Do CFOs need new KPIs in the era of AI?

Not necessarily new KPIs, but better ones. AI offers two big opportunities: measuring what we thought was unmeasurable and finding patterns we couldn’t see before.

Leading indicators, for instance, can predict future outcomes and will be increasingly visible with AI’s help. These indicators have always been harder to find and often overlooked. Many organizations only track what’s easy to measure, but that can lead to missed opportunities. Leading indicators usually require more creativity and collaboration to define, but they offer a much greater window for proactive decision-making. One example: a pension company noticed that people who checked their fund valuation were likely to transfer out. That behavior became a leading indicator.

The key is for CFOs to look for behaviors, signals, or early events that statistically tend to precede the outcomes you care about. Often, it’s tracking engagement before results — like website visits before sales, or employee pulse survey scores before attrition. Some other examples: low usage of internal tools might predict future tech support tickets; increased time-to-fill in recruiting could signal a pending talent bottleneck.

Treat leading indicators with care once you find them—they can change over time or lose relevance as behavior adapts. The process isn’t one-and-done; it’s iterative.

How can CFOs identify proxy indicators when direct measurement isn’t possible?

Proxy indicators may not be perfect, but they offer valuable insight when you can’t measure direct metrics or when tracking them is too costly or unreliable.

Start by deeply understanding your processes. As a CFO, you need to talk to frontline experts. For example, in call centers, customer frustration can be detected in real time through sentiment analysis.

I use the Holmes Method, a three-step process:

1. Has anyone done it before? Research how others have solved similar problems.
2. Can we find a proxy? Look for indirect signs of the behavior or outcome.
3. Use Fermi estimates. Break big unknowns into smaller, bounded estimates.

For example, if you managed public health and wanted early warning signs of a COVID outbreak, you could use proxy methods such as tracking Google search trends for related symptoms or sampling wastewater to measure virus density in the sewage system—both indicators that rise well before positive test results appear.

How can KPIs influence company culture and behavior?

KPIs affect behavior significantly. But most companies focus on extrinsic motivation (bonuses, rewards). Intrinsic motivation is far more powerful—pride, peer recognition, and personal achievement matter more. For example, in one factory, we wrote performance results on a whiteboard. Without offering bonuses, managers began checking the board first thing every day out of pride.

Another example: in a major retail bank that transformed its approach to performance measurement, the Continuous Improvement team went from ‘uninvited guests’ to ‘oversubscribed resource’ once the operations teams received full ownership and accountability for their own performance.

Transparency, trust, and regular feedback turn KPIs into conversation starters, not just metrics.

How can CFOs help non-finance leaders understand and act on KPI data? 

A CFO needs to present KPIs as answers to business questions. Don’t just show data—explain relevance. Non-finance leaders don’t need every number. They need clarity, insight, and confidence to act.

Not as many CFOs as you’d think know how to communicate KPIs. CFOs are excellent with financial KPIs, but a common issue is that finance teams often present them in ways that overwhelm non-financial audiences. Visual storytelling is key. Use intuitive layouts, color coding, traffic-light indicators, trend lines, and simple annotations to highlight the “so what” of the data.

You also need to reduce cognitive friction. Avoid financial jargon unless it’s necessary. Instead of “EBITDA margin delta,” say “change in operating profit percentage.” And include context wherever possible: how does this quarter compare to last, or to targets? Are we trending up, or flatlining? One effective trick is to show a KPI as part of a mini narrative: “We saw churn rise 12% after the price hike; this coincides with a drop in NPS, suggesting a value perception issue.”

CFOs need to pull in the expertise of others in their company and link KPIs to what I call the “Big Six” strategic outcomes: Profit, Growth, Innovation, Risk Management, Compliance, and ESG. These form the framework for any meaningful KPI strategy.

What’s your top advice for CFOs right now? 

Use a structured method to develop KPIs that link directly to strategy, span the entire business, and balance leading and lagging indicators. At the same time, embrace complexity. Automation is taking over the routine work.. The future lies in navigating the spaces between disciplines—the difficult, cross-functional, messy problems that AI can’t yet solve. That’s where strategic CFOs will thrive.

One CFO who I know embedded analysts directly into business units, acting as translators between finance and ops. It elevated the conversation. The best CFOs don’t just ask “What happened?” They ask, “What could happen, and what should we do about it?”

KPIs only work if people trust them and see their value. Bring stakeholders into the process. Don’t just inform—involve. Ownership and engagement are what make metrics matter.

Think of it like gardening. If you want plants (metrics) to thrive, they need the right soil (culture), regular care (review), and the right climate (context). Otherwise, you just get dashboard weeds no one looks at.