Cost Curve Starts too late

Your Cost Curve Starts Too Late

Why 1 Percent a Year Is Not a Talent Problem. It Is a Decision and Permission Problem Month twenty-four. The leadership team is staring at a chart that should embarrass everyone and yet somehow embarrasses no one. Cost per labor hour consumed is down, but barely. The line is polite. It slopes in the right direction. It does not move the P and L in a way a board would call meaningful. Someone says the sentence that ends more value than any competitor ever took. “We are improving. About one percent.” That sentence is how slow decline gets rebranded as progress. The uncomfortable truth is not that people are lazy. The uncomfortable truth is that the curve started too late. Not because the team did not try. Not because the plant did not run kaizens. Not because the engineers did not have ideas. It started too late because the organization spent its first year deciding what it already knew and then spent another year asking permission to do what it had already decided. By the time execution began, the window where the improvement would have mattered most was gone. Most companies live in the one to two percent band. Some will argue the economy only gives you that. Some will say labor markets, inflation, supply volatility, and regulation. Those are real

forces. They are not the explanation for the flat curve. The flat curve comes from inside the fence. The flat curve comes from the time it takes to decide, the time it takes to get permission, and the time it takes to execute. Execution time is real time. Steel has lead time. Installations require outages. Training takes repetition. Culture changes at the speed of lived experience. In major companies, anything that truly moves the P and L often takes eighteen to twenty four months to execute. That is not a failure. That is physics plus scale. The failure is the year before execution even starts. The year lost in decision and permission. That part is compressible. It is also where most of the value goes to die.

The Chart That Tells the Truth

Start with a simple measurement. The y axis is percent cost improvement per labor hour consumed. That phrasing matters. It forces clarity. It is not cost per unit, which can be gamed by mix. It is not total cost, which can be hidden by volume. It is cost improvement per labor hour consumed. It asks a hard question. How much cost did the organization remove for every hour of labor it still had to spend to run the business. Time sits on the x axis. Not as calendar decoration, but as the conveyor belt of consequence. The x axis is not only “how long did it take.” It is what the company did during that time. There is time to decide. There is time to get permission. There is time to execute. Then there is time for outcomes to show up in the financial statements. Most companies assume the curve is a talent story. They assume the slope comes from better people doing better work. That belief contains some truth. It is also incomplete in the way incomplete ideas always are. It frames what is visible and ignores what is causal. If the belief is that exemplars should produce five to six percent improvement per year, then the curve is not a motivational poster. It is a system. It is the output of capability, capital, and time. Here is the first correction that changes everything. The y axis is not driven only by the effect of talent across capability states from poor to fair to good to excellent. The y axis is also driven by access to cash. Not abstractly. Mechanically. A business with no access to cash behaves differently than a business with full access. A business under restrictive access behaves differently than a business under semi restricted access. Access to cash changes what can be improved, how quickly assets can be renewed, which bottlenecks can be removed, and whether the organization can sustain capital and assets rather than harvest them. Then there is the second correction. The effect of talent is not only the ability to run operations well. It is the capability to use that cash and capital well. It is the capability to improve and sustain capital and assets so gains do not evaporate. Capital that is spent without capability becomes future cost. Capability without capital becomes stalled intent. The y axis is the result of

both.

So the chart is not one line. It is a family of lines, and the gap between them is not explained by effort. It is explained by two multipliers and one clock.

The Two Multipliers Hiding Inside “Productivity

” Think of the slope as the product of two forces. One force is capability. That is where talent actually shows up. Not in resumes, but in what the organization can reliably do. A poor capability state produces weak, fragile improvement. A fair state produces local wins that do not hold. A good state produces repeatable improvement in defined domains. An excellent state produces compounding improvement because learning is captured and reused, and because leaders can make decisions without rearguing first principles every month. The second force is capital access. It is not enough to say, “we have cash.” Cash on a balance sheet is not the same as access to deploy it. Access is governed. Access is constrained by covenants, internal hurdle rates, allocation politics, risk committees, and fear. That governance can be rational. It can also become a machine that protects yesterday at the expense of tomorrow. Now add the missing connective tissue. Capability determines how well capital is used. Capital determines how quickly capability can be scaled and sustained. This is why cost improvement is never just “a talent problem.” It is a talent and capital utilization problem.

