This is not a staffing problem or a management problem but an infrastructure problem, and it is far more expensive than it appears because the cost is not the task itself but what the task is displacing.

The mechanism that puts the wrong work on the right people

In most organisations, the allocation of operational tasks follows an unofficial rule: when something needs to get done and there is no clear system for doing it, it goes to the person most likely to do it correctly. That person is almost always one of the most capable people in the business, which is why the task gets done reliably, and also why it stays with them indefinitely.

The mechanism is self-reinforcing - The capable person completes the task well, which confirms that they are the right person for it, which means it comes back to them the next time it needs doing. Over months and years, senior people can accumulate a significant portfolio of tasks that arrived this way: each one small enough to seem manageable, together consuming a meaningful fraction of the time they should be spending on work that actually requires them.

The organisation rarely notices this happening because the output is maintained. The work gets done, the capable person absorbs the load without visible complaint because absorbing load is part of how they became capable, and the cost sits in the work that does not happen because of the hours this person is spending on tasks a well-designed system could handle automatically.

What this costs in practice

The direct cost of misallocated senior time is straightforward to calculate once it is made visible. A person earning $150,000 per year costs the business approximately $75 per hour, meaning five hours per week on work that does not require their skills costs $375 per week, $1,500 per month, and $18,000 per year, and that is before accounting for the opportunity cost of the thinking and judgment that did not happen in those hours.

Across an organisation of 50 to 200 people, where this pattern is replicated across multiple senior staff, the total misallocation of skilled capacity is typically significant, often running to several hundred thousand dollars per year when the full picture is assembled honestly.

The harder cost to quantify is the one that does not appear as an expense at all: the strategic work that was deferred, the client relationship that was not developed, the operational improvement that was never examined because the person who would have examined it was occupied with a task that a system should have been handling for the past three years.

The categories of misallocated work that appear most consistently

Across the businesses we work with, certain categories of misallocated work appear regardless of industry or scale.

The first is exception handling. Most operational processes produce exceptions, and exceptions require judgment, but in many organisations the volume of exceptions reaching directors and senior leaders is far higher than it needs to be because the upstream process is poorly designed, generating cases that a better-built system would triage and resolve automatically before they surface as decisions requiring senior intervention. The leader becomes a de facto quality control layer for a process that should never have reached them, and the cost is not just their time but the interruption cost of context-switching out of higher-order work every time an exception lands.

The second is performance and operational reporting. In many businesses, the work of assembling board packs, operational dashboards, and divisional performance summaries involves significant manual effort from people whose value to the business lies in interpreting what the data means, not in producing it. The data exists across multiple systems, and the effort of extracting, reconciling, and formatting it into something a leadership team can act on falls to senior finance or operations staff who spend two or three days per month on work that contains no analytical judgment whatsoever. That time is recoverable in full.

The third is cross-functional coordination and governance that should be handled by infrastructure. In organisations where significant work crosses team or entity boundaries, the coordination of handovers, approvals, and status visibility often defaults to senior people because they have the organisational authority to make things move. The activity looks like leadership because it involves senior people and consequential decisions, but much of it is process administration: chasing sign-offs, consolidating inputs, and maintaining visibility over workflows that a well-designed system should be tracking automatically. The senior person is not adding judgment to these interactions but providing organisational friction to compensate for the absence of infrastructure.

Why it persists

The reason this pattern persists in otherwise well-run organisations is that addressing it requires stepping back from the day-to-day to examine it, which requires exactly the time and attention that the pattern itself is consuming.

There is also a deeper, more human reason. The people carrying this work are often reluctant to raise it because doing so feels like complaining about their own effectiveness, and the organisations they work in have implicitly rewarded their willingness to absorb it. Asking a capable person to stop doing something they do well in order to free them for work that has not yet been defined requires a degree of structural confidence that many leadership teams do not have available when they are running at capacity. This is why these kinds of problems benefits from an external perspective, because the internal people best placed to solve the problem are the ones most embedded in the conditions that created it.

