# Expanding business models with AI

Author: Marin Niehues, Niehues Consulting · Published: 2026-06-01 · Updated: 2026-11-16 · AI Strategy & Business Case
Source: https://www.niehuesconsulting.com/blog/expanding-business-models-with-ai/

A company expands its business model with AI when it turns existing data and processes into an offer that customers pay for and then delivers that offer reliably. For most use cases the technology is available; the bottlenecks lie in data, integration and operations. Three questions above all remain for the executive team: which source of revenue should emerge, which unit is accountable for it, and how will its success be measured? This article organizes the options in stages, compares pricing models and describes what an organization must put in place to deliver.

Because new offers frequently rely on customer and machine data, the topic also touches the legal framework. This article provides an overview and does not constitute legal advice.

**A business model grows with AI when offer and accountability fit together and a team delivers the new service every day. The choice of technology follows from this clarification.**

## What stages of expansion are there?

There are three stages of expansion: the improved existing offer, the additional data service and the changed revenue model. Executive teams sort their options by the depth of intervention. The deeper the stage, the more the revenue logic and organization change, and the higher the demands on delivery.

- Stage 1: The existing offer is improved. AI is built into the product or service, for instance as an assistant function in software or as an automated quotation check in the customer portal. The price stays the same or rises moderately; the benefit shows in customer retention and margin.
- Stage 2: Existing data gives rise to an additional service that is sold separately, for instance a forecasting service based on machine data or an analysis for customers who want to benchmark their own operations.
- Stage 3: The revenue model changes. The company charges by usage or by outcome; with outcome-based billing it takes on part of its customers' risk. A well-known pattern is billing by the availability of an asset.

The sensible entry point is Stage 1 or Stage 2. There the existing business remains intact, and the company gains experience with customer response and operating effort. Stage 3 follows once the earlier stages run reliably. Entering it too early commits the company to an outcome its own operations cannot yet guarantee.

The starting position is favorable. According to the German Federal Statistical Office, 26 percent of companies in Germany with ten or more employees used AI in 2025, rising to 57 percent among companies with 250 or more employees. Most of these deployments serve internal efficiency. Bringing AI into the customer offer as well opens a source of revenue that competitors with purely internal use do not have. How this becomes a lasting lead is described in [Competitive advantage from AI](/blog/ai-competitive-advantage/).

## Decision criteria for the selection

Four criteria separate promising expansions from ideas that sound good in a workshop. The first is willingness to pay: are there customers with a dedicated budget for the new service, and can this be tested in conversations before building? The second is data access. A service based on customer data needs available data of sufficient quality and a clarified basis for its use. The third is fit with the core business. A new service can displace revenue from the existing business, and this effect belongs in the calculation from the outset. The fourth is delivery capability: who builds and operates the service, and who supports customers as soon as the first one uses it?

The business units and finance assess the criteria together on a scale of one to five. An expansion enters the business case once it reaches at least four points on willingness to pay and data access; with fewer points, the basis on which a prototype could prove anything is missing. The full process is set out in the articles on [Prioritizing AI use cases](/blog/ai-use-case-prioritisation/) and on the [AI business case](/blog/ai-business-case-benefits-and-costs/).

## How do you price an AI feature?

You price an AI feature as a premium, by usage or by outcome, and the choice follows from willingness to pay and delivery capability. The built-in feature without its own price fits as long as customers will not pay a premium; the separate data service fits once a customer has a budget for it; outcome-based billing fits only when your own operations deliver the outcome measurably and reliably.

**Pricing models for AI features**
Criterion | Premium or add-on module | Usage-based | Outcome-based
Billing | Fixed amount per license or month | Per transaction, query or operating hour | Per outcome achieved, such as availability or savings
Risk for the provider | Low | Medium, cost per use fluctuates | High, part of the outcome risk shifts to the provider
Prerequisite | Clear added value that customers see immediately | Cost control per transaction, including model usage | Measurable outcome that provider and customer assess the same way
Suitable from | Stage 1 | Stage 2 | Stage 3, after reliable operation

With usage-based billing, a close look at your own costs pays off. Every call to a language model incurs running costs, and a flat fee without a usage cap can erode the margin on heavy users. A tiered price with a free allowance and volume pricing limits this risk.

## May a company use customer and machine data for new offers?

A company may use customer and machine data for new offers when a legal basis exists: under the GDPR for personal data, and under the rules of the EU Data Act for data from connected products. For Stage 2 and Stage 3, this clarification is part of the business case.

The EU Data Act governs access to data generated through the use of connected products and related services. Users of such products gain a right of access to this data, and the manufacturer as data holder uses non-personal product data on the basis of a contract with the user. For a machine builder developing a forecasting service from the operating data of its installed equipment, data use therefore belongs in the customer contracts.

As soon as data relates to individuals, such as operator data on a machine or contact data in the customer portal, the GDPR applies as well, with its requirements on legal basis and purpose limitation. Clarify both questions with data protection and legal before the prototype works with real customer data.

## What are examples of AI business models in mid-sized companies?

Typical examples in mid-sized companies are the assistant function in specialist software, the forecasting service based on machine data and billing by availability. The following patterns describe how such expansions are structured.

Assistant function in specialist software. A software provider for trade businesses adds a function to its solution that drafts a quotation from photos and notes. The company starts at Stage 1, measures usage and, after a few months, offers the function as an add-on module.

