Why Productivity Never Compounds

MICHAEL CARROLL | THE ONE-DEGREE DISPATCH | PRODUCTIVITY, CULTURE, AND DECISION ARCHITECTURE

Why Productivity Never Compounds

The hidden economic role of culture is to determine whether coherent action survives long enough for learning to become institutional capability.

By Michael Carroll

Industrial Transformation Leader | Research Fellow, LNS Research | Board Advisor | Chief Strategy and Operations Officer, Trek.AI

THREE TAKEAWAYS

At 6:12 a.m., the dry-erase marker squeaks across a production board that still carries the ghost of last year's priorities. Beneath the new headings, the old words remain visible: recurring defect, late material, changeover loss, approval waiting. Someone has taped a fresh program name across the top, but the work underneath it is familiar.

The operators do not need a speech to understand what happened. A leader changed, the measures were rearranged, ownership moved, and work that had only begun to produce learning was returned to its starting position. Nothing in the plant looks idle. The meetings are full, the actions are current, and the enterprise is once again paying to relearn what it already knew.

This is one reason productivity can remain flat inside organizations full of technology, talent, capital, data, and improvement programs. The missing variable is not always knowing what to do or knowing how to do it. It is the ability to preserve coherent action long enough for learning to become institutional capability and for that capability to produce value repeatedly.

The Reset Hiding Inside the Work

The common explanation for weak productivity is usually technical. The enterprise has the wrong equipment, too little automation, poor data, an aging workforce, inadequate skills, or an operating model that no longer fits the market. Each explanation can be true, and no serious account of industrial performance should pretend that culture overrides capital intensity, product mix, supply conditions, demand, regulation, or technological change.

Those explanations still leave an operating question unanswered. Why do organizations acquire capable technology, hire capable people, adopt capable methods, and fail to retain the gains? Why does a strong improvement weaken when it crosses a functional boundary or disappear when its sponsor leaves? Why does the next initiative begin by rediscovering lessons the previous initiative already paid to learn?

The answer is often hidden inside the reset. An enterprise can improve an operation without converting that improvement into institutional capability. It can solve a problem once without changing the conditions that cause it to return. It can execute well for a quarter without building the trust, permission, decision architecture, and operating memory required to execute well again under different pressure.

Productive capability is not the same as current output. It is the repeatable ability to detect what matters, reason about cause, decide close enough to the event, act with bounded authority, learn from the result, stabilize the gain, and carry that learning into the next cycle. The enterprise has not compounded anything until the next action is better because the previous action occurred.

What Actually Compounds

Culture does not compound value directly. Coherent action allows learning to accumulate. When that learning is retained in routines, evidence, permissions, decision rights, and operating memory, it becomes institutional productive capability. Capability then produces performance repeatedly, and repeated performance is what accumulates into value.

The full chain matters because the terms are not interchangeable. Improvement is an event. Learning is what the enterprise discovers from the event. Capability is the retained ability to use that learning again. Value compounds only when each cycle begins with more usable knowledge, stronger judgment, and less recurrence than the cycle before it.

Productive performance therefore requires more than choosing the right work and doing that work correctly. The enterprise must stay with the work through enough operating cycles for evidence to mature, errors to be corrected, routines to stabilize, and knowledge to transfer beyond the original team. The neglected productivity variable is time under coherence.

A useful managerial heuristic is that choosing the right things and doing them right may account for less of the eventual result than staying with them long enough for them to matter. The survey does not establish a measured 20-20-60 law, and the arithmetic should not be presented as one. The principle beneath it is stronger than the numbers: direction, execution, and persistence work multiplicatively, and a collapse in any one can erase the value of the other two.

Research on industrial learning has shown for decades that experience can reduce cost and improve performance, but also that knowledge depreciates when it is not retained, transferred, and reused. A factory does not benefit from yesterday's learning merely because yesterday happened. The routines, measures, permissions, and people that carry the learning must survive long enough to make the next action better than the last.