The structural fix

The resolution is not a conversation about priorities or a reorganisation of responsibilities. It is the construction of infrastructure that removes the task permanently, by building a system that handles it reliably without requiring senior involvement.

When this is done well, the capable person does not lose work but gains back the hours that were being consumed by tasks beneath their skill level, and those hours become available for the judgment-intensive work that only they can do. The organisation does not get smaller when this happens: it gets more of what it was already paying for.

The question worth asking honestly is not whether this pattern exists in your business, because it almost certainly does, but how many hours per week your best people are currently spending on your worst work.

Karst is an operations and technology firm that designs and implements Intelligent Operating Systems to deliver commercial outcomes. Built by operators. Powered by AI. Measured in outcomes.

This is not an implementation failure in the dramatic sense. Nobody lost significant money, nobody made a catastrophic decision, and the technology worked as described. What failed was the gap between a working system and a system that actually changes how the business operates, and that gap is where most technology investment goes astray.

The reason is almost never the technology. J.P. Morgan Asset Management's research found that while nearly 90% of companies have invested in AI technology, fewer than 40% report measurable gains, and the explanation they give is consistent: most organisations apply AI to discrete tasks rather than redesigning how work gets done. Good implementation is less about selecting the right tool and more about building the right conditions for the tool to land in.

The condition that determines almost everything else

Before any system is built or any tool is selected, there is a prior question that determines whether the implementation will succeed: is the problem defined in commercial terms.

Not in technology terms, not in process terms, but in commercial terms, meaning what is this costing the business right now, what does fixing it free up, and what does a successful outcome look like as a number rather than a description.

When this question is answered honestly before the work begins, the implementation has a success condition, everyone involved knows what they are building toward, the technology selection follows from the outcome rather than preceding it, and the business has a basis for evaluating whether the investment paid off. When this question is skipped the implementation has no anchor, and a system that technically functions correctly can still fail to change anything of commercial significance, which is the single most common reason implementations underdeliver: not a technical failure but a definitional one, where the business invested in capability before it had defined the return.

What the sequencing looks like when it is done well

The implementations that produce durable commercial outcomes follow a consistent sequence, and it is worth describing what that sequence looks like from the inside.

The first stage is commercial scoping, where the problem is quantified, the cost of the current state is established, and the target outcome is defined. This is not a discovery workshop in the consulting sense but a direct conversation about what the business is trying to change and what changing it is worth, typically taking hours rather than weeks and producing a brief that defines success before any system design begins.

The second stage is process mapping, not as an end in itself but as a tool for identifying exactly which steps in the current workflow require human judgment and which do not, because this distinction determines what gets automated and what does not and getting it right at this stage prevents the most common implementation error: automating the wrong things and leaving the actual cost-drivers in place.

The third stage is a contained build, beginning with the highest-value, lowest-risk component of the system and making it work properly before expanding. Businesses that try to implement comprehensively from the start tend to produce systems that are theoretically complete and practically fragile, while starting contained and expanding from a working foundation produces something the organisation can trust, and trust is the precondition for genuine adoption.

The fourth stage is embedding, which is where most implementations fail even when everything before it has gone well. A system that the organisation does not use is a system that does not exist commercially regardless of how well it functions technically, and embedding means changing the workflow rather than just providing the tool, staying engaged until the new way of working has replaced the old one rather than until the build is technically complete.

What makes adoption actually happen

Adoption is the part of implementation that technology vendors consistently underestimate and operators consistently know determines everything.

The organisations where new systems embed quickly share a common characteristic: the people using the system understand why it exists and what it is designed to free them from. They are not being asked to use a tool for reasons that have been explained to their manager but not to them, and they can see specifically what the system does and what it means for how their work operates.

When adoption is poor, the cause is almost always one of three things: the system was introduced without the team understanding the problem it was solving, the transition period asked too much of people who were already at capacity, or the new workflow required more steps than the old one in the short term without enough visibility of the longer-term benefit. None of these are technology problems, and all of them are conditions that can be managed, which is what separates an implementation that changes something from one that produces a new line item in the software budget.

