:
- Old model: 15% revenue share (XASS)
- New model: Flat tiered fees based on API usage
Why this matters for OYN:
- Existing apps are scrambling. They need to rebuild billing, re-acquire payment info, optimize API calls. Their engineering bandwidth is diverted to migration, not innovation.
- New entrants start clean. We build on the new pricing from day one. No migration, no legacy.
- Developer sentiment is souring. Multiple developers publicly angry about the changes. Some leaving the ecosystem. Less competition.
Key quote from a developer:
"This would have completely killed our ambition to integrate" — Jacques Malan, Xama Technologies
MTD for Income Tax (April 2026)¶
Making Tax Digital for Income Tax Self-Assessment launches for UK earners >£50K in April 2026. This is a regulatory compliance deadline driving new demand for tax-aware Xero tools. While OYN isn't a tax app per se, the ability to surface tax-relevant insights (VAT efficiency, expense categorization quality) directly serves this cohort.
OYN's Moat¶
Why can't Fathom just "add AI"?
-
Architecture. They're template engines. Adding AI analysis to a template engine is like adding a jet engine to a bicycle. You need to rebuild from scratch.
-
Data model. OYN is designed to reason about financial data. Fathom is designed to display it. The underlying data structures, the API integration patterns, the output formats — completely different.
-
Speed of iteration. Fathom has 557 reviews worth of users expecting a specific product. Any pivot to AI disrupts their existing base. OYN has zero legacy expectations.
-
The "Anti-App" positioning. The market is experiencing App Fatigue. Accountants don't want another dashboard. They want less work. OYN isn't a reporting tool — it's a teammate. "Hire OYN for £29/month. It does the work of a junior analyst."
-
Accountant channel leverage. One accountant = 10-50 client deployments. The "Firm-Wide Audit" play: connect OYN to an accountant's entire client list, generate a free report showing missed opportunities across the portfolio. The accountant becomes your salesperson.
What OYN Actually Does Today¶
OYN is a Weekly Cash Report for small business owners and founders. It connects to Xero, pulls live financial data, and uses AI to produce an actionable one-minute briefing every week.
The Core Output: Weekly Cash Report¶
From the website (ownyournumbers.com):
"We use AI to give founders and CEOs instant financial insights, risk alerts, and actionable priorities in a founder-friendly format."
A typical OYN report delivers:
- Cash Position — Cash in bank, owed to you, owed by you, net cash
- Cash Flow Trend — 3-month average burn rate, cost base
- Cash Cushion — Months of runway at current burn (the number every founder needs)
- Risk Rating — Red/Amber/Green based on cushion months (<2 = High Risk, 2-6 = Medium, 6+ = Low)
- Key Highlights — Most recent sales invoice, top 3 debtors, largest upcoming payments
- 5 Actionable Recommendations — Specific, named: "Chase Randstad for £25k overdue" not "improve collections"
Example output:
Hi James, here's your one-minute weekly cash update. Cash in Bank: £40,323 | Owed to you: £161,479 | Net Cash: £88,883 Cash cushion: 5.8 months | Risk: MEDIUM Action 1: Chase Randstad for £25k — 15 days overdue
Technical Architecture¶
- Backend: Python/FastAPI, PostgreSQL
- AI Pipeline: Multi-model — Gemini 2.5 Pro (primary), GPT-5, Claude 4.1, Grok 4 via OpenRouter
- Xero Integration: OAuth2 with PKCE, pulling Reports API (Trial Balance, Aged Receivables, Aged Payables, General Ledger)
- Security: Customer data never used for AI training (Vertex AI Enterprise guarantees). Financial artefacts auto-deleted in customer environments. No PII in logs.
- Dual ingest: Xero API (primary) + Excel upload (fallback for non-Xero users)
- Infrastructure: Hetzner VPS + Ansible deployment, Cloudflare CDN
Data Flow Per Report Session (Weekly)¶
| Data Type | Xero Endpoint | Typical Size |
|---|---|---|
| Trial Balance | /Reports/TrialBalance |
0.1-0.5 MB |
| Aged Receivables | /Reports/AgedReceivablesbyContact |
0.1-2 MB |
| Aged Payables | /Reports/AgedPayablesbyContact |
0.1-2 MB |
| General Ledger | /Reports/GeneralLedger |
0.5-20 MB |
| Total per session | 4 API calls | ~5 MB typical |
API Cost Structure¶
Xero Developer Platform Costs¶
| Tier | Monthly Fee | Max Connections | Egress Included |
|---|---|---|---|
| Starter | Free | 5 | n/a |
| Core | £15 | 50 | 10 GB |
| Plus | £95 | 1,000 | 50 GB |
| Advanced | £745 | 10,000 | 250 GB |
At ~20 MB/customer/month (4 weekly reports × 5 MB), egress is never the limiting factor. It's connections. And Xero charges per connected org, not per API call — fully predictable.
