10 Min. Reading Time<\/p>\n
Around 70 percent of all AI projects in German companies fail. Not because of the technology, but due to resistance from the workforce. Those who introduce artificial intelligence without considering change management simultaneously burn budget and trust. A guide for decision-makers who want to do it better.<\/strong><\/p>\n There’s a dangerous misconception prevalent in German boardrooms. AI implementation is treated as an IT project: allocate budget, buy a tool, roll it out. Done. However, reality looks different<\/a>. AI doesn’t just change processes, but roles, hierarchies, and the self-image of entire departments. Ignoring this provokes resistance.<\/p>\n The numbers confirm this. McKinsey has consistently quantified the failure rate of change initiatives at around 70 percent for years. The main reason: inadequate involvement of those affected and lack of commitment from leadership. In AI projects, this problem is exacerbated because the technology triggers existential fears. Employees don’t ask if a new tool is more convenient; they ask if they’ll still have a job next year.<\/p>\n The Boston Consulting Group published a comprehensive study on AI usage in the workplace in 2025. The result initially sounds positive: 72 percent of employees worldwide now use AI tools. However, beneath the surface, a different picture emerges. Only among knowledge workers is penetration high, and the ROI question<\/a> becomes more pressing. Among operational employees, usage stagnates at 51 percent.<\/p>\n Even more crucial: only 26 percent of surveyed companies systematically invest in redesigning workflows and training their teams. The rest leave employees to use ChatGPT on their own. The result is isolated productivity gains among individuals, but no organizational transformation. The gap between companies that introduce AI strategically and those that leave it to chance will continue to widen in 2026.<\/p>\n There’s a school of thought that argues the opposite. Particularly in Silicon Valley, it’s claimed that good technology sells itself. If an AI tool is truly useful, employees will use it voluntarily. Change management is seen as an outdated concept from the industrialization era that only slows down the natural adoption process.<\/p>\n There’s some truth to this position. The BCG study indeed shows that many employees discover and use AI tools on their own, often without the IT department’s knowledge. The problem lies elsewhere: without guidance, a patchwork of tools, processes, and data silos emerges. And the departments that could benefit the most are left out. Change management isn’t about persuading employees; it’s about creating the framework for scaled adoption.<\/p>\n Siemens took a different approach when introducing its internal AI platform. Instead of a centrally controlled rollout, the company appointed AI ambassadors in each division. These aren’t external consultants or IT specialists but subject matter experts from their respective areas who have undergone a three-day AI training.<\/p>\n These ambassadors know their team’s pain points. They understand which tasks are tedious, where errors occur, and where time pressure is high. Their task is to identify concrete use cases and test them with colleagues. No slide deck, no town hall meeting. Just a colleague showing how they save an hour a day with AI support. Siemens reports an 80% acceptance rate in departments using this model.<\/p>\n \n“67 percent of change management practitioners are already using AI or planning to implement it. The biggest challenge remains acceptance within the affected teams, not the technology itself.”<\/p>\n Prosci, AI in Change Management: Early Findings (October 2023)<\/span>\n<\/p>\n<\/blockquote>\n Analyzing failed and successful AI implementations reveals five phases that companies should systematically go through. This isn’t a rigid framework, but a guiding structure that each organization must adapt to its situation.<\/p>\n Phase 1: Diagnosis (Weeks 1 to 4).<\/strong> Before selecting a tool, it’s essential to identify where AI has the most significant impact. This involves interviews with team leaders, process analyses, and an honest assessment of data quality. Many projects fail in this phase because executives skip the diagnosis and jump straight to the pilot.<\/p>\n Phase 2: Identify Quick Wins (Weeks 5 to 8).