AI Value: A 4-Step Guide to Business Impact

Step 1: Ideate – Identify the Value Drivers

The first step toward generating real value with AI is understanding which parts of your business process actually drive that value. In other words: what are the key levers that influence the outcome of the process you want to improve?

To get there, it’s helpful to ask a few focused questions: Which factors have the greatest impact on performance? How frequently do they occur? Can AI realistically influence them? And most importantly – where could AI make a difference that truly adds value?

Strong value drivers usually share three characteristics: they show up often, they significantly impact either costs or revenue, and they can be meaningfully influenced by AI. If all three apply, you’re onto a high-potential lever – and well on your way to building a business case that holds up.

A simplified example: A speciality chemicals manufacturer identifies the optimisation of maintenance for its complex, failure-prone machinery as a key strategic value lever. An in-depth analysis and discussions with plant management reveal a clear value driver: the number of maintenance needs detected early.

Why? Because the earlier potential defects are identified, the more targeted the deployment of technicians can be to prevent breakdowns. The result: fewer unplanned production stoppages – and with them, a significant reduction in revenue losses caused by downtime.

Step 2: Relate – Setting up the Equation

Once the key value drivers of a business process have been identified, the next step is to link these with a possible AI solution in a meaningful way. To do this, each driver is scrutinized in detail - this requires both creativity and close collaboration between specialist and tech expertise:

What specific output from an AI solution (e.g. an automatically generated error report or an intelligent product recommendation) really influences the respective value driver? What type of AI solution would be best suited to deliver precisely this output?

These considerations are crucial for making the potential impact of an AI solution both realistic and tangible — and they serve as the first real test for any AI initiative. They reveal whether the project is truly worth pursuing — or better left on the shelf.

In the example of the speciality chemicals manufacturer, the AI solution, value drivers and financial benefits were linked by the following value hypothesis: New machines are to be equipped with heat and vibration sensors. An AI Agent - a kind of virtual maintenance manager working around the clock - could then automatically recognize anomalies in the sensor data, derive maintenance requirements and directly trigger an order to a technician. This would prevent impending breakdowns at an early stage.

The result: fewer production interruptions, less lost sales - and therefore a measurable contribution to business success.

This provides the basic arithmetic framework for the business case: avoided loss of sales = (avoided loss of sales per recognized maintenance requirement) × (recognized maintenance requirements). Beyond this simplified form, further earnings effects were included (e.g. reduced emergency maintenance costs, reduced scrap, reduced depreciation from total losses).

Step 3: Estimate – Filling in the Numbers

Once the basic framework of the value equation is in place, the third step is to substantiate the previously rather qualitatively formulated assumptions with figures - in other words, to quantify the expected added value.

This step involves taking a close look at each value driver, following the principle: “Facts first.” Hard data takes priority — such as measured values like process costs, cycle times, or insights from previous feasibility studies. Where concrete data isn’t available, well-grounded expert estimates or benchmarks from similar projects can fill the gap.

This step is essential to make the added value of an AI project tangible for the first time - in figures that can be verified and communicated convincingly.

Back to the example of the chemical company: for the planned AI agent in maintenance, the previously defined value drivers were quantified based on empirical values from similar projects and assessments by the maintenance team.

The underlying assumption: the "AI maintenance manager" could boost early detection of maintenance needs by 16–22%. At present, downtime accounts for 4% of total production time, resulting in annual revenue losses of approximately €17.04 million.

In the most realistic scenario (‘middle case’), the team assumes that an increase of 19% in recognized maintenance requirements would reduce the downtime rate to 3.24% - which would reduce the loss of revenue to around €13.81 million.

The result: the AI Agent would secure around €3.24 million in additional annual revenue through avoided production downtime alone - and thus deliver clearly quantifiable added value.

Step 4: Substantiate – Making the Equation Waterproof

The final step is to substantiate all previous figures, assumptions and estimates and make them reliable. This is primarily done by critically reviewing each variable in the value equation by specialised and technical experts, as well as cross-comparisons with the results of similar projects. Empirical values are critical here. This step is crucial to establish credibility for the business case through transparency about data sources and assumptions, and to make the case justifiable to decision-making bodies. Experience has shown that validation by a team of internal and external experts creates the greatest trust.

In the chemical company example, every figure and variable was backed by clear data sources, assumptions, and scenario analyses — all with the goal of strengthening the credibility and robustness of the value equation. The key assumptions here were that similar improvements could be achieved as in comparative projects, that all recognized maintenance requirements would be translated into timely and successful maintenance, and that the AI agent would not generate any false positives (i.e. superfluous maintenance requirements). It was crucial to clearly document the impact of each individual assumption on the overall added value. This way, if an assumption is challenged—such as during a business case discussion—it’s possible to pinpoint exactly how much it influences the outcome. The key advantage: the overall value of the AI project isn’t immediately put into doubt. Instead, uncertainties can be assessed in a targeted, transparent way and evaluated separately.

Conclusion: Unlocking AI’s Value with the Value Equation

Our experience shows: In practice, simplicity and clarity are particularly convincing when it comes to communicating the added value of AI. The four steps of our ‘Value Equation’ were developed with precisely this guiding principle in mind.

Of course, there are more complex — and potentially more precise — methods out there. But the four steps outlined above are essential. They form the foundation of any solid AI business case.

What matters in practice is a balanced, pragmatic approach to implementation: Where can reliable estimates and benchmark data be sourced? What role does the project’s time horizon play? How should implementation costs be factored into the value calculation? And just as importantly: how can the projected value be strategically positioned within the internal political landscape? Answering these questions is key to turning a good idea into a convincing case for investment. They reveal why every business case needs to be customised. This is the only way to create robust, convincing arguments - and thus a real basis for making a decision to invest in AI.