Year 1 Xero costs: ~£615 (3 months free Starter → 3 months Core → 6 months Plus)
AI Model Costs (Per Report Session)¶
OYN uses a multi-model pipeline via Vertex AI and OpenRouter:
| Model | Input Rate | Output Rate | Typical Cost/Report |
|---|---|---|---|
| Gemini 2.5 Pro | $1.25/M tokens | $10/M tokens | ~$0.05-0.15 |
| Gemini 2.5 Flash | $0.30/M tokens | $2.50/M tokens | ~$0.01-0.03 |
| GPT-5 (OpenRouter) | ~$2/M | ~$8/M | ~$0.05-0.20 |
| Claude 4.1 (OpenRouter) | ~$3/M | ~$15/M | ~$0.08-0.25 |
Typical cost per weekly report: $0.05-0.15 (using Gemini 2.5 Pro as primary) Monthly AI cost per customer: $0.20-0.60 (4 reports/month)
Combined Unit Economics¶
| Scale | Xero Fee | AI Cost/mo | Total COGS/Customer | Revenue/Customer | Gross Margin |
|---|---|---|---|---|---|
| 50 customers | £15 (Core) | £25 | ~£0.80/customer | £29-39 | 97-98% |
| 500 customers | £95 (Plus) | £250 | ~£0.69/customer | £29-39 | 97-98% |
| 1,000 customers | £95 (Plus) | £500 | ~£0.60/customer | £29-39 | 98% |
| 5,000 customers | £745 (Advanced) | £2,500 | ~£0.65/customer | £29-39 | 98% |
Verdict: API costs (Xero + AI combined) are <3% of revenue at any scale. This is a software-margin business.
Path to App Store¶
| Phase | Timeline | Tier | Milestone |
|---|---|---|---|
| Build | Months 1-3 | Starter (£0) | OAuth2 integration, 5 test orgs |
| Beta | Months 4-6 | Core (£15/mo) | 10 beta customers (certification requirement) |
| Certify | Month 7 | Core (£15/mo) | Submit for Xero App Store certification |
| Launch | Month 8 | Plus (£95/mo) | App Store listing goes live |
| Scale | Months 9+ | Plus (£95/mo) | Grow to 100+ customers |
Key gate: Xero requires 10 beta customers actively using the integration before you can submit for certification. The "Firm-Wide Audit" play (connect to one accountant's entire client list) could get you there in a single onboarding.
Revenue Model¶
At £29-39/month per connected org:
| Stage | Connections | MRR | ARR | Timeline |
|---|---|---|---|---|
| Alpha | 25 (free) | £0 | £0 | Months 1-3 |
| Beta | 50 | £1,500-2K | £18-23K | Months 3-6 |
| Growth | 500 | £15-20K | £180-234K | Year 1-2 |
| Scale | 1,000 | £29-39K | £348-468K | Year 2-3 |
| Mature | 5,000 | £145-195K | £1.7-2.3M | Year 3-5 |
Xero platform fee impact: At 500 connections, the monthly platform fee is ~£125. That's 0.8% of MRR. Negligible.
Customer LTV: At Xero's own 1.03% monthly churn rate and £39/mo ARPU, LTV = £3,786 per customer. At 3:1 LTV:CAC, you can afford £1,262 to acquire each customer. The accountant channel makes acquisition cost closer to £50-100 per org deployed.
What OYN Needs to Be¶
Based on all of the above, OYN should be:
Core Value Proposition¶
"Your AI financial analyst. Plug it into Xero, get the insights your accountant charges £200/hour for."
