Workshops: AI Automation & Agents

Workshop: How Do You Measure the ROI of AI?

This workshop shows decision-makers and controllers how to measure the value of AI licences and projects robustly instead of guessing. You develop a KPI set, a realistic cost model and a business case that stands up to critical questions.

Who is this workshop for?

  • Executives and CFOs
  • Controlling and finance teams
  • AI, digital and transformation leads
  • IT leaders who need to justify licence spend
  • Owners of running AI pilots

What you will take away

  • You distinguish hard and soft benefits and know how each can be measured.
  • You capture a baseline before rollout so before-and-after comparisons are meaningful.
  • You build a complete cost model including licences, API costs, training and operations.
  • You define usage, quality and impact metrics per use case.
  • You write a business case with assumptions, scenarios and stop criteria.

Agenda

  1. Module 1

    Why AI ROI is hard to measure

    • Time saved is not the same as cost saved
    • Effects spread across many employees
    • Typical mistakes in AI business cases
  2. Module 2

    KPI framework

    • Usage: active users, frequency, use cases
    • Efficiency: cycle time, handling time, volume
    • Quality and impact: error rate, customer satisfaction, revenue
  3. Module 3

    Capturing all costs

    • Licences for ChatGPT, Claude, Copilot or Gemini
    • API and infrastructure costs, development, maintenance
    • Training, change management, governance
  4. Module 4

    Building the business case

    • Exercise: business case for one of your use cases
    • Scenarios: conservative, expected, optimistic
    • Presenting to leadership and supervisory boards

From gut feeling to defensible numbers

Many organisations have rolled out AI licences and sense a benefit but cannot quantify it. At contract renewal or when approving new projects, that gap becomes a problem.

The workshop provides a pragmatic approach without elaborate studies: a few well-chosen metrics, a clean baseline and regular review.

An example: for a Copilot rollout, knowing the number of licences and logins is not enough. More telling is how handling times in selected processes have changed and whether the time gained flows into value-adding work.

Measurement methods at a glance

Different methods suit different use cases, often in combination.

  • Time sampling before and after rollout
  • Comparison groups with and without AI support
  • Process metrics from ticketing, ERP or CRM systems
  • Usage statistics from AI tool admin consoles
  • Structured surveys on quality and acceptance

Data protection in usage measurement

Analysing usage data can allow conclusions about individual employees. We show how to rely on aggregated data, create transparency and involve employee representatives so measurement is not perceived as performance monitoring.

Frequently asked questions

Do you provide benchmark figures for AI productivity?

No. Generic percentages rarely transfer to your organisation. We help you collect your own figures that hold up to scrutiny.

Does this work for AI rollouts already under way?

Yes. If there is no baseline, we show ways to reconstruct one using comparison groups or historical process data.

Is the workshop suitable for individual projects?

Yes. The framework applies to broad licence rollouts as well as single automation or agent projects.

Workshop

Get in touch

Ready to prove the value of your AI investments?

Tell us a little about your team – we will get back to you within 24 hours with a proposal for content, format and dates.

Prefer to talk?

We only use your details to handle this request. See our privacy policy.