Why 95 percent of AI projects fail
According to an MIT study, 95 percent of all AI pilot projects do not deliver measurable results. The Boston Consulting Group estimates: 10 percent of success depends on algorithms, 20 percent on technology and data, and 70 percent on people and their processes. 31 percent of German companies cite lack of employee acceptance as the biggest obstacle, according to Bitkom, and only 28 percent have a change management strategy for AI implementation.
Recommendation: Before introducing any AI, ask yourself three questions: Have we asked the people who will actually use it? Do they understand what will change for them and why? Do they have the skills to use the tool responsibly?
Link to Radar newsletter: https://kompetenz-radar.de/ausgabe/2026-10-03-warum-95-prozent-der-ki-projekte-scheitern
How can I learn AI? The great skills crisis According to a survey by the Stifterverband, 79 percent of executives say their employees lack basic AI skills. In small and medium-sized enterprises, more than half cite missing competencies as the reason for not using generative AI. Only 16 percent of employees receive training before new AI tools are introduced, and just 24 percent feel well prepared—compared to 77 percent of executives who assume their teams are.
Recommendation: Focus on systematic qualification instead of lip service—train employees before introducing new AI tools rather than relying solely on learning by doing.
Link to Radar newsletter: https://kompetenz-radar.de/ausgabe/2026-10-06-wie-kann-ich-ki-lernen-die-grosse-kompetenzkrise
AI costs rise 48% in 2026—despite cheaper prices German companies are expected to spend €28.7 billion on AI in 2026, an increase of 48 percent year-over-year, even though token prices have fallen by 41 percent since March. Only 21 percent of SMEs have a defined AI strategy according to a study by the ILIN Institute, and only 26 percent are aware of their real-time AI costs, according to KPMG. Lower token prices lead companies to use AI more broadly and more frequently—the consumption grows faster than the per-unit cost savings.
Recommendation: Implement model routing, prompt caching, usage limits, and transparency about which queries go to which AI systems to prevent efficiency gains from turning into unplanned cost drivers.
Link to Radar newsletter: https://kompetenz-radar.de/ausgabe/2026-10-07-ki-kosten-steigen-2026-um-48-trotz-guenstigerer-preise
113 minutes a day: These five AI tools save time Employees in Germany save an average of 113 minutes per day through AI use. Five tools cover different tasks: Microsoft 365 Copilot for emails and scheduling, ChatGPT and Notion AI for text and documentation, GitHub Copilot for development, and Zapier for workflow automation without coding. According to Deloitte, companies with structured AI adoption report 30 percent higher productivity in administrative processes.
Recommendation: Start with one tool for your most time-consuming task, test it for two weeks, measure the time saved, and then scale up.
Link to Radar newsletter: https://kompetenz-radar.de/ausgabe/2026-10-08-113-minuten-taeglich-diese-fuenf-ki-tools-sparen-zeit
Heilbronn tests AI—but who asked the clerks? Since October 2026, the Heilbronn Foreigners’ Office has been testing an AI system to process residence permit applications, a year ahead of the EU’s strict requirements for high-risk systems coming into force in December 2027. International studies covering over 16,000 businesses show that IT projects with active participation from employees and works councils achieve success rates of 67 to 78 percent, while projects without such involvement reach only 32 to 45 percent. In AI adoption, projects with works council involvement showed 30 percent higher acceptance and 25 percent fewer issues post-implementation.
Recommendation: Involve the employees who will actually use the new system from the start—this doubles the success rate of IT and AI projects, according to the cited studies.