The Cost-Effectiveness of AI in Small and Medium-Sized Businesses: How to Manage AI Budgets

Do you know how much your company spends on AI? And do you manage that budget effectively? If not, you’re not alone: Many small and medium-sized businesses lack a comprehensive overview of their AI costs. That’s not surprising. After all, numerous AI licenses are now in use, mostly spread across many different AI tools: from ChatGPT and Copilot to Gemini, Claude, and Perplexity. And more and more AI features are being integrated into existing software, some of which come with a fee. To make matters worse, OpenAI, Microsoft, and Google love to update their licensing models—as do Anthropic and Perplexity. This makes it increasingly difficult to monitor and control AI costs. We’ll explain how you can still maintain an overview and avoid unnecessary AI costs.

Gespräch 2 Personen über Wirtschaftlichkeit von KI im Mittelstand

Key Takeaways

  • To gain a complete overview, companies should track all AI services and their costs as centrally as possible.
  • Budgets, usage limits, and notification features help manage expenses effectively and avoid unexpected costs.
  • Users should be made aware of AI costs so they can help keep costs low through their usage behavior.
  • Costs alone say little about cost-effectiveness. Small and medium-sized businesses should therefore always consider how intensively AI solutions are used and what specific added value they create.
  • The goal should not be to operate AI as cheaply as possible, but rather to strike a sensible balance between costs and benefits.

Why AI Costs Are So Hard to Track

With traditional software, the calculation is usually relatively simple: number of users multiplied by the license price. With AI tools, however, additional factors come into play. In addition to user licenses, for example, API-, model-, and token-based costs may arise, while various features (such as Copilot Work) are available within AI tools that can incur additional costs. In addition, many existing systems—such as CRM or ticketing systems—also include AI features that aren’t always free.

All of this already makes it difficult to keep track of AI costs. However, changes that providers make to their licensing and pricing models make it even more challenging—especially when it comes to budget forecasting.

So it’s no surprise that, according to a KPMG survey, only 35 percent of companies have a complete overview of their AI expenditures. Given the rising use of AI, an ever-larger portion of costs remains uncontrolled.

At the same time, costs alone do not indicate whether an AI application actually creates added value. To operate AI cost-effectively, companies need more than just a list of existing licenses. They must clarify: What does our AI cost? And what benefits does it provide? Only by considering both perspectives together can AI be effectively managed in small and medium-sized businesses.

Controlling and Managing AI Costs: 6 Steps

  • 1. Track all AI services

    Which AI solutions are officially in use? This includes not only ChatGPT, Claude, and Microsoft Copilot, but also AI features within other SaaS applications and in-house solutions.

  • 2. Centralize cost visibility

    Admins should always keep an eye on the costs and usage data for these tools. With Microsoft, for example, you can view the number and percentage of active users as well as the number of prompts submitted. However, it’s even better when all costs across all tools are consolidated in one place, so that IT managers can easily get a comprehensive overview. The data should update automatically on a regular basis.

  • 3. Manage Budgets

    Budgets and usage limits should be defined for each tool to prevent costs from spiraling out of control. Notifications when limits are approaching or reached should help IT managers maintain transparency, control costs, and manage usage. This is precisely what our EBF Insight solution offers.

  • 4. Raising Employee Awareness

    Users also influence AI costs through their usage behavior. Costs can be reduced by avoiding unnecessarily long conversations, using smaller models for simple tasks, and making a conscious decision about when AI use is actually necessary. That’s why training is extremely important. Effective AI enablement in small and medium-sized businesses must also foster cost awareness among employees.

  • 5. Monitor Price Changes by Providers

    It is important to regularly monitor whether AI providers such as OpenAI, Microsoft, Google, Anthropic, and Perplexity—as well as software vendors like HubSpot, Salesforce, and others—are changing their licensing models, and what impact these updates will have on your own company.

  • 6. Analyze usage and added value

    A paid license alone does not create added value. Therefore, in addition to the costs, you should assess how intensively and effectively the provided solutions are actually being used. The results should then be taken into account when selecting AI tools and managing budgets.

Insights into practice

EBF-Mitarbeiter

“A paid AI license does not automatically translate into productivity gains. That’s why, in our client projects, we assess costs, actual usage, and the value AI delivers for specific tasks together. This enables companies to allocate licenses according to need and make informed decisions about which applications to expand, adapt, or discontinue.”

Robert Hamel Solution Architect EBF

What tips can help employees avoid unnecessary AI costs?

  • Not every task—no matter how small—needs AI. For example, you can usually write emails to colleagues yourself, and they’re often shorter and more authentic that way.
  • Not every task requires the most powerful model. A smaller model may be perfectly sufficient for simple tasks, while expensive models add value for more complex ones.
  • You can save tokens by summarizing chats regularly instead of letting them go on endlessly—as Claude, for example, points out time and again.
Illustration eines Mädchens, das eine Glühbirne wie einen Ballon hält

Our conclusion: Companies need true cost-effectiveness rather than mere cost control.

AI costs for small and medium-sized enterprises will be significant in the future. It is therefore important to keep an eye on these costs to avoid any surprises.

However, the goal of cost control should not be to operate AI as cheaply or restrictively as possible. What matters most is the balance between costs and benefits: Where does AI accelerate processes? Where does it improve results? Where is AI used intensively? Where is a company paying for solutions that are merely gimmicks without delivering any added value? By continuously answering these questions, a company lays the foundation for cost-effective and scalable AI deployment.

Contact

Do you have any questions? We’d be happy to talk with you.

Would you like to gain transparency into your AI usage and costs? We can help you centrally monitor and effectively manage your AI services.

AI Governance Beratung