The Digital Data Design Institute at Harvard is now the Harvard Business School AI Institute.

Measure the Complete Relay to Capture True AI Value

Illustration of business men running a relay race

Building on the last two months’ analysis of how better metrics clarify AI value, we are now putting those frameworks into practice. Recent data from McKinsey’s State of AI survey underscores the urgency of this approach. While 80% of respondents report gains in personal productivity, only 37% of organizations experience a tangible impact on earnings. Furthermore, the 6% of companies that qualify as high performers owe their success primarily to intelligent work design.[1]

In a relay race, equipping runners with elite footwear yields little benefit if the baton drops during handoffs. A similar vulnerability appears in how organizations evaluate performance. Standard AI dashboards frequently focus on isolated task metrics, such as prompt volume or individual hours saved. Consequently, executive teams lose sight of complete operational flow, which only culminates when work crosses the finish line to deliver value to the customer.

The Baton Gets Dropped at the Handoff

Evidence of lost efficiency appears across multiple corporate studies. According to BCG’s 2026 survey of nearly 12,000 workers, 42% of regular frontline AI users gain a full day or more each week. However, two-thirds receive little guidance on directing these reclaimed hours, leading over half to absorb the extra capacity without shifting toward higher-value tasks.[2] At the same time, Workday found that nearly 40% of generated time savings disappear during quality checks and output corrections.[3]

Software engineering provides a clear illustration of this pattern because technical teams measure workflows continuously from start to finish. In an analysis of 22,000 developers, Faros AI found that while task completion rose 34% and code-related task completion rose 210%, review times surged by about 200% on average. Consequently, delivery lead times quadrupled as individual output gains were absorbed by severe review bottlenecks.[4]

Timing the Complete Workflow

Addressing these handoff bottlenecks begins with a shift in measurement strategy. Management scorecards establish true operational pace by tracking overall duration. Three flow metrics can clarify where work moves or stalls:

  • End-to-End Cycle Time: Total duration from customer request to final result, such as processing a claim or submitting a proposal.
  • Wait Time: Duration work remains idle in queues and approval chains between handoffs.
  • First-Pass Yield: The proportion of work completed accurately on the initial attempt without requiring revision.

To illustrate the potential bottlenecks, consider a hypothetical commercial insurer that has deployed automated tools to assist adjusters with claims assessments. Adjusters cut drafting time from three hours to forty minutes. However, total payout duration remains at 18 days because files spend nine days awaiting two layers of supervisory sign-off. These approvals survive from an earlier period when drafts were slow and error prone. Streamlining these handoffs aligns internal acceleration with customer results.

Redesigning the Exchange Zone

In relay races, overall outcomes depend heavily on the exchange zone where the baton changes hands. When flow metrics pinpoint where work stalls, leaders can focus on restructuring these critical handoffs. Organizations that successfully convert time savings into bottom-line performance begin by removing legacy signoffs that existed to catch obsolete errors, often replacing them with risk-based thresholds. At the same time, they shift decision authority to specialists closest to the work, allowing teams to eliminate supervisory bottlenecks and replace rigid sequential reviews with synchronized, parallel workflows. Achieving these improvements requires a fundamental operating-model decision that belongs squarely on the executive leadership agenda.

This Month’s Action Item for Leaders

Trace a high-value customer workflow end-to-end to establish baseline turnaround speed and queue duration before modifying operations.

Trim redundant approval steps and handoffs where work sits idle, setting 90-day targets centered on total delivery speed.

Transfer decision rights to frontline teams closest to the work, providing explicit guidance on directing reclaimed capacity toward strategic priorities.

One Other Thing

A complete performance evaluation also factors in supervision load alongside primary flow metrics. Survey data from BCG shows that 47% of workers spend more time managing automated tools than performing primary tasks.[2] Because oversight requires real labor, factoring management time into performance equations ensures that accelerating initial drafting does not simply shift hours into expanded supervisory review.

Next Month at a Glance

As organizations optimize operational flow, attention shifts toward how individuals work alongside personal agents including OpenAI’s Dots, Anthropic’s Claude Duet, Microsoft’s Copilot Autopilot and Instinct. These are the new persistent agents that keep working after you log off. Next month, we will examine how leaders can construct clear usage guardrails that protect sensitive data while empowering employees to harness agentic capability. We will address what they can do, what they cost, and the governance questions that every leader should ask.

Sources

[1] https://www.mckinsey.com/capabilities/quantumblack/our-insights/the-state-of-ai

[2] https://www.bcg.com/press/3june2026-ai-reshaping-jobs-faster-than-companies-reshaping-work

[3] https://www.barchart.com/story/news/37034540/new-workday-research-companies-are-leaving-ai-gains-on-the-table

[4] https://www.faros.ai/research/ai-acceleration-whiplash


Meet the Author

Mike Grandinetti artistic headshot

Mike Grandinetti is an Executive Fellow at Harvard Business School. He’s a serial tech entrepreneur, board member, AI & innovation consultant, VC EIR, and award-winning professor in the practice. A former Silicon Valley engineer and McKinsey consultant, Mike has been C-Suite leader roles across 8 tech startups, resulting in 2 NASDAQ IPOs and 7 strategic exits.

He’s led senior executive workshops for Berkeley, Brown, Carnegie Mellon, Columbia, Cornell, Harvard P&ED, NYU & Oxford. He’s been a senior advisor and organizing team member for the MIT CIO Symposium for a decade.

https://www.linkedin.com/in/mikegrandinetti
www.mikegrandinetti.com