Forecasting service based on machine data. A machine builder analyses the operating data of its installed equipment and predicts wear. The service is sold as an annual subscription and replaces part of the unplanned call-outs with scheduled maintenance. The data lead from many installations is hard for competitors to replicate.

Billing by availability. The principle is older than AI: aero-engine manufacturers have billed by operating hours for decades. AI makes this model accessible to mid-sized companies, because forecasts make the probability of failure calculable. A compressor manufacturer then sells compressed air per cubic meter instead of the machine.

All three patterns build on data and customer access that the company already has.

**Three routes to a new AI offer**
Criterion | Workshop brainstorming | Technology-driven pilot | With Niehues Consulting
Starting point | Participants' ideas | An available technology | Customer value and business case
Proof of benefit | Estimate | Demonstration in a test environment | Prototype in the real process
Accountability afterwards | Open | With IT | Named product owner with an outcome target
Route into operations | To be clarified later | Handover after the project ends | Transfer with an operations team and rollout

## Organization: who is accountable for the new revenue stream

Every expansion needs an owner with accountability for results. A proven setup is a product lead accountable for revenue and operating costs who, as product owner, also decides on scope. This person needs access to the teams that deliver. Without this assignment, the service drifts back and forth between sales and IT, and nobody decides whether a defect takes priority over a new feature.

Decision paths come on top. The executive team defines who decides on prices and who decides on changes to the scope of service. Sales and service need enablement so that they can explain and sell the service. Delivery management ensures that deadlines and quality remain visible and that dependencies between teams surface early.

Four metrics show the executive team the status of the expansion: the revenue share of the new offer, the usage rate among targeted customers, the margin after operating costs and the lead time from customer request to delivery. They belong in regular reporting, in the same meeting as the figures for the core business.

## Niehues Consulting as a partner for the expansion

[Niehues Consulting](/) in Düsseldorf supports executive teams in exactly this undertaking, from selecting the application to its rollout. Marin Niehues works as a management consultant from the delivery perspective: he builds organizations that deliver and brings AI into operations. His projects range from VC-funded scale-ups to DAX 40 groups, with assignments in 20+ countries on six continents and a home base in the DACH region.

The approach follows the Niehues Delivery Model with the phases Understand, Build and Implement. Three of its test questions matter most for a new offer. In Understand: which business result should change measurably? In Build: is the business case confirmed with real figures? In Implement: which metric shows after 90 days that the solution is landing? The prototype shows, with real users and real data, whether the service is adopted in a customer's process; transfer into operations follows, internally or with external teams. The route into regular operations is described in [From pilot to scale](/blog/from-pilot-to-scale-ai-operations/), and the framework for steering all applications in the article [AI management: what executive leadership must steer](/blog/ai-management-what-executives-must-steer/).

Delivery management and organizational design ensure that roles and decision paths fit the new service. The free initial consultation lasts 30 minutes and clarifies at which stage your company should start. You can arrange it via the [contact form](/#kontakt).

**The route in brief**
Executive teams expand their model in stages, select by four criteria and assign an owner to each expansion. The pricing model follows delivery capability, and data use is clarified before the prototype. The prototype in the real process provides the proof, and transfer into operations turns it into an offer.

**Sources**
- Federal Statistical Office of Germany (Destatis): Use of information and communication technology in enterprises, 2025
- Regulation (EU) 2023/2854 on harmonized rules on fair access to and use of data (Data Act), 2023
- Regulation (EU) 2016/679 (General Data Protection Regulation), 2016

## Frequently asked questions

### How long does it take for an AI offer to generate revenue?

As a planning value, allow three to six months to the first measurable effect for a Stage 1 feature and six to twelve months to the first paying customer for a standalone Stage 2 data service. An outcome-based revenue model requires reliable operations and follows only afterwards. Check after 90 days in operation whether usage is ramping up as planned, and adjust price or scope before scaling.

### How can a company generate new revenue with AI?

A company generates new revenue with AI by selling a service based on its own data or expertise for which customers have a budget. Common forms are add-on modules, forecasting and analytics services and usage-based billing. The starting point is a conversation with customers before anything is built.

### What is a data-driven business model?

A data-driven business model earns revenue from data generated in the company's own operations or through the use of its products. Typical offers are analyses, forecasts and benchmarks for customers. AI makes such models economical because it analyses large volumes of data without effort growing with each customer.

### Who should be accountable for an AI business model?

An AI business model should be owned by a product lead accountable for revenue and operating costs, backed by a sponsor on the executive team. This person needs decision rights over scope and priority and firm access to the delivering teams.

### How do you prevent a new AI offer from cannibalising the core business?

You prevent cannibalisation by quantifying the displaced revenue in the business case beforehand and aiming the new offer at customers or services that the core business does not cover. Where displacement is unavoidable, it goes into the calculation as a cost. If the new offer brings more margin overall than it displaces, it is worth pursuing, because otherwise a competitor will fill the gap.

### How do you recognize the success of an expansion?

You recognize the success of an expansion by metrics that report the new offer separately. Revenue share and usage rate show market acceptance; margin and lead time show the quality of delivery. The trend over several quarters says more than any single month.

### What AI business models are there?

Three AI business models are common: the product with a built-in AI feature, the data service as a separate offer and billing by usage or outcome. The built-in feature fits as long as customers will not pay a premium. The data service fits once a customer has a budget for it, and outcome-based billing only with reliable operations.
