Build Estimate — Housing Association Business Planning Platform¶
February 2026
The Core Thesis¶
Building a modern replacement for MRI HousingBrixx is a domain-complex but technically achievable project. The critical insight is this:
The financial modelling domain expertise — how housing associations model their 30-year plans, the accounting treatments, regulatory rules, covenant definitions, and stress test methodologies — is the most valuable and hardest-to-replicate input. The software engineering is the commodity part.
If the domain expertise comes from an advisor who already knows how every HA in the country builds their business plan, the build becomes a software engineering exercise around well-understood requirements. This dramatically reduces cost, risk, and time-to-market compared to a team that must learn the domain from scratch.
What Makes This Different from a Typical SaaS Build¶
This is not a CRUD application. The core challenges are:
- 30-year integrated 3-way engine correctness — P&L ↔ Balance Sheet ↔ Cash Flow with monthly/annual granularity over 360 periods, all deterministically linked
- Regulatory compliance outputs — FFR template with 132 financial lines + 160 assumption lines, SORP/FRS 102 conventions, RSH viability expectations
- Treasury/covenant logic under stress — configurable covenant definitions (varying by lender), headroom analysis, breach detection across multiple simultaneous scenarios
- Trust parity with incumbent — housing association treasurers will verify every number against their existing models. One rounding error = lost credibility
However, with the domain expert providing the financial logic, formulas, and accounting rules, the engineering team can focus on what they do best: building reliable, scalable, well-designed software.
Architecture Overview¶
┌─────────────────────────────────────────────────────────────┐
│ Client Applications │
│ Modern Web App (React/Next.js) │ Board/Read-Only Portal │
└──────────────────┬──────────────────────────────┬────────────┘
│ │
▼ ▼
┌──────────────────┐ ┌──────────────────────┐
│ API Layer │ │ Auth & Access │
│ (REST/GraphQL) │ │ (SSO, RBAC, Audit) │
└────────┬─────────┘ └──────────┬───────────┘
│ │
├────────────────────────────────┘
▼
┌──────────────────────────────────────────────────────┐
│ Core Financial Engine │
│ • Component compiler (building blocks) │
│ • Time-series calculation graph (360 months) │
│ • 3-way linkage (P&L, BS, CF) │
│ • Covenant monitoring & breach detection │
│ • Scenario branching & comparison │
└──────┬────────────────────────────┬──────────────────┘
│ │
▼ ▼
┌────────────────────────┐ ┌──────────────────────────────┐
│ Regulatory Reporting │ │ Specialist Engines │
│ FFR export, SORP maps, │ │ Treasury/loans, development │
│ VfM metrics, board packs│ │ appraisals, asset lifecycle │
└──────────┬─────────────┘ └──────────────┬───────────────┘
│ │
└────────────┬───────────────────┘
▼
┌──────────────────────────────────┐
│ Data Layer │
│ PostgreSQL │ Object Storage │
│ Redis/Queue │ Optional Warehouse │
└──────────────────────────────────┘
│
▼
┌──────────────────────┐
│ AI Layer (Phase 2+) │
│ Narratives, anomaly │
│ detection, drafting │
└──────────────────────┘
Key Technical Decisions¶
| Decision | Recommendation | Rationale |
|---|---|---|
| Calculation engine | Bespoke, deterministic | Core domain moat; must be fully auditable and traceable. Cannot rely on third-party engine for regulatory-grade outputs |
| Frontend framework | React/Next.js | Large talent pool, strong ecosystem, server-side rendering for performance |
| Backend | Python (FastAPI) or TypeScript (Node) | Either works; choose based on team strengths. Python preferred for AI integration |
| Database | PostgreSQL | Reliable, well-understood, excellent for financial data; managed cloud options available |
| Hosting | AWS or Azure UK regions | UK data residency required for sector procurement; ISO 27001 achievable |
| Authentication | Auth0 or similar managed service | Buy, don't build. SSO/SAML support needed for enterprise clients |
| FFR export | Custom Excel generation (openpyxl or similar) | Must match NROSH+ Bulk Import Template exactly |
| Charting/dashboards | Modern charting library (e.g., Recharts, D3) | Board-quality visual output is a differentiator vs incumbent |
| AI/LLM | API-based (GPT-4/Claude) | Assistive only; never authoritative for compliant numbers |
Build vs Buy Components¶
| Component | Decision | Notes |
|---|---|---|
| Calculation engine | Build | Core IP and domain moat |
| Component framework | Build | Sector-specific building blocks |
| FFR export engine | Build | Must exactly match RSH template |
| Scenario manager | Build | Core differentiator |
