The Real Test of AI Investment: Can Businesses Change How They Work to Deliver Lasting Value?

McKinsey research suggests businesses gain more value from AI when they redesign workflows, improve decision-making and build employee skills. Leaders should measure business results, strengthen accountability and support training to turn AI investment into lasting productivity gains.

Businesses investing in artificial intelligence need to change how they work if they want better results, according to McKinsey research. Buying AI tools can help employees complete tasks faster, but the wider benefits depend on how companies make decisions, organise teams, develop skills and measure success.

The report, published on September 9, draws on a survey of 701 executives and senior leaders across industries and regions. It finds that organisations fundamentally redesigning their operations around AI, described as "reinventors", report stronger organisational performance than those at earlier stages of adoption.

For business leaders, investors and governments, the findings raise an important question: how can spending on AI translate into lasting improvements in productivity, services and competitiveness?

Better Tools Need Better Ways of Working

AI can speed up individual tasks without improving the overall business. An employee may prepare a document quickly, but its approval could still take days. A customer-service system may answer questions instantly, while customers continue waiting for another department to resolve their problems.

These examples show why companies need to examine complete processes. The aim should be to improve the outcome for the customer or business, with clear responsibility at every stage.

McKinsey recommends bringing employees from different departments together around shared goals. This could mean organising a team around resolving customer complaints or completing orders, rather than leaving each department responsible only for its own activity.

For investors, the implication is straightforward: AI adoption alone does not demonstrate business value. Companies should explain whether their investments improve service quality, reduce overall costs or increase revenue.

They also need to account for training, maintenance, system integration and supervision. Time saved becomes a financial benefit only when the organisation uses it effectively.

Faster Decisions Must Deliver Measurable Results

The research suggests that leading organisations maintain a clear long-term direction while adjusting budgets and staffing as new evidence emerges. Some use quarterly targets; others update their plans more frequently.

This approach could help businesses expand successful AI projects and stop those that fail to deliver. However, frequent reviews should not become pressure to produce immediate savings from every investment. Improving data systems or training employees may take time.

Leaders therefore need realistic targets and clear measures of progress. A useful review would examine whether a project improves the entire process, how much it costs and whether its benefits can be sustained.

Rewards also matter. If managers receive bonuses only for improving their own department's performance, they may overlook problems elsewhere in the business. Combining individual rewards with team results could encourage cooperation while preserving personal accountability.

Faster decisions must also have clear limits. Employees need to understand what they can approve themselves and when human review or senior management involvement is required.

Skills and Partnerships Will Shape the Benefits

AI is likely to change the tasks within many jobs. Companies will need to identify which activities can be automated, which require human judgment and where people and technology should work together.

McKinsey recommends organising talent around relevant skills rather than relying entirely on fixed job descriptions. For employees, this could create opportunities to take on new responsibilities. It could also create uncertainty if roles change without adequate training or clear career pathways.

Businesses should connect training to actual work. Employees need opportunities to practise new skills and understand how those skills will influence assignments and promotion.

Companies must also decide which capabilities to develop internally and which to obtain from partners. External providers can offer expertise and help accelerate implementation, but businesses still need the knowledge to assess their performance.

Data access, reliability, responsibility for mistakes and the ability to change suppliers should form part of these decisions. Partnerships work best when they support a clear business goal and management retains effective oversight.

Governments Need to Support Practical Adoption

For governments and development partners, the findings suggest that access to AI software is only one part of business readiness. Smaller firms may also need affordable training, better information systems and support to improve management practices.

Assistance programmes could help businesses identify suitable processes, test changes and measure results. Success should be assessed through improvements in productivity, service quality and business performance, rather than simply counting software purchases.

The findings also require careful interpretation. The survey reflects individual leaders' responses, and only 53 respondents were classified as reinventors. It does not prove that organisational redesign alone caused better outcomes.

Nevertheless, the practical message is clear. Companies should begin with a specific business problem, establish how performance will be measured and give teams the skills and authority to improve it. The value of AI will depend on how effectively businesses turn its capabilities into reliable results.