AI is now ubiquitous in the workplace, but studies have found that most companies do not see its impact on enterprise earnings or growth. A global CEO survey found that 56 percent of CEOs haven’t yet seen significant financial benefits from AI. A recent International Labour Organization paper discussed the AI aggregation paradox—strong evidence of individual productivity gains but far weaker evidence at an enterprise, industry or macroeconomic level.
A principal reason for this value-capture gap is how organizations measure, encourage and reward AI adoption and activities. Given the pace of change in the market, management teams often look for the quickest evidence of progress. Many organizations now track activity metrics such as AI usage rates, tokens consumed, licenses purchased, seats deployed and daily adoption rates. Measures of value such as better decisions, faster processes, improved financial KPIs, higher quality of work and commercial results are harder to define and attribute.
Even leading AI adopters can focus too heavily on activity rather than outcomes. One example is “tokenmaxxing”—maximizing the use of AI tokens without regard to the output, which became a growing phenomenon even in Silicon Valley until companies began curtailing AI spending per employee.
This paradox highlights a simple truth: AI usage is a variable cost to the business, just like human labor. Organizations must therefore demonstrate how AI creates value and closes the value-capture gap. To address this gap, directors may consider three fundamental, interconnected questions:
- How do we know if our AI initiatives are truly successful?
- How do we measure such success, often defined as return on investment (ROI)?
- How should we hold the management team accountable for these metrics?
Driving the right accountability
Management incentive plans are one of the most powerful tools directors can use to align priorities. Well-designed incentive plans influence behavior throughout the organization. Proponents of AI activity metrics may point to the benefits of creating a culture of adoption and signaling to the market ongoing business transformation. On the other hand, investors expect executive compensation to reflect value creation from AI and will question activity-based metrics.
WTW’s analysis of a sample of S&P 500 companies’ 2026 proxy filings shows that incorporating AI metrics into executive incentive plans remains uncommon. Only 8 percent of sampled companies included an AI-related metric in their executive incentive plans (vast majority in the annual incentive plan). Among the companies that use AI incentive metrics, most include it as a component of a broader strategic or individual performance scorecard rather than a standalone metric. When included in a broader scorecard, these AI metrics tend to be qualitative.
What good looks like
Activity metrics are not inherently bad when taken in the right context as they provide useful insights into how employees adopt AI, develop new skills and experiment with new ways of working. Employee sentiment can also help measure cultural change. However, these metrics are less effective when assessing executives’ performance or determining their compensation.
Directors can hold management accountable for defining business-specific metrics to track value over time, allowing them to monitor progress and advise management in evolving the company’s AI enablement strategy with a focus on value capture through growth, cost reduction, improving employee experience and motivation. Some examples of outcome-based AI metrics may include:
- Revenue from AI-enabled products and services, either measured in absolute terms or as a share of overall revenue mix, connecting AI value directly to growth.
- Gross or operating margin improvement in processes where AI is used, capturing efficiencies directly impacting the P&L.
- Customer outcomes such as retention, satisfaction or resolution rates on products or services where AI is deployed to enhance features or improve customer service.
- Cycle time for key processes such as sales or product development, measured against a historical baseline to assess if productivity gain positively impacts the customer experience.
- Improvements in employee productivity, engagement and wellbeing, which shows that the workforce is aligned with the company’s AI vision and that the company is maximizing human potential in creating value.
Adoption was never the goal; value was
Some may argue that traditional measures such as revenue growth, profitability and stock price already capture the value creation from AI. Using more specific AI metrics can provide more targeted incentives when stakes are high (e.g., high-profile AI investments, AI development shakes the core of a business such as in the semiconductor sector), directing attention to AI-related strategic priorities within the business’ context.
Motivating and rewarding value metrics narrows the AI value capture gap by holding management accountable to what matters and by cascading value-based organization priorities throughout the organization. That often is the first step towards driving meaningful cultural change.