| Authentication | Buy (Auth0, Clerk) | Commodity |
| Email/notifications | Buy (SendGrid, Postmark) | Commodity |
| File storage | Buy (S3/Azure Blob) | Commodity |
| PDF generation | Buy (Puppeteer, WeasyPrint) | Commodity |
| Hosting/infrastructure | Buy (AWS/Azure managed services) | Commodity |
| Monitoring/logging | Buy (Datadog, Sentry) | Commodity |
| AI/LLM capabilities | Buy (API access) | Use best-available models |
Modules¶
| Module | Complexity | Description |
|---|---|---|
| Core Financial Engine | Very High | 30-year 3-way integrated model, dependency graph, monthly/annual rollups, audit trace per number |
| Component System | High | ~100+ building block templates (tenures, loans, costs, development), formula-locked outputs |
| Scenario Manager | High | Copy-on-write branching, assumption overrides, multi-scenario comparison |
| Loan & Treasury | Very High | Multiple loan types, covenant calculation (configurable per lender), headroom analysis, breach detection |
| Regulatory Reporting | Very High | FFR export (exact template match), VfM metrics, SORP-aligned outputs, validation checks |
| Consolidation | High | Multi-entity consolidation with intercompany eliminations |
| Development Appraisal | High | Scheme-level modelling: land, build costs, tenure mix, grant, cross-subsidy |
| Asset Lifecycle | High | Component accounting, stock condition, decent homes, building safety, decarbonisation |
| Dashboard & Reporting | Medium-High | Board-ready dashboards, custom reports, PDF/Excel export, branding |
| Data Import/Export | High | Excel/CSV import, HMS/finance system connectors, migration tooling |
| User Management | Medium | Multi-tenant, RBAC, audit trail, SSO |
| AI Layer | Medium | Scenario narratives, anomaly detection, board paper drafting |
| Frontend Application | Very High | Planner UX, interactive modelling, scenario comparison, responsive design |
Phased Build Plan¶
Phase 1 — Lean MVP (6–9 months)¶
Goal: Deliver a credible, auditable alternative for 3–5 pilot housing associations, with the advisor providing hands-on domain guidance throughout.
Team: Advisor (part-time domain lead) + 2–3 senior developers + 1 designer (part-time)
Includes:
| Capability | Detail |
|---|---|
| Core 3-way engine | Monthly + annual rollups, 30-year horizon |
| Initial component library | Income (by tenure), operating costs, capital programme, loans, equity |
| Scenario manager | Base case + 3–5 stress scenarios |
| Loan/covenant essentials | ICR, gearing, liquidity headroom (configurable definitions) |
| FFR export v1 | SOCI, SOFP, cash flow — ready for NROSH+ upload |
| Basic reporting | Board summary, covenant dashboard, VfM metrics |
| Import/export | Excel/CSV templates for data onboarding |
| User management | Tenant/user roles + full audit trail |
| Hosting | Cloud-native deployment on AWS/Azure UK |
Defers: Deep connectors, advanced consolidation, development appraisals, AI features, asset lifecycle depth, benchmarking.
Target cost: £100,000–£200,000
This assumes:
- 2–3 senior developers at £500–£800/day (mix of employed and contract)
- Part-time designer and part-time product/domain input from the advisor
- AI-assisted development (Copilot, Claude, Cursor) accelerating code output by 30–50%
- Modern frameworks reducing boilerplate dramatically vs 10 years ago
- Cloud infrastructure costs minimal at pilot scale (<£500/month)
- The advisor provides the financial modelling logic, formulas, and test data — this is the single biggest cost reduction vs a team learning the domain from scratch
Key milestone: Successfully produce an FFR-compliant output that reconciles against a pilot client's existing HousingBrixx output.
Phase 2 — Production Platform (+6–9 months)¶
Goal: Harden for 20–50 clients with reduced advisor involvement per implementation.
Team: +2–3 additional engineers, +1 QA, +0.5 SRE
Adds:
| Capability | Detail |
|---|---|
| Asset lifecycle module | Decent homes, building safety programmes, EPC-driven decarbonisation |
| Consolidation v1 | Multi-entity groups with intercompany eliminations |
| Development appraisal | Scheme pipeline, grant assumptions, tenure mix, cross-subsidy |
| Richer dashboards | Custom reports, PDF board packs, branded output |
| Integration layer | Xero, selected HMS/GL imports |
| Migration toolkit | Import adapters for HousingBrixx and common Excel model patterns |
| Compliance hardening | Regression test harness against known FFR examples |
| SRE maturity | SLAs, disaster recovery, monitoring, support workflows |
Target cost: £200,000–£400,000 additional
Key milestone: Three clients successfully submit their FFR using the platform without parallel running against HousingBrixx.
