RocaSalvatella · Tech Solutions Manager · Business Case

Hydronia doesn't have a software problem.

It has a channel visibility and product data problem. A €15M digital backbone transformation, framed from first principles.

€450MAnnual Revenue
30%YoY Growth × 3yr
12,000Certified Installers
€15MTransformation Envelope
18moProgramme Timeline
Scroll to explore
01 · Diagnosis

Three structural tensions that drive everything

Before reaching for a tech stack, the problem must be framed in business terms. These three tensions are the actual architectural drivers.

01
Channel Data Ownership
Hydronia knows distributors intimately but is blind to 12,000 installers and nearly all end customers. Distributors treat this opacity as their competitive moat. CMO wants to bypass them. Head of Sales won't allow it. This tension is the architectural spine of the entire programme.
Structural · Political
02
SAP ECC Constraint
ECC is end-of-life but stable. S/4HANA migration is a separate €8M project starting 2027. Everything built in this programme must integrate cleanly with ECC today and survive the migration afterwards. This brutally constrains integration design choices.
Technical · Timeline
03
Capacity vs Ambition
4 FTE internal team. 20-item stakeholder backlog. Only 8–10 are executable in 18 months. 60% capex / 40% opex by Year 2. The sourcing logic has to be realistic, not aspirational — build economics only work where AI-assisted development genuinely changes the calculus.
Organisational · Budget
02 · Target Architecture

Four logical layers. One coherent system.

A layered view of the future-state ecosystem. Each layer has clear ownership, clear interfaces, and is designed to survive the S/4HANA migration.

Layer 04 · Intelligence
Intelligence & AI Layer
iPaaS · Data Platform · IoT Ingestion · AI Agents · Predictive Analytics
MuleSoft / Azure IntegrationMicrosoft FabricRAG / LLM AgentsIoT Cloud Connector
Layer 03 · Channel
Distributor & Installer Digital
Renovated Distributor Portal · Installer Mobile App (MVP) · Warranty Registration · Certification Management
React NativeSSO / Magic LinkReal-time StockMobile-first
Layer 02 · Product
Product Information Management
Single Source of Truth · Energy Labels · F-gas Traceability · CSRD Reporting · Distributor Catalogue Sync
Akeneo GrowthContentservEU RegulatoryMonth 3 Go-live
Layer 01 · Core Transactional
ERP + Commercial Backbone
SAP ECC (untouched) · New CRM as commercial spine · Master Data Management foundation
SAP ECC — FreezeSalesforce / Dynamics 365S/4HANA-safe
Explicit Exclusions — What Is Left Out

S/4HANA migration (separate project, 2027) · Full FSM deployment (Phase 2, post-Month 12) · Direct-to-consumer model · Factory operations digitisation. Calling these out is as important as the inclusions.

03 · Sourcing Logic

Buy, build or reuse — and why.

Criteria applied: TCO over 3 years, time-to-value, S/4HANA survivability, vendor lock-in risk, and the capacity of a 4 FTE team to govern each platform.

ComponentDecisionPlatformRationale
ERPReuseSAP ECC — FreezeProhibitive risk and cost to touch. S/4HANA migration already planned for 2027. Any work here competes with backbone budget.
CRMBuy SaaSSalesforce Sales Cloud / MS Dynamics 365Speed to value, native SAP connectors, ecosystem depth. Decision gate: which iPaaS is chosen — MuleSoft favours Salesforce, Azure Integration Services favours Dynamics.
PIMBuy SaaSAkeneo Growth / ContentservEU regulatory timeline (F-gas, CSRD, energy labels) doesn't wait for a build. Must be in production by Month 3.
Installer AppBuildReact Native + FastAPI + SupabaseNo off-the-shelf fit without deep customisation. AI-assisted development compresses build timeline to 10–12 weeks. IP stays with Hydronia.
iPaaSBuy SaaSMuleSoft / Azure Integration ServicesThe current point-to-point hairball is an existential risk. Chosen platform determines CRM path. Multi-vendor delivery model requires clean API governance.
Data PlatformBuy SaaSMicrosoft Fabric / DatabricksStack follows CRM selection. Fabric is natural if Dynamics 365. Power BI already in use — minimise re-tooling.
04 · Deep Dive

The Installer App: where the real architecture problem lives.