A company in a poor capability state with full capital access can spend a fortune and still stay stuck in the one to two percent band. It purchases systems it cannot operationalize. It buys automation it cannot maintain. It funds projects that never make it into standard work. It converts cash into complexity. The finance function learns the wrong lesson and tightens access. The next budget cycle becomes restrictive. The company says it has a capital problem. It does not. It has a capital competence problem. A company in an excellent capability state with no capital access can still improve, but it will hit ceilings quickly. It can tighten planning. It can reduce rework. It can stabilize flow. It can improve maintenance discipline. It can remove waste that is behavioral and procedural. Then it reaches the moment where the next step requires asset renewal, line redesign, or technology. Without capital access, the curve flattens. The company says it has a talent problem. It does not. It has an access problem. That is why the y axis must include both. If you want to visualize it, picture the same capability state under different cash access states. The slope changes even if talent does not. Then picture the same cash access state under different capability states. The slope changes even if cash does not. Now introduce time.

Where Time Hides the Real Constraint

You already gave the most important empirical anchor. In major companies, anything that moves the P and L often takes eighteen to twenty four months to execute. That is the execution span. That is build time, install time, train time, stabilize time, and the time to get the new way of working into muscle memory. You also gave the second anchor. Decision and permission often take another year. That is the compressible part. So the x axis is not one bar. It is three. Time to decide. Time to get permission. Time to execute. The only one that is not easily compressible is execution. Even excellent organizations cannot compress reality past certain limits. A new maintenance program still requires cycles to take hold. A supply chain redesign still requires suppliers to adapt. An asset still has lead times. Decision time and permission time are different. They are not physics. They are architecture. Decision time is mostly lost in three places. First, the organization does not know what it believes about cause and effect, so every decision becomes an argument about the model, not an argument about the choice. Second, the organization cannot surface and reconcile competing incentives, so decisions get delayed until the conflict exhausts itself. Third, the organization has

trained itself to treat decisions as events rather than as outputs of an operating system. When decisions are events, they require calendars, meetings, and presentations. When decisions are outputs, they require evidence and authority. Permission time is lost in a different set of places. Capital allocation systems behave like courts. They demand proof. They demand certainty. They demand polished narratives. They demand alignment. None of those are wrong in principle. They become wrong when they are used to delay action until certainty arrives, because certainty does not arrive on schedule. In that delay, the business pays a tax. It pays in overtime. It pays in churn. It pays in quality losses. It pays in opportunity cost. Then, when the capital is finally approved, the organization frames the coming eighteen to twenty four months as “the work.” It is already late. Here is the central claim of this article. Your cost curve starts at the moment you begin deciding and seeking permission, not at the moment you begin executing. If you treat the year of decision and permission as overhead, the curve starts too late. If you treat that year as the first year of execution, the curve starts on time. This is why many companies never reach five to six percent. They are trying to average exemplar improvement while spending the first third of the window not moving. A simple illustration makes it obvious.

Decision + permission often consume Year 0 to Year 1. Execution often consumes Year 1 to Year 3.

If the curve only starts moving in Year 2, the average cannot reach exemplar levels. That diagram is not a math trick. It is a leadership mirror.

