AI’s allure is undeniable, and businesses invest heavily in its promise. Bain & Company reports that companies today invest an average of $5 million annually into generative AI initiatives, with some even exceeding $10 million. This pursuit is fueled by AI’s potential to improve productivity, cut costs, and unlock new opportunities that may create new revenue streams. However, many companies find the path to successful AI challenging because of fundamental mistakes in their approach.
Klarna, a Swedish fintech company, made headlines in 2024 by replacing a significant portion of its customer support staff with an AI-powered chatbot. This move aimed to reduce costs and enhance the customer experience, but it proved a fundamental mistake. Eventually, Klarna rehired human representatives, recognizing that while the AI technology met its organizational needs, it could not satisfy customer expectations. Klarna’s experience and our firm’s corporate advisory work illustrate that as companies rush into AI, they must consider four dimensions: corporate needs, organizational capabilities, customer expectations, and value realization. Corporations frequently make nine fundamental mistakes across these four dimensions.
Mistakes associated with corporate needs
- Technology-First Approach: Many companies begin their AI journey by acquiring powerful AI tools without first identifying the strategic problems they must solve. This technology-first mindset often leads to misaligned investments and disappointing outcomes. Bain’s research highlights that only 35% of companies have a clearly defined vision for creating business value from AI.
- Underestimating Costs: Companies frequently underestimate the true costs associated with AI deployment. According to Deloitte, 78% of surveyed companies expect to increase their AI spending next year, as many realize that scaling AI requires substantial investments in data infrastructure, workforce upskilling, and governance frameworks. Depending on the problems they are trying to solve and the infrastructures they plan to use, such as API calls to frontier models or fine-tuning open-source large language models (LLMs), such increases may still prove inadequate.
Mistakes associated with organizational capabilities
- Data Deficiencies: Poor data quality is a pervasive challenge. Companies struggle with inconsistent, inaccurate, and disconnected data, limiting the effectiveness of AI systems. Improved data management is a top priority for successful AI model development and deployment.
- Talent Gap: Companies often lack the technical talent required to build, test, manage, and bring to market AI systems effectively. According to BCG, only 4% of companies have the capabilities required to generate substantial AI value at scale. Firms that invest in upskilling employees and hiring AI specialists are more likely to succeed.
- Overreliance on Existing Organizational Structures: Traditional corporate hierarchies often struggle to accommodate AI’s iterative and experimental nature, where people and machines collaborate to solve both hard and mundane problems. Companies that adopt agile methods and flatter organizations that focus on cross-functional collaboration achieve better outcomes.
- Cultural Resistance: Employee resistance to AI adoption is common, especially when automation disrupts established workflows. Successful companies invest in trust-building, emphasizing AI’s role as an augmentation tool rather than a replacement.
- Neglecting Change Management: Organizations frequently overlook the effort required to embed AI into daily operations. Without clear change management initiatives, AI projects risk becoming isolated pilot programs that fail to scale effectively.
Mistakes associated with customer expectations
- Failure to Align with Customer Preferences: Companies often deploy AI solutions focusing solely on cost reduction or internal productivity improvements. Today, few use AI solutions to create new revenue streams. Robotaxi providers, such as Waymo, are the exception rather than the rule. In deploying their AI-based productivity improvement or cost-reducing solutions, corporations may replace too many human interactions and overlook customer expectations and the quality of the resulting customer experience. As seen with Klarna, even technically successful AI systems can fail if they do not meet customer expectations for responsiveness, empathy, or quality of service. Successful AI deployment balances automation with meaningful human interaction, ensuring customer trust and satisfaction are maintained.
Mistakes associated with value realization
- Ethical Blind Spots: Companies that ignore AI’s ethical risks, such as algorithmic bias and privacy concerns, face reputational and regulatory risks. BCG emphasizes that leaders excel by prioritizing AI transparency, performance monitoring, and explainability to ensure models adapt appropriately to changing conditions.
- Data Security and Privacy Concerns: Given AI’s reliance on data, businesses must secure sensitive information to maintain customer trust. Data security remains one of the top barriers to scaling AI solutions.
The Path to AI Success
To succeed with AI and avoid these fundamental mistakes, businesses must align AI initiatives with strategic goals, invest in high-quality data infrastructure, upskill their workforce, and implement robust governance frameworks. Companies that prioritize steps will be better positioned to unlock AI’s transformative potential while avoiding the fundamental mistakes that have derailed so many others.
You can watch the entire interview here under podcasts.