Most AI budget requests fail before they reach a decision. Not because the initiative lacks merit, but because the proposal reads like a technology pitch instead of a business case. The Geschäftsführung does not evaluate AI on technical sophistication. They evaluate it on three things: how much, how long, and what happens if it fails.

After structuring business cases across 25+ DACH engagements, we have identified the pattern that gets approved — and the patterns that stall.

Why most AI proposals fail

The typical AI proposal arrives as a slide deck with a use case description, a reference to "efficiency gains," a vague timeline, and a budget range so wide it communicates uncertainty rather than planning. It reads as: "We think this could work. We need money to find out."

That framing triggers every risk instinct a Geschäftsführer has. It sounds like R&D disguised as a business investment.

The proposals that get approved do something fundamentally different: they frame AI as a process improvement with known inputs, measurable outputs, and a defined payback period. They answer the question the executive is actually asking: "What do I get for this, and when?"

The structure that works

A business case that survives executive review has five sections. Not ten. Not three. Five.

Section 1: The operational problem in numbers. This is not "our customer service could be better." It is: "Our support team processes 800 tickets per week. Average first-response time is 4.2 hours. 60% of tickets follow predictable resolution patterns. Current cost per ticket: €14.30." If you cannot fill in these numbers, you are not ready for a business case — you need discovery first.

Section 2: The proposed intervention. What will the AI system do, specifically? Not "enhance customer service with AI" but "classify incoming tickets by urgency, route to the correct department, and draft initial responses for the predictable 60%." The level of specificity from workflow readiness directly determines the quality of this section.

Section 3: The cost model. Mittelstand AI initiatives typically fall into three budget tiers. A Level 1 single-workflow deployment: €30,000–80,000. A Level 1 deployment with legacy integration work: €60,000–150,000. A multi-workflow Level 2 initiative: €120,000–300,000. Your business case should specify which tier, with line items for engineering, infrastructure, and change management. Do not hide costs — hidden costs discovered later destroy trust faster than high costs disclosed upfront.

Section 4: The payback calculation. This is where most proposals collapse into speculation. The fix: calculate payback using only the operational metrics from Section 1. If the AI system reduces first-response time from 4.2 hours to 0.8 hours for 60% of tickets, and each hour of support costs €42, the weekly savings are calculable. No hand-waving. No "potential revenue uplift." Just arithmetic from measured baselines to projected outputs. See measuring AI ROI for the complete metric framework.

Section 5: The risk mitigation plan. What happens if it does not work? The answer should never be "we lose the entire investment." It should be: "Phase 1 costs €35,000 and delivers a working prototype on real data within 8 weeks. Go/no-go decision at week 8 with maximum €35,000 at risk. Phase 2 only proceeds if Phase 1 metrics meet threshold." Staged funding with kill points is the single most effective pattern for executive approval.

The numbers executives actually want

Three financial metrics close the conversation:

Payback period. Not IRR, not NPV — payback period. How many months until the investment is recovered through operational savings? For most Level 1 Mittelstand deployments, the target is 6–12 months. If your payback period exceeds 18 months, the business case needs rework.

Cost per unit improvement. The Geschäftsführung thinks in operational units. Cost per ticket, cost per invoice, cost per claim. Show the current cost, the projected cost after deployment, and the delta. This is the metric that makes abstract "efficiency gains" concrete.

Risk-adjusted investment. Not the total project cost — the maximum capital at risk before the first kill point. A €120,000 project with a €35,000 Phase 1 gate is a €35,000 decision, not a €120,000 decision. Frame it that way.

Common mistakes

Over-scoping. The business case tries to justify an enterprise-wide AI strategy instead of a single workflow deployment. Start small. The first business case should fund one initiative that proves the model.

Technical language. If the proposal mentions "fine-tuning," "RAG architectures," or "vector databases," it is a technical proposal, not a business case. Translate everything into process and financial terms.

Missing the comparison. Every AI business case competes against "do nothing." Explicitly calculate the cost of inaction — the ongoing operational cost, the compounding inefficiency, the competitive exposure. See the cost of AI inaction for the full framework.

No sponsor alignment. The business case is written but no named executive has committed to champion it. Without decision authority, even the best business case dies in review cycles. Secure the sponsor before writing the document.

The approval pattern

In the Mittelstand, the pattern that works is remarkably consistent: a named Geschäftsführer or Bereichsleiter champions a single-workflow initiative with a Phase 1 budget under €50,000 and a clear 8-week decision gate. The business case fits on three pages. The payback calculation uses existing operational data. The risk framing is honest.

That is not a sophisticated framework. It is a clear proposal from someone who understands that executives approve investments they can evaluate, not technologies they have to trust.

Build the business case around the process, not the technology. The technology is a means. The process improvement is the investment.