AI Readiness Assessment: A 5-Step Checklist for DACH SMEs

AI Readiness Assessment: A 5-Step Checklist for DACH SMEs

43% of German mid-sized companies have no concrete AI plans. Meanwhile, 91% of large enterprises already consider AI business-critical. The gap is not about technology — it is about readiness.

A 2025 study by the Wirtschaftsministerium found that German SMEs cite three recurring blockers: unclear business case (58%), missing data foundations (44%), and lack of internal skills (41%). None of these are solved by buying an AI tool.

An AI readiness assessment identifies where your organization stands before procurement decisions are made. Without it, you risk purchasing tools your infrastructure cannot support, processes cannot absorb, or your team cannot operate.

Here is the five-step checklist we use at debored.ai when evaluating whether a DACH SME is ready to deploy AI — and what to fix if they are not.

Step 1: Audit Your Data Foundations

AI models are data consumers. They produce outputs proportional to the quality and structure of the data they ingest.

Start with three questions:

• Where is your operational data stored? Spreadsheets, email attachments, CRM, ERP, shared drives, paper. If more than 30% sits in unstructured or decentralized storage, your data layer needs work before AI can add value.
• Is your data machine-readable? PDF invoices, scanned contracts, handwritten notes — these require OCR preprocessing before any AI pipeline can consume them. Extraction from truly structured formats (databases, APIs, CSV exports) costs 10x less to integrate.
• Do you have data governance? Who owns each dataset? What is the retention policy? How do you handle DSGVO deletion requests? AI pipelines multiply data surface area — if your governance is undocumented, you introduce compliance risk immediately.

Common DACH SME finding: Most companies operate with 5-15 disconnected data sources. Consolidation is the prerequisite, not the AI implementation.

Step 2: Map Your Processes — Not Your Tools

AI adoption fails when organizations automate a broken process. The automation simply produces broken output faster.

Map your three highest-volume operational processes. Document each one as:

• Inputs: what triggers the process (email, form submission, customer call, internal request)
• Steps: the actual sequence of actions, including handoffs between people and systems
• Decision points: where a human currently decides based on context, rules, or discretion
• Outputs: what the process produces (approved invoice, commissioned system, signed contract, support ticket)

For each process, calculate the current cost: hours per cycle × volume per month × loaded hourly rate. This number is the baseline against which any AI investment is measured.

Reality check: If you cannot document a process in under 30 minutes, it is too undefined to automate. Fix the process definition first.

Step 3: Assess Team Readiness and Skills

The team that operates the AI system determines whether it delivers value or becomes shelfware.

Three indicators of readiness:

• Digital literacy baseline: Can your team export a CSV, use a template document, and follow a structured workflow in a shared system? If not, train these fundamentals before introducing AI interfaces.
• Prompt vs. policy: AI tools require clear instruction. Teams accustomed to “figure it out” work styles struggle with the precision AI demands. Teams that follow SOPs and checklists adapt faster.
• Change capacity: How many new tools has your team onboarded in the past 12 months? Companies running 3+ concurrent tool changes see adoption drop below 30%.

Mittelstand pattern: The Geschäftsführer drives AI adoption, but operations depend on 1-2 key employees who are already at capacity. Adding an AI tool to an overstaffed workflow creates resentment, not efficiency. Start with a process that reduces someone’s workload, not adds to it.

Step 4: Evaluate Infrastructure and Security

AI tools consume compute, storage, and network bandwidth — and they introduce new security vectors.

Check these before any deployment:

• Network: Does your internet connection support sustained cloud API usage? Many German SMEs run on consumer-grade connections. A single AI pipeline processing documents can saturate 50 Mbps upload.
• Identity management: Does the tool integrate with your existing auth (Google Workspace, Azure AD)? Standalone AI logins create shadow IT and audit gaps.
• Data residency: Where does the tool process and store your data? If contracts, customer PII, or financial records go through US-based infrastructure, you need a DSGVO-compliant data processing agreement (AVV) in place. Many AI vendors do not provide one standard.
• API costs: Most AI tools charge per-call or per-token. A process running 5,000 operations/month at €0.02/call costs €1,200/month before you see any ROI. Model the cost against your Step 2 baseline.

Step 5: Define Success Before You Buy

The most expensive AI project is the one without a measurable target.

Write the success criteria before evaluating vendors:

• Quantitative: “Reduce invoice processing time from 18 minutes to 4 minutes per invoice within 8 weeks”
• Qualitative: “Customer response time drops below 2 hours during business hours”
• Compliance: “All automated decisions are logged with audit trail for DSGVO and GoBD requirements”

If you cannot articulate what success looks like in measurable terms, you are not ready to buy. You are ready to plan.

Beware the pilot trap: 68% of AI pilots never reach production (Gartner 2025). The reason is almost never technical — it is that the pilot lacked a defined production path from day one. Before starting any pilot, state explicitly: “If this works, here is how it goes into production, who maintains it, and what it costs at scale.”

The Takeaway

AI readiness is not a technology question. It is a process, data, team, and infrastructure question. The 43% of German SMEs without an AI strategy are not behind because they lack the tools — they are behind because they skipped the readiness assessment that would tell them what to fix first.

Start with Step 1 today. Audit your data foundations. Map your three highest-volume processes. Run the readiness checklist against your operations. The output is not a purchase order — it is a build plan.

For companies running Google Workspace, data consolidation usually starts with centralized document management and standardized access controls. From there, we deploy AI operations that integrate with your existing infrastructure — no new tools, no shadow IT, no DSGVO gaps.