How can the AI value contribution be measured effectively?

 

The market for artificial intelligence (AI) is growing unabated. According to our Metrics survey on the IT Agenda 2026, AI expenditure as a percentage of the IT budget is set to rise to over 5.3 per cent in 2026, 80 per cent more than in the previous year. However, this uncontrolled proliferation poses a structural problem: many AI projects emerge in a decentralised and unplanned manner, initiated by individual departments and often lacking overarching management, coordinated prioritisation, a shared vision and, above all, any measurement of benefits.

How is the value contribution of AI determined?

While AI budgets are being spent and projects are ramping up, the question of AI's actual value contribution in practice often remains unanswered. Why is this, and how can it be changed? Our established methodology for measuring IT value contribution in client projects distinguishes four dimensions. These form the conceptual framework for a customised key performance indicator framework that can be applied to a company’s AI value contribution:

1. AI increases effectiveness in business units

IT systems support business processes, enabling services to be delivered with greater quality and customer benefit. The value contribution primarily manifests itself in improved market and customer orientation. In the context of AI initiatives, this effect can be measured using the conversion rate, i.e. the proportion of customer contacts that lead to a successful transaction.

2. AI increases efficiency in business units

IT-supported optimisation of operational processes allows services to be delivered with the same or a higher level of quality while using fewer resources. This is reflected directly in lower service delivery costs. Typical KPIs include average processing time per transaction and degree of automation (i.e. the proportion of process steps that do not require manual intervention).

3. AI invreases efficiency in IT

Automation, consolidation and standardisation in IT operations make it possible to deliver equivalent-quality services at a lower cost. The value contribution here affects IT's own cost base. Relevant metrics include the amount of IT effort (in hours or full-time equivalents) required for a defined service request, as well as the proportion of incidents resolved through AI automation.

4. Other positive AI effects

Some IT systems support strategic business objectives without a direct monetary link, such as improving employee satisfaction, ensuring information security, and promoting regulatory compliance. Suitable KPIs include the employee satisfaction index regarding the use of AI-supported tools, and the number of security incidents identified and prevented automatically.

 

How can AI metrics be derived in a strategic manner?

AI systems rarely have a one-dimensional impact. For example, AI in customer service can improve the quality of interactions, reduce processing times, and positively impact staff satisfaction. This multidimensionality makes it challenging to measure the value contribution of AI in a structured way. What are the 'right' metrics in each case? There is no universal set of KPIs for AI initiatives, nor would such a set be effective. Instead, each AI initiative requires the deliberate definition of the specific business objectives it addresses and the metrics that reflect these objectives in a way that is relevant to management.

When AI is employed to develop the business model, explore new business areas, or differentiate from competitors, traditional efficiency KPIs alone are insufficient as a benchmark. In addition, metrics must be collected to highlight potential conflicts of interest. For instance, an AI solution in the service desk could boost the first-call resolution rate while simultaneously reducing customer satisfaction. It is only through the interplay of primary and accompanying KPIs that a robust and honest picture of the actual value contribution of AI can be obtained.

How to measure the value contribution of AI

We recommend a three-step approach for companies wishing to systematically assess the value contribution of their AI initiatives.

1. Reflect on existing and planned AI initiatives across the four dimensions. The aim is to gain a clear understanding of the desired outcomes and the dimensions to which each initiative contributes, rather than achieving a rigid categorisation. This forms the basis for company-wide, coordinated AI management.

2. Define suitable KPIs derived from the organisation’s business and AI strategy. The objective and impact of each metric are explicitly documented. Primary metrics that directly measure the targeted value contribution should always be combined with supporting metrics that highlight potential conflicts of interest.

3. Use consistent performance measurement in real-world operations. This is where it becomes clear whether the defined target values are being achieved. Here lies the real challenge: AI impacts rarely occur in isolation, but rather in interaction with organisational changes, parallel initiatives, and changing market conditions.

These three stages emphasise the importance of carefully selecting the right KPIs from the outset. Only by clearly defining your goals can you accurately assess whether they have been met.

 

If you want to see the bigger picture, you can run a benchmark comparison to look beyond your own dashboard. This shows which KPIs are currently in use, ranging from model accuracy and latency to error rates and the ROI of individual use cases, and where your own AI initiatives really stand. This enables you to identify blind spots, adjust goals and allocate investments specifically to areas where improvement is needed. If you have any questions about the AI benchmark or the value contribution of AI, please feel free to send me an email.