The Hidden Link Between the Axes

Now we get to the claim you already see but most people miss. There is linkage between the x axis and the y axis. You cannot compress time to decision and permission without changing the effect of talent. That is the coupling. Why. Because speed is not only a scheduling question. Speed is a quality question. When you compress decision time, you reduce the time available for debate, polishing, and politics. That only works if the organization has a higher quality decision process. That process lives in capability. It lives in how evidence is generated, how causal claims are tested, how risk is bounded, and how accountability is assigned. When you compress permission time, you reduce the time available for socializing, lobbying, and consensus building. That only works if the organization has a higher quality permission system. That system lives in capital competence. It lives in how projects are scoped, how value cases are made real, how assets are sustained, and how capital risk is managed without paralyzing the enterprise. This is the reason the “talent problem” narrative is seductive and wrong. Leaders feel that talent is the only lever. They believe hiring and training is the path to five to six percent. Talent matters. But talent, by itself, does not fix decision and permission architecture. In fact, in many companies, stronger talent makes the delay worse, because talented people create more analysis, more alternatives, and more debate. Without architecture, talent increases the surface area of uncertainty. So the correct framing is not “hire better people.” The correct framing is “build the system that lets existing talent decide and act faster, and build the capital competence that lets the business fund change without requiring a year of courtroom procedure.” That sounds like governance. It is, and that is why it is hard. Governance is where power and risk live. Governance is where most companies choose comfort over speed and then call the outcome “realistic.”

What “Access to Cash” Really Means on the Y Axis

Access to cash spans a spectrum. At one extreme there is no access. The business is constrained by debt, covenants, survival, or external shocks. Every dollar is triage. At the other extreme there is full access. The business has balance sheet strength, confidence, and freedom to invest. Between those extremes sit restrictive access and semi restricted access. That is where most large companies live most of the time.

Those categories are not labels. They describe behavior. No access forces the business to harvest assets. It stretches maintenance. It delays replacement. It defers training. It keeps running what should be renewed. In the short term, it can protect cash. In the medium term, it increases labor hours consumed per unit of output, because the system requires more human effort to compensate for degrading assets. That pushes the y axis down. Restrictive access forces the business into small bets only. It funds compliance. It funds safety. It funds urgent fixes. It does not fund redesign. The organization learns to chase symptoms. It improves in patches. The slope becomes a sawtooth. That also pushes the y axis down, because gains do not hold. Semi restricted access funds larger moves but with heavy gating. It produces a portfolio of projects that are always starting and stopping. Work gets half done. Teams get reassigned. Every restart adds waste. Labor hours consumed rise while results lag. The curve gets delayed. The y axis looks like underperformance even when effort is real. Full access lets the organization place bigger bets and sustain them. It can renew assets on a schedule rather than on a crisis. It can fund capability building that does not pay back in the quarter but changes the slope for years. Full access does not guarantee improvement. It only removes one ceiling. Now integrate capability. A company with full access and poor capability converts cash into decay. A company with restrictive access and excellent capability can still produce meaningful improvement by selecting the right moves and sustaining assets intelligently. But even that company will reach a ceiling if it cannot renew capital at the pace the system demands. So the y axis is a joint output. It is cost improvement per labor hour consumed as shaped by capability and capital access, and by the competence to deploy capital into assets that stay improved.

The Part Everyone Underestimates. Sustaining Capital and Assets

Most improvement programs assume gains are permanent. They are not. Gains require upkeep. Assets drift. Processes drift. People drift. Turnover drains muscle memory. Demand changes. Quality challenges reappear in new forms. Sustaining gains is not a morale issue. It is an asset and system issue. If maintenance is deferred, labor hours consumed rise. If training is not repeated, rework rises. If engineering standards are not held, variation returns. If capital renewal is delayed, downtime rises. Every one of those pushes cost improvement downward or reverses it.

This is why the capability effect of talent must include the capability to sustain capital and assets. Many organizations build the skill to improve. Fewer build the skill to sustain. Sustainability is where exemplar curves separate from typical curves. In typical companies, a project delivers a short spike, then fades. The curve looks like progress, then reverts. Leadership calls it a culture issue. It is often a sustainment architecture issue. Sustainment requires stable standards, clear ownership, and recurring funding for the boring work that protects gains. When capital access is restrictive or chaotic, sustainment becomes optional. Optional sustainment becomes predictable erosion. Predictable erosion becomes one percent. If you want five to six percent, you cannot treat sustainment as a side activity. It is part of the slope.