Phase 3 — Scale & Differentiation (+12 months)¶
Goal: 100–200+ clients, category leadership, defensible moat.
Team: +3–5 additional engineers, +1 ML engineer, +1 data engineer
Adds:
| Capability | Detail |
|---|---|
| AI copilot | Scenario narrative generation, anomaly detection, board paper drafting |
| Benchmarking | Anonymised sector benchmarking across client base (opt-in) |
| Advanced treasury | Hedging instruments, refinancing optimisation, debt strategy modelling |
| API ecosystem | Open APIs, partner integrations, connector marketplace |
| Performance at scale | Large portfolios (50K+ units), concurrent scenario runs |
| Advisor operating layer | Multi-client command centre for advisory firms |
| International expansion | Scottish/Welsh/NI regulatory variants; potentially Irish/Australian markets |
Target cost: £400,000–£1,000,000 additional
Total Investment Summary¶
| Phase | Duration | Team Size | Cost Range | Cumulative |
|---|---|---|---|---|
| Phase 1 — Lean MVP | 6–9 months | 3–4 people | £100K–£200K | £100K–£200K |
| Phase 2 — Production | +6–9 months | 6–8 people | £200K–£400K | £300K–£600K |
| Phase 3 — Scale | +12 months | 10–15 people | £400K–£1M | £700K–£1.6M |
How This Compares¶
| Approach | Estimated Cost | Time to Market |
|---|---|---|
| Traditional enterprise build (no domain partner) | £3M–£8M | 30–42 months |
| Advisor-led build (domain expertise provided) | £700K–£1.6M | 24–30 months |
| Lean MVP only (prove concept, iterate) | £100K–£200K | 6–9 months |
The 3–5× cost reduction comes from three factors:
- Domain expertise is provided, not learned — the advisor knows exactly what the financial engine needs to calculate and why. No months of discovery, no expensive regulatory consultants, no getting it wrong and rebuilding
- Modern development tools — AI-assisted coding, modern frameworks, cloud-native infrastructure, managed services. A senior developer in 2026 produces 3–5× the output of a developer in 2015
- Focused scope — building only what housing associations need, not a general-purpose FP&A platform. Every feature serves the specific use case
What AI Can Automate¶
AI capabilities represent a genuine differentiator, not a gimmick. Housing association finance teams spend enormous time on tasks that AI can accelerate:
| Capability | Value | Phase |
|---|---|---|
| Scenario narrative generation | Auto-generate plain-English explanations of what each stress scenario means and why it matters — saving hours of board paper writing | Phase 2–3 |
| Anomaly detection | Flag unusual assumptions or data patterns (e.g., "your void rate assumption is 2× the sector average for your region") | Phase 2 |
| Board paper drafting | Generate first-draft business plan commentary from model outputs — the advisor reviews and refines rather than writing from scratch | Phase 3 |
| Natural language querying | "What's our peak debt position in the combined stress scenario?" — instant answer from the model | Phase 2 |
| Automated stress test design | AI generates plausible stress scenarios based on current economic conditions and RSH guidance | Phase 3 |
| Peer benchmarking insights | "Your cost per unit is 15% above the sector median for associations of your size — here's what's driving it" | Phase 3 |
| Regulatory change impact | When RSH updates the FFR template, AI helps map changes to existing models | Phase 3 |
Important Principle
AI is assistive only — it never produces authoritative numbers for regulatory submissions. All financial calculations remain deterministic and fully auditable. AI helps with interpretation, narrative, and productivity, not arithmetic.
The Advisor's Role in the Build¶
The advisor is not a traditional "subject matter expert" hired for workshops. They are the domain co-founder — the person whose expertise makes the entire venture viable.
What the Advisor Provides¶
| Input | Why It's Critical | Alternative Cost |
|---|---|---|
| Chart of accounts mapping | Which building blocks map to which FFR lines, SORP categories, and covenant calculations | 3–6 months of regulatory research + trial and error |
| Financial modelling logic | How rent, costs, development, treasury, and depreciation interact over 30 years | Would require hiring 2–3 specialist consultants at £1K+/day |
| Test data and validation | Real (anonymised) business plans to test against — the only way to prove the engine works | Cannot be synthesised; must come from real HA experience |
| Regulatory interpretation | What the RSH actually expects vs what the rules literally say | Years of sector experience |
| Covenant definitions | How different lenders define ICR, gearing, asset cover — and the edge cases | Proprietary knowledge from lender negotiations |
| Client relationships | Warm introductions to 3–5 pilot clients willing to test | 6–12 month cold sales cycle otherwise |
| Credibility | A trusted advisor recommending the tool carries more weight than any marketing campaign | Priceless |
What the Advisor Doesn't Need to Do¶
- Write code
- Design interfaces
- Manage cloud infrastructure
- Handle customer support at scale
- Build the sales and marketing machine (beyond initial pilots)
The advisor provides the what and why. The engineering team provides the how.