The measure of whether it worked

A good implementation has one test: did the commercial problem it was designed to solve get solved, and can that be demonstrated with a number.

Not whether the system is being used, not whether the team is satisfied with the tool, but whether the cost has come down, the capacity has been released, or the risk has been reduced by the amount that was agreed before the work began. Everything else is a proxy measure, and proxy measures are how businesses convince themselves that technology investments paid off without being able to demonstrate that they did.

The businesses we work with know what success looks like before the first line of code is written. That clarity is what makes the difference between an Intelligent Operating System that becomes a permanent part of how the business operates, and a technology project that was completed and then slowly forgotten after sharing the highlight in the board meeting.

Karst is an operations and technology firm that designs and implements Intelligent Operating Systems to deliver commercial outcomes. Built by operators. Powered by AI. Measured in outcomes.

The most painful processes are not always the most valuable to fix first, because they are the processes that generate the most complaints, consume the most management attention, and produce the most visible frustration, and while those qualities make them memorable they do not make them the best place to invest the first round of operational work. Starting with the loudest problem rather than the highest-value one is one of the more reliable ways to invest in operational improvement and produce a result that fails to change the commercial picture.

The sequence in which you build an Intelligent Operating System matters at least as much as the decision to build one at all, because getting the sequence right produces compounding returns while getting it wrong produces marginal improvements that disappear into the background noise of a business continuing to operate the way it always has.

The four questions that identify the highest-value starting points

There is a practical framework for identifying where operational investment will produce the most commercial return, and it comes from asking four questions about each candidate process.

The first is volume, because processes that run once a year, even painful ones, are not strong early candidates: the commercial case for systemising any process depends partly on how often it runs, given that a process running daily generates daily returns while one running quarterly generates quarterly returns, making volume the multiplier against which every other calculation operates.

The second is consequence of failure. Some processes carry significant downstream risk when they go wrong: a client onboarding sequence that fails damages a relationship before it has properly begun and risks early churn, a compliance process that fails creates regulatory exposure, and a reporting process that fails produces decisions made on incorrect information. High-consequence processes are prioritised not because they fail often but because when they do the cost is disproportionate to the time the process itself takes to run.

The third is senior capacity consumption. Processes that require experienced, well- paid people to complete are candidates for early investment regardless of whether those people raise them as a priority, because when a senior person's time is absorbed by a task that does not require their judgment the loss is not just the time the task takes but the thinking that did not happen while it was being completed. Asana's Anatomy of Work research found that knowledge workers spend 60% of their working time on tasks other than the skilled work they were hired to do, a pattern as pronounced for senior staff as for anyone else and considerably more expensive per hour.

The fourth is downstream dependency. Some processes feed other processes and getting them right makes everything that follows more reliable, while getting them wrong creates errors that propagate through the business: client intake data captured inconsistently produces inconsistent delivery records, inconsistent invoicing, and inconsistent reporting, meaning an investment in the front of that chain is an investment in the integrity of everything downstream, making the return larger than the process itself would suggest.

The output of this exercise is not a ranked list of processes to fix in sequence. It is a map of where the first components of a connected operating system should be built, and in what order, so that each one creates the conditions for the next. J.P. Morgan Asset Management's research found that while nearly 90% of companies have invested in AI technology, fewer than 40% report measurable gains, largely because most apply it to discrete tasks rather than redesigning how work gets done.

The four questions above are a guide to avoiding exactly that outcome: identifying not which tasks to automate but where to start building a system that redesigns how work flows through the business.

Why the obvious starting point is almost never the right one

The reason businesses default to their most painful process is understandable: it is visible, generates internal pressure, and fixing it produces goodwill that is immediately felt. There is a reasonable intuition that the amount of pain a process causes correlates with the value of removing it, but while the correlation exists it is weak, because pain is a function of visibility and frustration while value is a function of commercial impact, and they overlap without being the same thing.