Why Decision Time Exists. And Why It Can Shrink

Decision time exists because uncertainty exists. People do not delay because they like delay. They delay because they do not trust the model, they do not trust the data, they do not trust each other, or they do not trust what will happen to them if they are wrong. That last clause is the one most executive teams avoid saying out loud. Decision time can shrink, but it shrinks only when the organization changes what it means to decide. In many firms, deciding is a performance. It is a meeting. It is a deck. It is a request for endorsement. It is also a negotiation of blame. In exemplar firms, deciding is closer to engineering. It is evidence, assumptions, bounds, and accountability. It is clear who decides. It is clear what evidence is required. It is clear how risk is bounded. It is clear what happens after the decision. That clarity reduces the psychological load. When the psychological load drops, time drops. This is the first place where talent must change. The talent required to move faster is not the talent of “smart analysis.” It is the talent of building decision objects that carry evidence and bounds so decisions do not require theater. When that talent is absent, speed attempts fail. Leaders push for faster decisions and get worse decisions. That creates backlash. Backlash increases permission gates. The system slows down further. The curve starts even later. So the coupling is real. You cannot compress x without changing the talent effect that drives y. Speed is purchased with capability.

Why Permission Time Exists. And Why It Can Shrink

Permission time exists because capital is scarce and risk is real. It exists because the organization learned, often painfully, that uncontrolled spending and uncontrolled change can destroy value.

Permission time becomes toxic when it is designed to prevent failure instead of designed to produce outcomes with bounded risk. Preventing failure sounds wise until you measure the failure created by delay. The best way to see permission time is to trace how a project gets funded. The organization demands certainty of savings before spending. But certainty is often only available after learning. Learning often requires spending. So the system creates a paradox. It demands proof that can only be produced by the very investment it is withholding. To break the paradox, teams inflate claims. They polish narratives. They promise outcomes they cannot control. The finance function responds by increasing scrutiny. That increases delay. The cycle feeds itself. In that cycle, access to cash becomes less about actual cash and more about trust. Trust in the teams. Trust in the numbers. Trust that assets will be sustained. Trust that the organization will not abandon the work halfway through and then ask for more money later. That trust is the capital competence factor again. It is capability expressed in financial form. Permission time shrinks when permission becomes a staircase, not a wall. It shrinks when the organization defines bounded commitments that can be approved quickly, with clear stop rules and clear ownership. It shrinks when it funds learning deliberately instead of pretending learning is free. It shrinks when it separates two questions that most companies mash together. The first question is whether the idea is plausible. The second question is whether the organization is capable of executing it. When those questions are separated, permission becomes cleaner and faster. Again, speed is purchased with capability. The permission system can move only as fast as the organization’s ability to use cash responsibly and sustain assets afterward.

The Real Reason Execution Time Feels Sacred

Execution time feels sacred because it is hard. It is where real work happens. It is where people must change behavior. It is where operations are disrupted. It is where the project can embarrass leaders. So most leaders treat execution time as the main story. They obsess over project plans. They create dashboards. They run weekly reviews. They treat the eighteen to twenty four months as the defining grind. That focus is understandable. It is also incomplete. The execution window is already long. If you add a year of decision and permission, the total cycle becomes two and a half to three years before the P and L reflects the change. That means the organization is always working on yesterday’s problems. It is always funding yesterday’s ideas. It is always reacting to yesterday’s market signals. Then it wonders why competitors feel faster. This is why the curve starts too late.

An exemplar firm does not compress execution by pretending reality is different. It compresses the front end so execution starts earlier and starts cleaner. It spends less time debating and more time running bounded experiments that build evidence. It does not treat the year before execution as overhead. It treats it as the first year of conversion.

A Second Illustration. The Cost Curve Start Line

Picture two companies that both require eighteen to twenty four months to execute major initiatives. Company A spends a year deciding and securing permission. Company B compresses that year so that decision and permission take months, not a year, and it does so by changing capability and capital competence. The difference is not speed for its own sake. The difference is where the curve begins.

Company A begins real movement late in execution. Company B begins real movement at the start of execution, because the front end was compressed. If you want five to six percent, the curve cannot wait until late execution to move. It must move earlier, and it must hold.

That is the simplest explanation for why one percent is not a talent problem in the way most leaders mean it. It is a system design problem.