<\/strong> The first AI application shouldn’t be an ambitious transformation project, but a clearly defined use case with measurable results. Examples include automated sales proposal templates, AI-supported categorization of support tickets, or intelligent scheduling. The effect must be visible within two weeks.<\/p>\n Phase 3: Build Ambassadors (Weeks 9 to 16).<\/strong> Parallel to the first pilot, AI ambassadors are identified and trained in each department. Importantly, these don’t have to be digital natives. The best ambassadors are experienced professionals who are respected within their teams and open to new ideas.<\/p>\n Phase 4: Scaling with Guardrails (Months 5 to 9).<\/strong> Only when quick wins and the ambassador network are in place does broad implementation begin. Clear governance rules are needed: What data can flow into AI systems?<\/a> Who checks the results? How are errors handled? These rules should be developed collaboratively with the teams, not imposed from above.<\/p>\n Phase 5: Institutionalize Cultural Change (from Month 10).<\/strong> AI competence must be integrated into job profiles, onboarding processes, and performance evaluations. Companies that fail to do this will experience a relapse into old patterns after the initial enthusiasm. Anthropics’ current labor market study<\/a> shows that 60 percent of all job categories will shift their requirement profiles within two years.<\/p>\n Sources: Prosci Change Management Study 2023, BCG AI at Work Report 2025, Siemens Digital Industries<\/p>\n A common mistake is planning AI projects without involving the works council until the implementation is already decided. This not only creates legal risks because the works council has co-determination rights regarding performance and behavior-relevant systems. It primarily generates mistrust, slowing down the entire process.<\/p>\n Companies that involve the works council early on report significantly faster implementation. The reason is simple: when the works council supports the project, resistance among employees decreases substantially. A works agreement on AI usage creates legal and planning security for both parties. It regulates data protection, qualification, and the guarantee that AI implementation isn’t equivalent to job cuts.<\/p>\n Knowing the typical pitfalls can help you avoid them. These seven mistakes are the most common across industries:<\/p>\n \u25cf No Executive Sponsor:<\/strong> Without a board member visibly backing the project, it lacks legitimacy. Middle management blocks it because it’s unclear whether the project has political backing.<\/p>\n \u25cf Too ambitious first steps:<\/strong> A company-wide AI rollout as the first project overwhelms both the organization and IT. Better: one team, one use case, three months.<\/p>\n \u25cf Training as a mandatory event:<\/strong> Full-day training sessions without practical relevance create frustration rather than competence. Better: 90-minute workshops with real tasks from daily work.<\/p>\n \u25cf Lack of success measurement:<\/strong> If no one measures whether the AI implementation yields results, no one can argue why it should be continued.<\/p>\n \u25cf One-way communication:<\/strong> Newsletters and intranet articles are not enough. Employees need a channel to ask questions, express concerns, and contribute ideas.<\/p>\n \u25cf Data protection as an afterthought:<\/strong> AI systems processing personal data require a GDPR-compliant foundation. Clarifying this during rollout costs months. Compliance and AI<\/a> must be considered together from the start.<\/p>\n \u25cf No Plan B:<\/strong> Not every pilot works. Companies need to be willing to discard a use case and test another without questioning the entire AI program.<\/p>\n AI change management is not a project with a beginning and an end. It’s a permanent leadership task. The good news: companies that do it right see results quickly. According to Prosci, the adoption rate increases from 15 to 75 percent when professional change management is used. That’s a factor of 5 with identical technology.<\/p>\n The first step is not a tool evaluation or a strategy paper. The first step is a conversation. With five team leaders from different areas, with the works council, and with the employees who work daily with the processes that AI is supposed to change. Those who listen before deciding have a better chance of being among the 30 percent who succeed.