Measuring AI impact requires tracking engine metrics over financials

Wollmann addresses the difficulty of measuring AI impact in enterprise settings, noting that direct monetary ROI is rare and often delayed. He distinguishes between efficiency gains (cost reduction) and top-line growth (new markets, pricing). For most use cases, such as knowledge retrieval or process automation, immediate financial impact is not visible. Instead, he recommends measuring 'engine' improvements: shorter process cycles, fewer incidents, faster decision-making, and reduced need for reverts. He emphasizes that velocity alone is insufficient; quality metrics must be balanced against throughput to ensure that increased speed does not result in lower-quality outputs. He cites the example of a senior engineer who asks daily, 'why should not an AI do this?' to drive adoption, but stresses that organizations must first define their strategic position before automating. The goal is to improve the underlying operational engine, which should eventually translate to balance sheet returns, even if the link is not immediate. Wollmann argues that organizations should focus on leading indicators like process cycle time and quality rather than waiting for lagging financial indicators to validate AI investments.