AI Strategy & Business Case

Choosing an AI Consultancy: What Decision-Makers Should Look For

Marin Niehues2 July 202610 min read

Two businessmen discussing figures on a laptop

The right AI consultancy is the one that can show its results working in live operations and takes responsibility for getting them there. This criterion decides the choice because AI initiatives fail more often at implementation and unclear ownership than at the technology. This article explains what an AI consultancy does, sets out five evaluation criteria and the warning signs of weak providers, and describes a selection process that lets the executive team reach a well-founded decision.

What does an AI consultancy actually do?

An AI consultancy helps companies find the applications of artificial intelligence with the greatest business value, test them as prototypes in the real workflow and bring them into operation. Its work covers four fields. It analyses workflows and builds a business case. It sets the selection and sequence of applications. It builds prototypes and transfers them into live operation with internal or external teams. And it ensures that business units adopt the solution and that the portfolio is steered through metrics.

The market is broad. Strategy consultancies design target pictures, technology firms build solutions, and software vendors combine advice with their own products. Each model has its place. For the executive team, the practical question is which model fits the result the company wants to achieve.

The answer starts with the goal. An AI initiative succeeds when a specific workflow performs measurably better and the people who carry it out actually use the new tool. This result arises at the interface between the business unit and IT and requires decisions from management. This yields a simple selection principle. Assess consultancies by what is running in operation after the project, and give less weight to the list of methods on offer.

What should you look for in an AI consultancy?

Look for five criteria in an AI consultancy, each of which can be verified in conversation and through references.

  • Business focus: The consultancy starts with business analysis and a robust business case. Only then is the application and technology decided. Starting with the product or the model answers the question of value too late.
  • Proof in the real workflow: A prototype demonstrates value in the daily work of the business units. A demo environment provides no such evidence.
  • Transfer into operation: The consultancy describes how the prototype becomes a stable product with internal or external teams, one that covers operation and maintenance.
  • Implementation and enablement: Workflows and responsibilities are part of the scope, as is enabling users. Adoption by the business units completes the engagement.
  • Steering through metrics: The consultancy manages the AI portfolio through metrics and makes each application's contribution to business results visible.

Each criterion can be tested in a single conversation. Ask for an example in which a prototype was transferred into live operation, and ask to see the metrics used to measure success. Providers with delivery experience answer concretely and name the roles involved. Providers without it stay with methods and tools. Why many prototypes never get past the pilot stage is explained in the article Why AI pilots do not reach operation.

How do you spot a dubious AI consultant?

You can spot a dubious AI consultant by the fact that they promise results before they know your workflows and data. The market is young, the job title is not protected, and many providers emerged only with the spread of generative AI. Six warning signs separate weak providers from robust ones.

Two of these signals are enough to remove a provider from the shortlist. The reason is the asymmetry of costs. A rejected consultancy costs one meeting, while a wrong choice costs months and the business units' trust in the next initiative. Because legal questions are touched on here: This article provides an overview and does not constitute legal advice.

Do I need an AI consultancy or is a software vendor enough?

A software vendor is enough when a standard product covers a clearly defined workflow and your team steers the implementation itself. A consultancy pays off as soon as several units are affected, the business case is open or prototype and operation should be in one pair of hands.

The differences between the common models become visible when they are set side by side along the five criteria.

For initiatives intended to reach operation, the comparison leads to a clear recommendation. Every unmanaged handover between providers costs time and knowledge. The initiative therefore needs one party that owns the path from analysis to implementation end to end. Whether a large consultancy or a specialist boutique fits better is decided by scope and steering needs. Programs with dozens of parallel teams call for the capacity of a large firm. Initiatives with one to five use cases benefit from a boutique that puts the experienced person directly into the project. Criteria for this choice are set out in the article Choosing an organizational consultancy.

How much does an AI consultancy cost?