<\/p>\n Read more on MBF Network<\/p>\n AI Governance in the Boardroom: Why CEOs Can’t Delegate Artificial Intelligence<\/a> (Digital Chiefs)<\/p>\n AI in the Labor Market 2026: What the Anthropic Study Means for Decision-Makers<\/a> (MyBusinessFuture)<\/p>\n DORA and NIS2 Simultaneously: Double Compliance Pressure for Financial Service Providers<\/a> (SecurityToday)<\/p>\n<\/div>\n Plan for 12 to 18 months from diagnosis to stabilization. Quick wins should be visible after 8 weeks, and full scaling typically takes 9 to 12 months. It’s crucial that the process doesn’t end after the pilot phase.<\/p>\n<\/details>\n As a rule of thumb: 15 to 20 percent of the total AI project budget should be allocated for change management. For an AI project over 500,000 Euro, that’s 75,000 to 100,000 Euro for training, communication, ambassador programs, and accompanying research.<\/p>\n<\/details>\n Ideally, a C-level sponsor in combination with a dedicated change manager. The IT department alone is not enough, as AI change is primarily a cultural and organizational task. HR, works councils, and specialist departments must be involved.<\/p>\n<\/details>\n Yes, to a limited extent. AI-powered pulse surveys can measure sentiment in real-time, chatbots can answer frequent questions about AI adoption, and analytics platforms can identify areas with low adoption rates. However, strategic steering and personal conversations cannot be automated.<\/p>\n<\/details>\n Aim to establish a works agreement early on that regulates AI usage, data protection, and training guarantees. In practice, works councils rarely block AI projects out of principle. Often, it’s due to a lack of information, transparency, or a written assurance that AI adoption won’t lead to layoffs.<\/p>\n<\/details>\n Source title image: Pexels \/ MART PRODUCTION<\/p>\n","protected":false},"excerpt":{"rendered":" 10 Min. Reading Time Around 70 percent of all AI projects in German companies fail. Not because of the technology, but due to resistance from the workforce. Those who introduce artificial intelligence without considering change management simultaneously burn budget and trust. A guide for decision-makers who want to do it better. Key Takeaways According to […]<\/p>\n","protected":false},"author":205,"featured_media":89434,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"_yoast_wpseo_focuskw":"change management","_yoast_wpseo_title":"Change Management: Guide to successful AI project adoption","_yoast_wpseo_metadesc":"Digitalisierung: Boost your business with AI! Learn proven change management strategies to ensure your AI projects succeed. Read now!","_yoast_wpseo_meta-robots-noindex":"","_yoast_wpseo_meta-robots-nofollow":"","_yoast_wpseo_meta-robots-adv":"","_yoast_wpseo_canonical":"","_yoast_wpseo_opengraph-title":"","_yoast_wpseo_opengraph-description":"","_yoast_wpseo_opengraph-image":"","_yoast_wpseo_opengraph-image-id":0,"_yoast_wpseo_twitter-title":"","_yoast_wpseo_twitter-description":"","_yoast_wpseo_twitter-image":"","_yoast_wpseo_twitter-image-id":0,"featured_post_sortierung":0,"featured_post":0,"pre_headline":"","bildquelle":"","teasertext":"","language":"de","_wp_old_slug":[],"footnotes":""},"categories":[2214,1156,1276,2027],"tags":[],"class_list":{"0":"post-93982","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","6":"hentry","7":"category-artificial-intelligence","8":"category-digital-business-future","10":"category-kuenstliche-intelligenz","11":"entry"},"wpml_language":"en","wpml_translation_of":89435,"acf":[],"yoast_head":"\nKey Takeaways<\/h2>\n
\n
The Problem Starts in the Executive Suite<\/h2>\n
What the BCG Study Really Shows About AI Adoption<\/h2>\n
The Counterposition: Does AI Really Need Change Management?<\/h2>\n
The Siemens Model: AI Ambassadors Instead of Top-Down Rollout<\/h2>\n
\n
Five Phases for Sustainable AI Change Management<\/h2>\n
Why the Works Council Isn’t an Adversary<\/h2>\n
Checklist: The 7 Most Common Mistakes in AI Change Projects<\/h2>\n
What Decision-Makers Should Do Now<\/h2>\n
Frequently Asked Questions<\/h2>\n
How long does a typical AI change process take?<\/strong><\/summary>\n
What does professional AI change management cost?<\/strong><\/summary>\n
Who should be responsible for AI change management?<\/strong><\/summary>\n
Can AI itself help with change management?<\/strong><\/summary>\n
What to do if the works council blocks AI projects?<\/strong><\/summary>\n
Further Reading<\/h2>\n
\n
More from the MBF Media Network<\/h2>\n
\n