Not ERP. Not CRM. The installer ecosystem is the one non-core application that unlocks the channel visibility problem — which is the core problem.

Why This One
The Channel Visibility Unlock
12,000 installers are currently invisible to Hydronia. They're managed via WhatsApp groups and a low-engagement portal. No off-the-shelf product solves Hydronia's specific combination of certification management, territory logic, distributor relationship mediation, and mobile-first UX at premium brand standards.
  • Most strategically differentiating non-core application
  • No market product fits without deep customisation
  • Build economics improved ~40% with AI-assisted dev
  • CMO, Head of Aftersales, and installers all need it
Build Path
Technical Approach
Scoped build with external team of 4–6 developers, CI/CD from day one, three environments. Zero additional logins: SSO via distributor or magic link email.
  • React Native — iOS + Android from single codebase
  • FastAPI backend — async, typed, documented
  • Supabase/PostgreSQL — RLS-enforced multi-tenancy
  • PIM integration via iPaaS — spec sheets served live
  • Offline-capable for on-site installation context
AI-Assisted Development Impact
With Cursor + Claude Code + Copilot, MVP delivery compresses from 20+ weeks to 10–12 weeks. Build cost falls ~40%. IP stays with Hydronia. This changes the TCO equation fundamentally.
MVP Scope + Success Criteria
What Shipped in Month 12 Looks Like
Spec sheets
On-device in <30 seconds · Offline cache
Warranty reg
At point of install · No paper · Auto-populated
Certification
Digital card · Training history · Expiry alerts
65%
Installer activation target · Month 3 post-launch
70%
Warranty registration rate · From 30% baseline
<30s
Spec sheet retrieval · Literal installer ask
05 · AI Use Cases

Where AI is the product — not just the enabler.

Two use cases with business cases. Both have clear value, defined dependencies, and honest feasibility assessments. Not hand-wavy.

Case 01 · Predictive Maintenance Agent
Know when a heat pump is about to fail — before the customer does.
New heat pump models already ship with cellular IoT modules. That telemetry lands in a vendor cloud that Hydronia doesn't yet integrate. An anomaly detection model on that data stream can alert the assigned installer proactively — before a failure creates a furious customer call 4 business days later.
ValueMTTR from 4 days → <4 hours. Upsell preventive service contracts. NPS improvement across installer and end-customer touchpoints.
DependencyIoT vendor cloud integration via iPaaS (prerequisite). GDPR data flow sign-off with Legal.
Feasibility High — data already exists, integration is the only blocker
Case 02 · AI Spec Assistant (RAG on PIM)
"Which pump for a 180m² house in a cold zone with underfloor heating?"
The installer at a customer's house asks a natural language question — voice or text — and receives the right spec sheet, installation guide, and a pre-filled warranty registration form. All in under 30 seconds. Solved directly by RAG over the PIM catalogue.
Value~40% reduction in technical support calls. Faster close on complex installations. Installer loyalty and NPS lift.
DependencyPIM in production (required as RAG corpus). Structured product data quality. Installer App as delivery channel.
Feasibility Medium-high — PIM is prerequisite, which is already a programme priority
06 · Delivery Model + 90-Day Plan

How the work actually gets delivered.

Tech Solutions Manager as orchestrator. Multi-partner delivery. Monthly SteerCo. First measurable value at Month 6 — non-negotiable.