Risks and Mitigations¶
| Risk | Likelihood | Impact | Mitigation |
|---|---|---|---|
| Financial engine inaccuracy | Medium | Critical | Golden test packs from real HA models; parallel running against HousingBrixx; formal validation with sign-off before each pilot goes live |
| FFR compliance gaps | Medium | High | Embed regulatory SME (the advisor) in product loop; regression test harness against known FFR submissions; annual update process |
| Data migration friction | High | High | Import adapters for common Excel patterns and HousingBrixx exports; assisted reconciliation (difference checker by statement/period); dual-run period tooling |
| Long sales cycles | High | Medium | Advisor-led distribution bypasses procurement; start with 3–5 lighthouse pilots from advisor's portfolio; publish quantified outcomes |
| Incumbent lock-in | High | Medium | Time market entry to coincide with SORP 2026 / rent settlement model refreshes; position as "run alongside" not "rip and replace" initially |
| Regulatory changes | Ongoing | Medium | Annual FFR template mapping updates; advisor monitors regulatory pipeline; flexible engine designed for rule changes |
| Advisor availability | Medium | High | Formalise advisor commitment early; document domain knowledge systematically; build internal capability over time |
| Security/procurement barriers | Medium | Medium | ISO 27001 certification path from day one; UK data residency; standard HA procurement questionnaire readiness |
| Performance at scale | Low (Phase 1) | High (Phase 3) | Architecture designed for horizontal scaling; performance testing with large portfolios from Phase 2 |
Ongoing Costs (Post-Launch)¶
| Category | Monthly Range | Notes |
|---|---|---|
| Cloud infrastructure | £500–£5,000 | Scales with client count; managed services reduce ops overhead |
| Support & customer success | £5,000–£15,000 | Part-time initially, growing with client base |
| Continuous development | £10,000–£30,000 | Ongoing feature development and regulatory updates |
| Annual FFR template update | £5,000–£10,000 | One-off annual exercise (typically March, for June deadline) |
| Security & compliance | £2,000–£5,000 | Penetration testing, ISO 27001 maintenance |
| Total monthly burn | £17,500–£65,000 | Scales with maturity and client count |
Revenue Required for Sustainability¶
At an average annual contract value of £15,000–£20,000:
- Break-even on ongoing costs: 15–50 clients
- Break-even on total investment (Phase 1–2): 25–60 clients
- Profitable at scale: 100+ clients
Given the TAM of ~227 core buyers and the advisor's existing relationships, reaching 50 clients within 3 years of launch is a realistic target.
Timeline to Market¶
| Milestone | Target Date | Assumption |
|---|---|---|
| Partnership agreement signed | Q2 2026 | — |
| Domain blueprint + chart of accounts mapping | Q2–Q3 2026 | Advisor-led, 6–8 weeks |
| Core engine + component framework + first UI | Q3–Q4 2026 | 2–3 senior devs |
| Scenarios + covenant monitoring + FFR v1 | Q4 2026–Q1 2027 | — |
| Pilot onboarding (3–5 clients) | Q1 2027 | From advisor's portfolio |
| Parallel testing against 2027 FFR cycle | Q1–Q2 2027 | June 2027 FFR deadline |
| Production hardening + Phase 2 features | Q3 2027–Q1 2028 | Growing team |
| Scale to 20–50 clients | 2028 | Advisor-led distribution |
| AI features + benchmarking + 100+ clients | 2029 | Category leadership |
Summary¶
A modern replacement for MRI HousingBrixx is technically feasible, commercially viable, and strategically well-timed. The key enabler is the advisor partnership:
- Without domain expertise: This is a £3M–£8M, 3-year enterprise software project with high regulatory risk and uncertain market fit
- With domain expertise: This is a £100K–£200K MVP that can be in pilot within 9 months, scaling to a £700K–£1.6M platform serving 100+ clients within 3 years
The software engineering is genuinely the commodity input. Modern tools, cloud infrastructure, and AI-assisted development mean that a small team of senior engineers can build in months what would have taken years a decade ago.
The domain expertise — knowing exactly what 227 housing associations need their 30-year business plan to produce, how every covenant is calculated, what the regulator actually expects, and having the relationships to get pilots through the door — that is the irreplaceable asset.
This document is part of a four-part research package. See also: Executive Summary · Product & Competitive Intelligence · Regulatory Requirements