According to the fee study of the German Association of Management Consultants (BDU), external consulting is typically billed at around 1,000 to 2,500 euros per consultant day in Germany. Total cost is driven mainly by scope and duration. A potential analysis with prioritized use cases takes a few weeks, a prototype in the real workflow several weeks to a few months, and the transfer into operation additionally requires integration with existing systems. How an analysis runs in four weeks is shown in the article AI potential analysis.

For small and medium-sized companies in Germany, federal and state funding programs exist whose conditions change regularly. The federal funding database lists the programs open at any given time. The publicly funded Mittelstand-Digital centers also offer free initial guidance.

A selection process in four steps

A structured process shortens the decision and makes it traceable for colleagues and business units.

First, define the goal and the metric. State which workflow should improve and how the improvement will be recognized. This goal becomes the yardstick for all proposals. How the executive team translates such a goal into a steerable portfolio is described in the article AI management: what executive leadership must steer.

Second, build a shortlist of two to four consultancies and send them all the same brief. Comparable proposals only emerge when everyone answers the same question.

Third, have the candidates work on a real case. A work sample of a few hours, in which the business unit and the consultancy sketch the business case together, reveals more about the way of working than any presentation.

Fourth, check references and structure the contract so that milestones are tied to the operational result.

Niehues Consulting as a partner for AI initiatives

Niehues Consulting in Düsseldorf advises companies from a delivery perspective. Founder Marin Niehues builds organizations that deliver and brings AI into operation. 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 services follow the five criteria of this article:

  • AI management: business analysis and business case, plus the selection and sequencing of applications and steering of the AI portfolio through metrics.
  • AI prototypes, operationalisation and implementation: prototypes that prove value in the real workflow, maturing until the business units adopt them, and transfer into operation.
  • Delivery management and organization design: steering initiatives as delivery manager or product owner, and roles with clear decision paths so that the organization delivers reliably.

The approach follows the Niehues Delivery Model with its phases understand, build and roll out. Three of its review questions serve as a yardstick for any consultancy, including competitors. Which business outcome should change measurably? Does the prototype prove its value in the real workflow with real data? Who owns operation from day one? The entry point is a free 30-minute initial consultation: Book an initial consultation.

Frequently asked questions

What does an AI consultant do?

An AI consultant identifies the AI applications with the greatest business value, tests them as prototypes in the real workflow and brings them into operation. This includes the business case, prioritization, implementation in the business units and steering the portfolio through metrics.

How do I recognize a good AI consultancy?

You recognize a good AI consultancy by its track record in operation. It describes the path from business analysis through the prototype to implementation, names responsibilities and defines metrics. References showing a workflow in live operation carry more weight than presentations of methods.

What are the red flags when hiring an AI consultant?

Red flags are savings promised before a baseline has been measured and a tool chosen before the analysis. Further signals are references without any application in live operation and a proposal that says nothing about operation. Two such signals justify exclusion.

Which questions should I ask in the first meeting?

Ask about the goal and about operation after implementation. Which workflow should improve? Who takes responsibility afterwards? How is the contribution to business results measured? The answers show whether the consultancy thinks from the result.

Do I need a consultancy or is a software vendor enough?

A software vendor is enough when a standard product covers a clearly defined workflow and your team steers the implementation itself. A consultancy pays off as soon as several units are affected, the business case is open or prototype and operation should be in one pair of hands.

How much does AI consulting cost per day?

In Germany, consultant days are typically billed at around 1,000 to 2,500 euros. Total cost is driven by scope and duration, ranging from a potential analysis over a few weeks to transferring a prototype into operation over several months.

Is there funding for AI consulting for SMEs in Germany?

Yes, the federal government and the states fund consulting for small and medium-sized companies through changing programs. The federal funding database shows which programs are currently open and their conditions. The Mittelstand-Digital centers additionally offer free initial guidance on AI.

How does a selection process for an AI consultancy work?

The process has four steps. You define the goal and metric, build a shortlist with an identical brief, have a work sample produced on a real case and, after checking references, tie the contract to the operational result.

Let’s talk about your initiative.

In 30 minutes we clarify where your organization stands and which step will pay off most now.

Request a consultation

30 minutes · free · no obligation