Must-Have Features (v1)¶
- Financial health score — 1-100, with plain-English explanation
- Automated ratio analysis — Profitability, liquidity, efficiency, leverage
- Trend detection — "Your gross margin has declined 3 months in a row"
- Cash flow forecasting — Simple, based on actuals + patterns
- Plain-language insights — Not charts. Words. "You'll run out of cash in 47 days if receivables don't improve"
- Monthly report generation — One-click PDF/email for accountants to send to clients
Differentiators vs Fathom/Syft/joiin¶
- AI reasoning, not templates — OYN interprets, doesn't just display
- Proactive alerts — "Unusual expense detected" without being asked
- Multi-entity awareness — One accountant, 50 clients, one dashboard
- Natural language queries — "How's cash looking?" → answer
- Sub-£40/month pricing — Undercut Fathom ($39-389) and Spotlight ($49-749)
Technical Architecture Decision¶
The Journals endpoint question. Journals provides the deepest data (every GL entry) but requires Xero Advanced tier (~£735/mo for the customer). Reports API works on cheaper tiers.
Recommendation: Launch on Reports API. This gives you access to the vast majority of Xero users (Starter/Standard/Premium). Add Journals as a premium feature for Advanced-tier users later. Don't gate your addressable market behind a £735/mo customer prerequisite.
Risks to Name¶
| Risk | Severity | Mitigation |
|---|---|---|
| Xero builds native AI analysis | Medium | JAX is focused on automation, not analysis. Xero's track record building their own apps (Shopify integration: 2.96★) is poor. |
| Fathom adds AI | Medium | They'd need to rebuild. 557 reviews of users expecting the current product. Takes 12-18 months minimum. |
| Reddit alpha competitor ships first | Low-Medium | Solo developer, alpha stage. OYN has Ciaran's FD expertise + Steve's engineering. Speed advantage if you move now. |
| API pricing makes unit economics worse | Low | New pricing is flat tiered, not usage-based per call. Webhook-first architecture minimizes API usage. |
| Low adoption / can't get to 50 users | Medium | Accountant channel de-risks this. 5 firms × 10 clients each = 50. Free tier removes friction. |
| AI hallucinations in financial data | High | This is the biggest technical risk. Financial analysis must be accurate. Mitigation: grounded AI (never generate numbers, only interpret actuals from API), human-in-the-loop for accountant review, clear disclaimers. |
Recommended Next Steps¶
This Week¶
- Decide: Reports API vs Journals endpoint — We recommend Reports API for v1
- Register at developer.xero.com — Free Starter tier, 25 connections
- Define the v1 feature set — Financial health score + ratio analysis + trend detection + monthly report
This Month¶
- Build the Xero OAuth + data sync — Pull P&L, Balance Sheet, Aged Receivables, Cash Flow
- Build the AI analysis layer — LLM prompts grounded in actual financial data
- Recruit 5 alpha firms — Ciaran's network, UK accounting firms
Next 90 Days¶
- Beta with 10 paying customers — Core tier, £29/mo
- Submit for Xero App Store certification
- Collect reviews aggressively — First 20 reviews determine your trajectory
- Iterate based on accountant feedback — They know what their clients need
The Bottom Line¶
The Xero ecosystem has 432 apps. 86 in reporting. Zero AI financial analysts.
Users hate their current tools when things get complex. Xero's own AI (JAX) is mocked. The market is explicitly asking for what OYN does.
The economics are exceptional: 0.3% platform fee, £3,786 LTV per customer, accountant-driven distribution.
There's a two-week window where every existing competitor is distracted by a mandatory pricing migration.
And there's exactly one known competitor — a solo developer in alpha.
Build fast. Ship to 5 accountants. Let the data do the talking.
The Team¶
Ciaran O'Donnell — Co-Founder, Finance¶
Award-winning fractional FD with 16 years advising 100+ startups. CFO/Finance Director of the Year (British Accountancy Awards 2019). Founded Virtual FD in 2010. Brings the domain expertise — he is the weekly cash report, and OYN automates his methodology.
Stephen Greszczyszyn — Co-Founder, Technology¶
20+ years at Cisco building AI systems that reduce 3-day analysis tasks to 3-minute reports. Obsessed with turning complex unstructured data into simple insights. Building AI agents that run while you sleep.
Together: the FD who knows what the report should say, and the engineer who can make AI say it accurately at scale.
Report prepared by Friday (AI Chief of Staff) — Data sourced from Xero App Store (432 apps, 16K reviews), Product Ideas (4K ideas), GitHub, Hacker News, Perplexity, and Grok research. Full intelligence database and methodology available on request.