Metodo·
Value Scorecard: How to Measure the ROI of AI in Your Firm
A 4-quadrant framework for measuring the return on AI investment. Concrete KPIs, baselines and use cases for professional firms.
4 min
Leggi Caricamento...
Caricamento...
How the AIRA method works in a typical AI adoption path: Assessment, Implementation, Results, Autonomy — activities, timelines and checklists for each phase.
Theory is easy to write. AI adoption in practice requires a structured method, not better tools. The AIRA method — Assessment, Implementation, Results, Autonomy — guides professional firms through a typical 90-day path: from their current situation (often: shadow AI, zero governance, tools used without policy) to a structured, measurable and sustainable adoption.
This article walks through the 4 phases with the typical activities of each phase, indicative timelines and the elements that distinguish an effective path from one that stalls halfway.
Before entering the 4 phases, it is useful to understand where an AIRA path typically begins. The starting point is not a firm that "does not use AI": it is a firm where AI is already in use — informally, without governance, with tools chosen individually by staff members.
In a professional firm of medium size, it is normal to find:
This is not negligence: it is the normal spontaneous evolution of AI adoption. The AIRA method starts from this reality and structures it, without pretending to reset it from zero.
The first phase does not consist of choosing tools. It consists of understanding where you are.
The initial assessment produces a complete inventory and gap analysis. A typical inventory for a medium-sized firm includes tools that management does not know exist: free tools used by individual staff, AI assistants embedded in ERPs without configuration, plugins installed without compliance evaluation.
The gap analysis assigns each identified problem a priority:
Immediate actions (within 7 days) Retirement or replacement of GDPR non-compliant tools. Temporary stop communicated to staff while the policy is being prepared. This part is often the most difficult on a human level: people who use a tool daily perceive it as a constraint. Change management begins at the assessment stage, not the implementation stage.
High priority (within 30 days) Team training, drafting of internal AI policy, preparation of client disclosure.
Medium priority (within 90 days) Optimisation of processes with approved AI, definition of measurement metrics, first review cycle.
The output of the assessment is not a document to be filed. It is the operational work plan for the weeks that follow.
Implementation is divided into two sprints of three weeks each.
Sprint 1 — Governance and training (weeks 4-6)
Before introducing any approved tool, the structure that makes it sustainable is built:
Sprint 2 — Pilot adoption (weeks 7-9)
The selection criterion for pilot use cases is: highest ROI / lowest risk. For HR consulting, tax advisory or legal firms, typical candidates include:
For each pilot use case, three things are defined: the structured reference prompt (developed in guided test sessions), the output review procedure, and the measurement metrics.
The measurement phase only makes sense if the baseline was collected before implementation. Typical AIRA path metrics:
| KPI | How to measure | When |
|---|---|---|
| Hours/week per key process | Activity log already used for billing | 4 weeks pre and post |
| Average client response time | From receiving request to sending response | 4 weeks pre and post |
| Team AI literacy score (1-5) | Anonymous self-assessment | Pre and post implementation |
| AI policy compliance | Monthly checklist | From start of implementation |
| Clients informed of AI use | % of total active clients | Target: 100% by end of Sprint 1 |
An important methodological note: measurements over the first 4-8 weeks include a novelty effect that tends to diminish in subsequent months. Preliminary data should be interpreted as indicative and re-evaluated at 6 months for a stable estimate. The value of the measurement cycle is not a single number — it is the ability to make decisions based on data rather than perception.
The objective of the third month is not to optimise further: it is to make the firm independent from the consultancy engagement.
Typical activities:
At the end of the path, the firm must be able to independently evaluate a new AI tool, train a new staff member on usage rules, and correctly answer a client's question about the use of AI in their services.
The AIRA method works when three conditions are present:
The path does not work when the expectation is that the tool will solve the problem on its own. AI is a capable assistant — it does not replace a working method.
If you are considering launching a structured AI adoption path in your firm, the starting point is the assessment. Explore the AIRA Method or the AI Act Ready service to start with a real diagnosis of your situation. If you would prefer to discuss your firm's specific context before committing to a path, contact us — a 30-minute conversation is usually enough to identify where to start.
AIRA was designed for professional firms (accountants, lawyers, HR consultants, advisors) of medium size: from 2 to 50 people. For larger firms or those with very specific processes, the method adapts in terms of timeline. It is not suited to large organisations requiring enterprise-level digital transformation.
Cost depends on the chosen path. The AI Act Ready service (Phase A of AIRA) starts at €2,000. The full adoption path (Clarity & Value Sprint + Guided Growth) starts at €6,000. We also offer the Fractional AI Officer for ongoing accompaniment from €5,000/month.
The final phase of AIRA (Autonomy) is designed precisely to answer this question: the firm must be able to proceed independently. We deliver updatable policies, documented processes, a 12-month roadmap and train an internal AI lead. The objective is independence, not perpetual dependency on the consultant.
Yes, for simple cases and very small firms (1-2 people with limited AI use) self-management is feasible, using resources such as our free guides. For firms with 5+ people, client-facing processes and compliance obligations, the DIY risk is high: lack of governance, shadow AI, AI Act non-compliance.
Ogni settimana: guide pratiche, novità normative e casi d'uso per studi professionali. Niente spam.
A 4-quadrant framework for measuring the return on AI investment. Concrete KPIs, baselines and use cases for professional firms.