What Exemplars Actually Do Differently

Exemplars do not have magical workers. They have an operating system that changes how talent behaves and how cash behaves. They define decision rights clearly. Not in a policy document that nobody reads, but in the lived reality of what gets decided without escalation. Escalation is where time goes to die. When escalation becomes the default, every decision inherits the calend ar of the most senior person in the chain. They standardize what evidence looks like. Not in the form of bigger decks, but in the form of decision objects that are consistent, small enough to be consumed, and explicit about assumptions. When evidence is standardized, the organization stops rearguing method and starts arguing substance. They bound risk. They stop trying to make decisions perfectly safe, and instead make decisions safe enough through clear bounds, stop rules, and ownership. That reduces the emotional burden of deciding. It also reduces the political burden, because blame games thrive in ambiguity. They treat permission as an investment discipline, not a courtroom. They fund learning deliberately. They separate learning capital from scaling capital. They demand different evidence at each stage, and they do not demand late stage evidence at the early stage. They build capital competence as a capability. That means teams learn to scope projects realistically, to track benefits honestly, to sustain asset health, and to show what was learned when outcomes fall short. When teams can do that, finance becomes less defensive. When finance becomes less defensive, access to cash expands. When access expands, the y axis can climb. Notice what this implies. The x axis improvement requires y axis capability improvement. The y axis improvement often requires x axis compression, because the organization must stop losing years before work begins. That is the coupling again.

The Counterexample That Proves the Rule

There are cases where a company with limited cash access still improves meaningfully. It standardizes work. It reduces unplanned downtime. It improves quality at the source. It trains supervisors to solve problems rather than escalate them. It tightens planning. It improves scheduling discipline. Those moves can produce cost improvement without large capital outlays. They can also improve labor hour consumption directly.

That counterexample matters because it prevents a lazy conclusion. The conclusion is not “give people money and the slope increases.” Money without capability can reduce the slope. The conclusion is that capability can drive improvement even under constrained cash access. But the same counterexample also proves the rule. That company eventually hits a ceiling. It reaches the point where asset renewal is required to sustain gains. Without capital access, the curve flattens. If leadership misreads that moment as a talent shortfall, it starts a morale campaign instead of fixing access. If leadership reads it correctly, it builds capital competence, rebuilds trust, and earns greater access. So cash matters, but not as a substitute for capability. Cash matters as the amplifier of capability and as the protector of sustainment.

The P and L Reality That Should Change How Leaders Think

Most executive teams still treat improvement as a project portfolio. Projects are selected annually. Funding is allocated annually. Reviews happen monthly. Benefits are measured quarterly. This creates a mismatch. The execution cycle is eighteen to twenty four months. The decision and permission cycle is a year. The measurement cycle is a quarter. The cadence is misaligned, so leaders get anxious and intervene. Intervention adds friction. Friction increases delay. Delay pushes benefits out. The board sees delay and demands more control. Control adds more permission gates. The loop continues. One percent becomes predictable. If you accept the real cycle time, then the only way to raise the average slope is to start earlier and to stop losing the year of decision and permission. That is the compressible part you already named. This is why “Your Cost Curve Starts Too Late” is not a slogan. It is a diagnosis. It tells you where to look. It tells you what to measure. It tells you what to fix. Measure the timestamps. When did the signal first appear. When did the organization agree on what it meant. When was the decision made. When was permission granted. When did execution begin. When did the outcome show up in labor hours consumed and cost. If those timestamps show that decision and permission consume about a year, then the organization has located the biggest controllable delay in its improvement system. Then look at capability. If the organization cannot make decisions quickly without fear, then capability is missing in decision quality, not in work ethic. If the organization cannot grant permission quickly without panic, then capability is missing in capital competence, not in cash itself. This is where the curve is rebuilt.

A Falsifiable Claim Worth Putting on the Table

If the time to decide and the time to get permission together consume another year, and if execution that moves the P and L still requires eighteen to twenty four months, then a company will struggle to sustain a five to six percent cost improvement slope because the curve begins moving too late in the cycle. That claim is falsifiable. Track the timestamps and track the slope. If a company maintains a year of decision and permission delay and still sustains exemplar slope over multiple years, then the claim fails. If the data shows what most leaders already suspect, then the claim holds. This is the kind of claim leaders should welcome. It turns frustration into a testable mechanism instead of a cultural sermon.

