When Performance Goes Dark: Why Organizations Need Better Visibility Before AI Takes the Wheel

Rapid growth creates an appealing problem: more customers, locations, employees, and opportunities. It also creates a less celebrated one, leaders increasingly lose sight of who is performing, which teams are functioning, and why. Geographic expansion, demographic change, hybrid work, new generations of employees, and artificial intelligence are making this performance-visibility gap a strategic issue. In short: you cannot manage what you cannot see and you probably should not automate it either.

Performance measurement is a classic practical application of industrial and organizational (I-O) psychology. Campbell and colleagues’ seminal research demonstrated that individual job performance is multidimensional rather than a single quantity called “productivity.”¹ Research on team effectiveness similarly shows that outcomes depend not merely on assembling talented individuals but on cognition, coordination, motivation, leadership, and other team processes.², ³ Thus, a dashboard displaying sales or production may show what happened without revealing how or why it happened.

Why visibility is getting harder

Growth magnifies the problem. A founder who personally knows 40 employees may have excellent informal performance intelligence; at 4,000 employees across multiple countries, informal observation becomes anecdote. Distributed work further separates supervisors from employees, while changing workforce expectations alter how employees seek feedback and development. Deloitte’s 2025 survey found that roughly half of Gen Z and millennial respondents wanted managers to teach and mentor them, while substantially fewer reported receiving that support.⁴

The critical distinction is between organizational causes and macro-cultural causes. Low performance after a restructuring, incentive change, new supervisor, poorly designed workflow, or confusing policy is potentially an organizational-design problem. Generational expectations, demographic shifts, labor-market conditions, attitudes toward authority, and changing societal expectations about work operate more broadly. Treating every performance problem as an employee problem is bad I-O Psychology and occasionally very convenient management.

Construction illustrates the distinction. In 2025, 92% of surveyed contractors reported difficulty filling positions and 45% reported project delays associated with worker shortages.⁵ Associated Builders and Contractors separately estimated that the industry needed 439,000 additional workers in 2025.⁶ A superintendent observing slower completion therefore cannot automatically conclude that crews are underperforming. Staffing shortages, skill mix, subcontractor availability, scheduling, policy changes, and management systems may explain the numbers. Context is not an excuse; it is a variable.

The U.S. Army provides an unusually relevant example because modern performance science and Army personnel research overlap historically. Campbell’s Project A research examined thousands of soldiers across entry-level occupations using job knowledge tests, hands-on work samples, ratings, and administrative measures rather than relying on one supervisor’s impression.¹, ⁷ The Army continues moving toward data-rich talent management: its current personnel analytics tools are intended to give leaders faster insight into readiness and talent, while Army research explicitly addresses both individual assessment and team assignment/performance.⁸, ⁹ The lesson for business is straightforward: visibility improves when organizations replace isolated opinions with multiple, job-relevant indicators — not when they simply produce a larger spreadsheet.

AI raises the stakes

Agentic AI turns performance visibility from an HR concern into an operating-system requirement. Microsoft reports that organizations are moving toward human-agent teams and that 82% of leaders surveyed expected digital labor to expand workforce capacity within 12–18 months.¹⁰ McKinsey similarly argues that agentic organizations will require performance systems emphasizing outcomes and people’s ability to orchestrate agents rather than simple task completion.¹¹ Microsoft’s 2026 research also identifies a “transformation paradox”: employees may experiment with AI while legacy metrics and incentives continue rewarding yesterday’s work.¹²

That creates a dangerous equation: poorly specified performance + automation = poorly specified performance at machine speed. If managers cannot distinguish excellent individual performance from a favorable territory, or excellent teamwork from heroic compensation for a broken process, an AI agent trained or governed using those signals can institutionalize the error.

Three practical interventions follow directly from I-O psychology. First, organizations should conduct performance-focused job analysis and build behavioral and outcome measures at both individual and team levels, including quality, adaptability, collaboration, safety, and judgment — not merely volume. Second, they should create multi-source performance sensing: objective operational indicators, validated supervisor assessments, customer outcomes, team-process measures, and frequent developmental feedback. Third, leaders should establish human-agent performance architecture before automating workflows: define who owns each outcome, what constitutes acceptable performance, where human judgment remains mandatory, and how AI actions will be audited.

The emerging principle is simple: measure the work before mechanizing the work. I-O psychology has spent decades developing scientifically defensible ways to understand individual and team performance. In the agentic-AI era, that science is becoming infrastructure. Otherwise, organizations may finally achieve perfect visibility into the wrong things.

Footnotes

1. Campbell, J. P., McHenry, J. J., & Wise, L. L. (1990). “Modeling Job Performance in a Population of Jobs.” Personnel Psychology, 43, 313–575. The study examined 9,430 Army personnel and developed a multidimensional model of job performance.
2. Kozlowski, S. W. J., & Ilgen, D. R. (2006). “Enhancing the Effectiveness of Work Groups and Teams.” Psychological Science in the Public Interest, 7(3), 77–124. Their influential review identifies cognitive, motivational, behavioral, leadership, training, and design mechanisms underlying team effectiveness.
3. Salas, E., Cooke, N. J., & Rosen, M. A. (2008). “On Teams, Teamwork, and Team Performance: Discoveries and Developments.” Human Factors, 50(3), 540–547.
4. Deloitte. (2025). “Gen Zs and Millennials at Work: Pursuing a Balance of Money, Meaning, and Well-being,” Deloitte Insights. The survey included more than 23,000 respondents across 44 countries.
5. Associated General Contractors of America. (2025). “Construction Workforce Shortages Are Leading Cause of Project Delays.”
6. Associated Builders and Contractors. (2025). “Construction Industry Must Attract 439,000 Workers in 2025.”
7. Campbell, C. H., et al. (1990). “Development of Multiple Job Performance Measures in a Representative Sample of Jobs.” Personnel Psychology, 43, 277–300. The Project A criterion-development effort explicitly treated performance as multidimensional and employed multiple measurement methods.
8. U.S. Army Research Institute for the Behavioral and Social Sciences. “Our Research.” ARI identifies holistic personnel assessment and team assignment and performance among its applied research areas.
9. U.S. Army, Integrated Personnel and Pay System–Army. “Service for Analytics and Business Intelligence Reports (SABIR).” The platform provides leaders with personnel analytics intended to support readiness and talent-management decisions.
10. Microsoft. (2025). “The 2025 Annual Work Trend Index: The Frontier Firm Is Born.”
11. McKinsey & Company. (2025). “The Agentic Organization: Contours of the Next Paradigm for the AI Era.” The article argues that human-agent organizations will require real-time governance and redesigned performance and talent systems.
12. Microsoft. (2026). “2026 Work Trend Index Annual Report: Agents, Human Agency, and the Opportunity for Every Organization.”

Contributor’s Note: Mark Wilson, Chief Analytics Officer and Senior Business Advisor. Comments, suggestions, reactions, examples, and questions are welcome. Reach me at mark.wilson@horizonperformance.com

---

When Performance Goes Dark: Why Organizations Need Better Visibility Before AI Takes the Wheel was originally published in Horizon Performance on Medium, where people are continuing the conversation by highlighting and responding to this story.