Business Case Development

A small efficiency request became a €4.19M board-ready roadmap

€4.19M: total modelled roadmap value
~947%: modelled ROI on ~€500k investment
5-6 weeks: modelled payback after deployment

Six chained nodes reading business goal, capability, feature, roadmap, investment and expected value; each node lights up in sequence. — Fig. 01 — The chain the business case was modelled on: business goal → capability → feature → roadmap → investment → expected value.

Case file

Industry: Construction project management consultancy (Tier 2 Mittelstand)
Region: Germany
Service: Business Case Development
Team: Wayne ArendseDelivery & Transformation LeadYurii KotulaCEO & Engineering Leader
Systems: Three disconnected systems, AHO Heft 9 / VOB / HOAI regulatory frame

Client context

A German construction project management consultancy acting as independent project partner for complex commercial builds.

The work is governed by AHO Heft 9, interlocked with VOB contract law and HOAI fee scales: sequential documentation, phase-gate approvals, financial auditing, and legal verification across more than 300 process steps.

Business problem

Project steering engineers were spending 30-40% of their billable capacity on administration: manual data extraction across three disconnected systems, line-by-line auditing of 50-200 page contractor invoices with a desk calculator, cross-referencing contract clauses for fee disputes, and chasing contractor documents.

The ask that started the engagement was modest: save some time on manual work. The diagnostic found a structural constraint underneath it.

The stakes

Roughly 2,550 manual processing hours per project.
28% of the total annual engineering payroll consumed by administration.
Approximately 44% of additional addressable revenue forfeited, because the firm could not scale without hiring.
Overpayment risk on multi-million-euro invoices, and working-capital drag from reporting delays.

What was broken

More than 300 granular process steps, mapped one by one
The top 20% of tasks caused 80% of the administrative burden
Example task: 30.5 hours cycle time, of which 11.5 active and 19 waiting
Three disconnected systems requiring manual data extraction
Invoice auditing done line-by-line against 50-200 page documents

Diagnostic approach

The diagnostic mapped all 300+ process steps, then modelled where automation, data flow, and AI could remove the constraint without touching legal accountability.

Strategy
What is the opportunity worth?
From a time-saving request to a growth case: the constraint was capping addressable revenue, not just cost.
- ~44% additional addressable revenue forfeited
- Growth blocked by hiring dependency
- EBITDA leverage identified
Operations
Where does the time actually go?
300+ steps mapped; the top 20% of tasks carried 80% of the burden. Waiting time dominated active time.
- ~2,550 manual hours per project
- 30-40% of billable capacity on admin
- Example: 30.5h cycle → 1.5h redesigned, ~15 min active
Technology
What can carry the load safely?
Three engines, designed non-invasively around the regulatory frame, with human sign-off preserved.
- Read-only integrations, no system replacement
- Deterministic auditing before AI judgment
- Locally hosted AI, legally cited outputs

What changed

Engine 1: Data liquidity
Read-only API connections across the three systems, auto-compiling compliance matrices and phase-gate documentation. Non-invasive by design.
Engine 2: Deterministic financial auditing
OCR plus logic: line-item extraction from invoices and bids, baseline comparison, and automatic flagging of discrepancies. Deterministic where money is counted.
Engine 3: Cognitive compliance shield
A locally hosted, retrieval-augmented AI trained on contract law, fee scales, and historical claims, producing legally cited draft responses. No AI output is ever legally committed without human sign-off; only Project Leaders sign.

Decisions

Non-invasive integration
Read-only connections instead of system replacement. The regulatory workflow stays intact; the manual burden does not.
Determinism where money moves
Invoice auditing runs on OCR and explicit logic, not model judgment. AI drafts; humans sign.
Local AI, cited outputs
A locally hosted vector store over contract law and fee scales, so every draft response carries its legal citation.
Self-funding sequence
~€500k investment (~€440k CapEx) phased over 44 weeks, MVP scoped to the 60 highest-yielding steps, modelled ~€1.4M of value in the first 4 months.

Results

Measure	Before	After
Admin share of billable capacity	30-40%	Redesigned workflows, example task 30.5h → 1.5h
Concurrent project throughput per PM	Baseline	+~67%, same person, same stress level
EBITDA margin (modelled)	~18.9%	~42.6%
Investment case	A vague time-saving idea	€4.19M roadmap, ~947% ROI, payback in 5-6 weeks

Strike Team

Senior engineering leaders find the real constraint in a fixed-fee engineering diagnostic, then stay to fix it. Project rescue, fractional CTO, architecture and business cases.

About the Strike Team