How to Start the Curve Earlier Without Breaking the Company

The hardest part is that the front end is not just process. It is power. Decision rights determine who matters. Permission systems determine who controls cash. Changing either will trigger resistance, because resistance is the natural defense of authority. So the path is not to announce speed. The path is to build trust by designing bounded arenas where speed is safe and learning is captured. That starts by selecting domains where outcomes are measurable in labor hours consumed and cost, and where risk can be bounded. When bounded arenas exist, leadership can predefine what evidence is required, what investment is allowed, and what stop rules apply. That changes permission from a wall into a sequence of steps. It also changes decision making from performance into engineering. As those arenas produce sustained gains, the finance function sees something it rarely sees. Evidence that capital is being used well and that assets are being sustained. That evidence increases access. Not because finance becomes generous, but because finance becomes confident. This is the moment where the y axis and the x axis start pulling together instead of against each other. Capability improves. Access improves. Decision and permission compress. Execution starts earlier. Gains show up sooner. The curve starts on time.

What This Means for Leaders Who Still Think This Is “Just Operations

” The slope of cost improvement per labor hour consumed is not an operations metric. It is a governance metric expressed in operational terms. If you are a CEO, the question is whether the enterprise has an operating system for deciding and acting that matches the speed of its environment. If it does not, strategy is theater.

If you are a CFO, the question is whether capital allocation is structured to fund learning and sustainment rather than structured to demand certainty and then punish teams for being human. If it is structured the second way, you will protect cash and still lose value. If you are a COO, the question is whether capability is being built where it matters most. Not only on the shop floor. In the decision system and the permission system. Because those systems determine how quickly the shop floor can be improved and how long the improvement will hold. If you are a board member, the question is simple. When you hear “one percent,” do you ask for more projects, or do you ask for timestamps. Do you ask how long it took to decide, how long it took to grant permission, and what architecture is being changed so the next cycle starts earlier. The curve starts too late when leaders accept slow front ends as normal. It starts on time when leaders treat decision and permission as part of execution, because that is what they are.

The Closing Reality

Most companies will keep talking about talent as the reason they cannot improve faster. That story is comforting because it implies the solution is hiring and training. Hiring and training matter. They will not, by themselves, compress a year of decision and permission. One percent is often the signature of an enterprise that begins execution after the competitive clock already started. Five to six percent is often the signature of an enterprise that begins the work at the moment the signal appears, because decision and permission are designed as fast, evidence-based systems, and because access to cash is earned through capital competence and sustained assets. Your cost curve does not start when the project plan kicks off. It starts when you decide and when you grant permission. If that part takes a year, the curve will start too late.

References

Labor productivity growth averages and context for the post 2005 slowdown are discussed in a Bureau of Labor Statistics Monthly Labor Review analysis. The slowdown in manufacturing productivity growth and similar post 2005 averages are summarized by Brookings. OECD experimental estimates on recent labor productivity growth, and discussion of how tighter financial conditions and constrained access to credit can impede investment and productivity, appear in the OECD Compendium of Productivity Indicators 2025 and related OECD materials. Firm level evidence linking access to finance, productivity, and investment is presented in an IMF working paper on financial constraints. The durability problem in corporate transformations is illustrated by survey based findings on sustaining performance improvements over time. Foundational theory on bounded rationality and decision making in organizations is treated in Herbert A. Simon’s work and scholarly summaries. Transaction cost explanations for why firms

organize decisions and permissions internally, and the costs that rise as coordination expands, originate in Coase’s “The Nature of the Firm” and later transaction cost economics literature. The option value of waiting under uncertainty, and why delay can be rational until it becomes value destructive, is treated in Dixit and Pindyck’s Investment Under Uncertainty. Deming’s emphasis on management systems, predictability, and sustaining improvement is treated in Out of the Crisis.