The national backdrop makes the question consequential. The United States Bureau of Labor Statistics reports that manufacturing labor productivity grew at only 0.1 percent per year from the fourth quarter of 2007 through the fourth quarter of 2019, compared with 4.7 percent during the preceding business cycle. The current cycle reached 0.5 percent through the first quarter of 2026, still well below the 2.1 percent long-term rate since 1987. Those national patterns have many causes and cannot be assigned to culture, but an enterprise that repeatedly destroys its own learning adds an avoidable operating loss to an already difficult productivity environment.

Figure 1. Coherent action allows learning to accumulate; retained learning becomes institutional capability; capability produces value repeatedly.

Culture Begins Downstream, Then Becomes Memory

Culture is often described as shared values, accepted behavior, or what people do when no one is watching. Those descriptions capture the visible pattern, but not how it formed. In operating terms, culture begins downstream. Environmental pressures meet capabilities and controls, those controls shape conditions, and people learn what the conditions make safe, rewarded, punished, and possible.

People then adapt rationally. They decide whether to raise a concern early, exercise a decision right, take a disciplined risk, protect attribution, wait for sponsorship, or remain silent until the outcome is undeniable. Repeated experience turns those choices into an operating forecast of how the enterprise will treat them when the result is uncertain.

That is where culture changes from consequence to feedback. Once the forecast is learned, it affects whether evidence is carried forward, whether permission is used, whether a decision remains closed, and whether people continue investing in work whose payoff lies beyond the next review cycle. Culture is formed by the system, then becomes part of the system's memory about what is worth preserving.

This does not make culture an independent actor or the first cause of productivity decline. People still make the decisions, and controls still shape the conditions in which those decisions become rational. Culture matters because it carries the memory of those conditions into the next decision, often long after the policy, leader, or program that created them has changed.

The Control-Experience Gap

The culture survey makes that memory visible through the gap between leaders and workers. In the leader-worker cut, 33.4 percent of leaders selected the strongest condition for learning risk being encouraged and rewarded, compared with 14.0 percent of workers. For explicit alignment between rewards and desired behavior, the comparable figures were 35.0 percent and 19.6 percent.

The gap continues where truth becomes consequential. Leaders were more likely to report that concerns were acted on quickly and transparently, 46.5 percent compared with 33.5 percent of workers, and that psychological safety was consistent across levels, 25.9 percent compared with 14.7 percent. These are self-reported perceptions, not audited control-performance measures, but the direction is consistent across the instrument.

The important conclusion is not that leaders are dishonest or workers are cynical. Leaders are more likely to see the policy, the governance design, the message, and the intent. Workers are more likely to experience the moment when the design is tested by an unfavorable outcome, an inconvenient fact, or a decision that crosses an organizational boundary.

A cultural control does not exist because leadership announced it. It exists when people can rely on it under pressure. The distance between control design and control experience is therefore not merely a communications gap. It is residual organizational risk.

Rewards become the truth system in this environment. People learn what the enterprise actually values by watching who is promoted, protected, funded, recognized, and allowed to recover from a reasoned failure. When those consequences contradict the stated culture, employees do not remain confused for long. They learn the operating rule and adapt to it.

Psychological safety works the same way. It is not mainly a belief about leadership's intentions. It is a learned forecast of what the system will do when someone introduces inconvenient truth. Concern handling is therefore a better operating test than another broad sentiment score because it observes what happens when truth enters the enterprise and carries a potential consequence.

Figure 2. Leaders more often report the controls they designed. Workers report whether those controls remain dependable under consequence.

Why the Return Would Look Like a Threshold

The June meta-analysis reported a striking relationship. Organizations above a cultural maturity threshold were estimated to have roughly 5.3 times higher odds of being clear market outperformers than organizations below it. That is not the same as a fivefold increase in productivity or value, and the source workbook available for this review was incomplete, so the exact estimate should remain directional until the calculation is reconstructed and replicated.

The exact multiplier is not needed to explain why a nonlinear return is plausible. The relevant controls are complementary. Psychological safety accomplishes little when the reward system punishes the person who exposes a problem. Clear decision rights accomplish little when senior leaders reclaim authority after an unfavorable outcome. Priority continuity accomplishes little when a leadership change discards the evidence, ownership, and operating logic that made the priority rational.