Weeks 1–4 · Discovery & Foundations
Validate assumptions. Make binding decisions early.
SAP ECC custom code audit with Partner 1. Legal review of distributor contracts for data exclusivity clauses. IoT vendor cloud due diligence. RFP to 3 partners per category. iPaaS and CRM decision by end of Week 4 — this is the critical path gate.
SAP AuditLegal ReviewRFP ×3iPaaS Decision
Weeks 5–8 · Design & Contracts
Sign. Setup. Kickoff.
CRM Phase 1 and PIM data model detailed design. Partner contracts signed. Formal kickoff with SteerCo. Dev/staging/prod environments live. CI/CD pipeline established before a single line of Installer App code is written.
Contracts SignedDev EnvironmentsCI/CD LiveSteerCo Kickoff
Weeks 9–12 · First Value
PIM in production. CRM in staging. Quick win visible.
PIM MVP live — the regulatory clock is running and this unblocks everything else. CRM Phase 1 in staging with distributor pilot group. Installer App wireframes validated with real installers. First IT spend report to CFO.
PIM LiveCRM StagingInstaller ValidationCFO Reporting
Months 4–9 · Channel Layer
Installer App MVP. Distributor portal renovated.
Installer App in production across pilot installer cohort. Distributor portal rebuilt — one login, real-time stock, accurate ETAs. CRM Phase 1 in production. IoT integration scoped and initiated.
App MVP LiveDistributor PortalCRM ProductionIoT Scoping
Months 10–18 · Intelligence & Scale
Data platform, AI use cases, CSRD reporting.
Data platform live with automated reporting. Predictive maintenance agent Phase 1. AI Spec Assistant in Installer App. CSRD and F-gas traceability reporting operational. Programme close and transition to BAU.
Data PlatformPredictive MaintenanceAI Spec AssistantCSRD Reporting
Partner Orchestration Model
Tech Solutions Manager
Programme orchestrator. Architecture governance. SteerCo. Partner management.
Partner 1 · Integration
SAP ECC wrappers. iPaaS setup. ECC connectors. IoT ingestion.
Partner 2 · CRM
Salesforce or Dynamics implementation. PIM if same vendor ecosystem.
Partner 3 · Product Digital
Installer App build. Distributor portal. AI use cases implementation.
07 · Risks & Failure Modes

What could derail this — and who owns it.

Honest risk assessment. The highest-probability risk is political, not technical. That's the one that kills programmes.

Distributor Channel Resistance
Critical
The 150 distributors see installer and end-customer data as their competitive moat. Any perception that the programme is a "bypass strategy" will generate organised resistance. Head of Sales has already signalled this.
Mitigation: Co-design the installer app with the 5 largest distributors from Week 1. Frame as channel amplification, not disintermediation. Formal distributor advisory panel.
SAP Custom Code Blocking Integration
High
Decades of SAP ECC customisation may make clean API exposure impossible without remediation. This is the single most common hidden cost driver in industrial backbone transformations.
Mitigation: Dedicated SAP audit in Weeks 1–4. Early decision: native IDocs via iPaaS adapter vs custom wrapper vs SAP BTP as abstraction layer.
Scope Creep from 20-Item Backlog
High
20 stakeholder asks. At least 5 contradict each other. Only 8–10 are executable in 18 months. Without governance discipline, each SteerCo becomes a scope negotiation.
Mitigation: Locked scope document signed by CIO and COO at kickoff. Change control from Week 1. SteerCo has "scope boundary" as standing agenda item.
iPaaS Vendor Lock-in
Medium
MuleSoft locks to Salesforce ecosystem. Azure Integration Services locks to Microsoft. Both have compounding cost implications. S/4HANA migration will test whichever choice is made.
Mitigation: Open API standards as vendor selection criterion. Architecture review against S/4HANA integration patterns before contract signature.
Key Partner Talent Attrition
Medium
External partners will rotate consultants across an 18-month programme. Without contractual knowledge transfer requirements, institutional knowledge walks out mid-delivery.
Mitigation: Knowledge transfer obligations in partner contracts. 2-week overlap for any key role transition. Architecture documentation as delivery artefact.
Regulatory Deadline Slip
High
EU Energy Efficiency Directive and F-gas regulation deadlines are fixed. CSRD applies from FY25. PIM is the single dependency for all regulatory deliverables. If PIM slips, the programme is non-compliant.
Mitigation: PIM is the highest-priority deliverable. Month 3 go-live is a hard constraint, not a target. Dedicated BA assigned to PIM from Week 1.
08 · KPIs

What success looks like at Month 18.

Five measurable outcomes. Baselines are real. Targets are defensible, not aspirational. If these don't move, the programme didn't work.

Installer App Activation
~5%
↓ baseline
>65%
Month 18 target
Warranty Registration Rate
30%
↓ baseline
>70%
Month 18 target
Product Update Propagation
6 weeks
↓ baseline
<48h
Month 18 target
Aftersales MTTR
4 days
↓ baseline
<1 day
Month 18 target
IT Spend Reporting
Annual
heroic effort
↓ baseline
Monthly
Automated from Month 3