Below a coherence threshold, the enterprise can possess many good practices and still fail to retain their combined value. Each strength is canceled by another weakness, reliability remains local and temporary, and the organization has pockets of capability rather than dependable enterprise capability. The system improves in places, then resets at the boundary, under pressure, or across time.

Above the threshold, the controls begin to reinforce one another. Truth arrives earlier because concern handling is predictable. Decisions remain closed because authority is clear. Learning survives early failure because rewards distinguish disciplined experimentation from carelessness. Priorities survive leadership changes because the causal logic is held by the institution rather than by one sponsor.

The threshold hypothesis is therefore not that one more values program suddenly creates five times the result. It is that productive capability begins to accumulate only after enough operating promises become mutually reliable for coherent action to survive pressure and time.

Figure 3. The coherence threshold is an illustrative causal hypothesis, not a fitted estimate. The reported 5.3x relationship remains directional until independently reconstructed.

The Cascade That Eats the Capacity to Recover

A reset does more than erase the value of the initiative that was abandoned. It consumes the capacity required to make the next initiative succeed. People spend time reconciling measures, rebuilding sponsorship, repackaging bad news, defending prior decisions, recreating teams, and explaining why the same problem has returned under a different name.

That mitigation load is easily mistaken for productive work. It appears as meetings, analysis, action registers, governance reviews, escalation calls, expedited freight, overtime, rework, additional controls, and another leadership search. Each response may be rational in isolation, yet together they remove engineering attention, leadership bandwidth, maintenance windows, and operating capacity from the capabilities that could have prevented recurrence.

The causal cycle then turns against the enterprise. Weak or inconsistent controls create unstable conditions. People adapt through delay, hedging, escalation, attribution protection, and risk avoidance. Those behaviors produce late truth, slower decisions, recurrence, value leakage, and turnover. The consequences consume capacity, leaving less capacity to strengthen the controls in the next cycle.

This is why cultural decline becomes a cascade rather than a static weakness. The organization is not only losing the gain. It is spending the trust, memory, attention, and improvement capacity required to recover from the loss. Each restart begins from a weaker position, and the workforce learns to wait for the next reset before fully committing to the current one.

The negative compounding is real even when positive compounding never begins. Repeated resets create more mitigation, mitigation consumes more capacity, lower capacity weakens control reliability, and weaker controls create more consequences. The enterprise can remain busy for years while the ability to improve is quietly being spent.

One Way Coherence Breaks: Collaboration Without Closure

Collaboration is usually treated as an unqualified good, and complex industrial decisions often require knowledge from operations, maintenance, engineering, quality, supply chain, finance, safety, and the people closest to the work. A narrow decision made without necessary context can be fast and wrong. The survey does not show that collaboration causes decision latency, and reported latency was similar across collaboration levels.

The defensible finding is narrower: better collaboration did not reliably remove latency. One plausible mechanism is the burden of unresolved plurality. More people can contribute more observations, but more observations make inference harder when no one owns synthesis, no evidence threshold closes the discussion, and every consulted party behaves like an approver or possible veto holder.

The answer is not less intelligence. It is a decision architecture that separates contribution from authority. Collaborate broadly enough to gather the context that can change the decision, then decide narrowly enough that one accountable owner can act within explicit bounds.

This is how collaboration connects to the reset. When decisions remain provisional, priorities remain provisional. Work cannot accumulate through repeated cycles when any meeting, sponsor, or adverse result can reopen the decision without new evidence. Participation creates context. Authority converts context into action, and closure gives action enough continuity to produce learning.

When Continuity Becomes Stubbornness

The strongest objection is that persistence can be dangerous. Organizations sometimes keep doing the wrong things because culture protects tradition, a leader refuses to admit error, or the operating system becomes too rigid to respond to evidence. A company can compound waste, defend obsolete methods, and call stubbornness continuity.

That objection is correct, and it marks the boundary of the argument. The goal is not to preserve every initiative or prevent every reset. The goal is to preserve coherent action until evidence, rather than politics, fatigue, sponsorship loss, or leadership fashion, justifies changing it.

A mature culture does not merely continue. It remembers why the work began, keeps measures connected to operating reality, allows contrary evidence to arrive early, and gives an accountable owner the authority to correct the path. It can stop bad work faster because truth does not have to survive a political trial before it becomes actionable.

Without cultural continuity, good work is abandoned before evidence can mature and bad work may persist because no one is safe enough to challenge it. With cultural continuity, the enterprise can distinguish disciplined persistence from institutional stubbornness. It can change the action without discarding the learning, evidence, ownership, and operating memory the action produced.

What Boards Should Measure

Boards and executive teams cannot manage this problem through engagement scores alone. They need measures that reveal whether coherent action is being preserved, whether learning is becoming institutional capability, whether authority survives consequence, and whether the enterprise is lowering recurrence rather than improving the explanation of recurrence.

The first measure is time from signal to decision and from decision to stabilized action. The second is the rate at which decisions are reopened without materially new evidence. The third is the percentage of improvement routines and strategic priorities that survive a leadership change without losing their causal logic, evidence base, operating memory, and accountable owner.

The fourth measure is recurrence across three, six, and twelve months. The fifth is the share of leadership, engineering, maintenance, and operating capacity consumed by mitigation rather than prevention and improvement. The sixth is the transfer rate: how often learning created in one team or site changes the next decision somewhere else.

The board should also compare the controls leaders believe they have installed with the conditions workers actually experience. Where those views diverge, the difference is not simply a communications problem. A control that does not hold under consequence cannot protect the outcome it was designed to shape, and a capability that exists only under one leader is not yet an enterprise capability.

The Work Must Survive the Calendar

The board can keep buying motion and calling each restart transformation. The plant will keep relearning what it already knew. Each reset will consume more of the capacity required to recover. The work either survives long enough to become capability, or the enterprise pays to begin again.

References, Sources, and Intellectual Lineage

This article draws first on the internal industrial culture survey and the June cultural artifacts meta-analysis supplied for this work. The survey provides directional evidence about the control-experience gap, reward-behavior alignment, concern handling, psychological safety, collaboration, decision rights, risk support, and cultural priority stability, while also carrying important limits: most measures are cross-sectional self-reports, several performance fields contain routing or interpretation problems, and the source workbook behind the reported 5.3 times outperformance estimate was incomplete in the review package. The threshold result is therefore treated as a hypothesis-strengthening association rather than a settled causal estimate. The productivity backdrop comes from the United States Bureau of Labor Statistics, including its March 2026 Monthly Labor Review analysis and its June 2026 long-term manufacturing productivity series, which report average annual labor productivity growth of 0.1 percent from the fourth quarter of 2007 through the fourth quarter of 2019, 0.5 percent in the current cycle through the first quarter of 2026, and a 2.1 percent long-term rate since 1987. The mechanism of accumulation and loss is informed by Linda Argote, Sara Beckman, and Dennis Epple's 1990 study of the persistence and transfer of learning in industrial settings; John Dutton and Annie Thomas's work on progress functions and the managerial conditions that shape learning returns; James March and Herbert Simon's account of bounded rationality, uncertainty absorption, and organizational choice; and Amy Edmondson's 1999 field study connecting psychological safety with learning behavior in manufacturing teams. W. Edwards Deming's work on fear, merit rating, variation, and management responsibility helps distinguish system design from individual blame, while Charles Goodhart's formulation of target-driven measurement failure clarifies why an indicator can displace the operating reality it was meant to represent. The argument extends Michael Carroll's prior article, When Culture Becomes the Escape Hatch, by resolving the apparent tension between culture as a downstream adaptation and culture as an active continuity mechanism: culture is formed by repeated operating conditions, then becomes the organizational memory that shapes what people preserve, challenge, reopen, and carry forward. It also advances Carroll's One-Degree Architecture, decision-latency framework, risk-centered causal networks on trust and agency, and the capacity cascade through which the consequences of weak controls consume the capacity required to restore them.