The most painful processes in many businesses are painful because they involve people directly: manual handovers between teams, approval sequences that depend on one person's availability, briefing processes that rely on institutional memory rather than documented structure. These produce friction and generate complaints without necessarily being where operational investment produces the largest financial return.

The highest-value processes are frequently the unnoticed ones: the financial reconciliation running across the business every month that absorbs three days of the finance function's capacity, the compliance review that nobody considers glamorous but that carries meaningful regulatory consequence if it goes wrong, the client reporting produced manually for four years because that is how it has always been done. These processes do not generate urgent complaints but generate chronic cost, and chronic cost is harder to see precisely because it never reaches the level of urgency that demands attention.

The sequence that builds connected infrastructure

When the four questions above are applied honestly, a priority order tends to emerge that is both commercially defensible and operationally achievable, and more importantly, each layer makes the next one more valuable.

The right first investment is almost always a process with high volume, clear downstream dependencies, and meaningful senior capacity consumption. Client intake and onboarding sits here for most professional services businesses, producing returns immediately through faster engagement initiation, more consistent early client experience, and the release of senior time that was previously absorbed by a process that should never have required senior involvement. But its deeper value is what it enables: accurate client data flowing into every system downstream, which makes the second layer of investment significantly more effective.

The second wave addresses high-consequence, lower-frequency processes, and it benefits directly from the data integrity established in the first. Compliance and regulatory workflows, financial reconciliation and reporting, processes where the consequence of failure is significant even if the frequency is monthly or quarterly: all of these become faster and more reliable to build when the intake and onboarding infrastructure feeding them is already producing clean, consistent data. The investment here is about risk as much as return, building the layer that means a process failure cannot become a business crisis.

The third wave addresses the management capacity question, asking what recurring decision points currently require senior involvement that a well-designed system could resolve, and what is landing on leadership desks because there is no infrastructure below to catch it. This wave becomes tractable precisely because the first two have freed the management capacity to examine it honestly.

The combined effect on how the business operates, with consistent data flowing through reliable processes and senior attention available for work that actually requires it, is what distinguishes a connected operating system from a collection of point solutions.

What this looks like in practice

The research confirms what this sequence is designed to prevent. J.P. Morgan Asset Management found that fewer than 40% of companies investing in AI technology report measurable gains, because most apply it to discrete tasks rather than redesigning how work gets done. The three waves above are not three discrete improvements. They are three layers of a system, where the integrity of each layer depends on the one below it.

Businesses that address operational problems reactively, fixing whatever is loudest at any given moment, tend to produce exactly the outcome the research describes: capable systems that underdeliver commercially because they sit inside an operating model that was never redesigned around them.

The question is not which process to fix but which sequence of fixes builds something that changes what the business can do.

Karst is an operations and technology firm that designs and implements Intelligent Operating Systems to deliver commercial outcomes. Built by operators. Powered by AI. Measured in outcomes.

The cost accumulates in the hours a senior person spends producing a report that could have been generated automatically, in the lag between a client signing and the delivery team being briefed because the handover depends on a process that lives in someone's head, and in the compliance task that takes three days each quarter because the underlying data sits in four different places and someone has to reconcile it by hand. These costs are real, recurring, and structurally embedded, but they do not have a line item.

The gap between the cost that appears on a P&L and the cost the business is actually bearing is, in most cases, significant, and understanding where it comes from and what it is worth closing is where the most valuable operational work begins.

The three forms of operational cost most businesses are not measuring

Operational inefficiency shows up in three distinct ways, and businesses tend to be aware of one while overlooking the other two.

The first is direct time cost. A task that requires a human being to complete it takes time, and time has a salary attached. When that task is manual and performed by someone whose rate reflects their experience and judgment, the cost of continuing to do it that way compounds across every repetition, because a compliance review that takes a senior analyst two days each month costs not just two days of salary but two days during which they are not doing the higher-value work that justified hiring them. Most businesses calculate the cost of the task in isolation and rarely the cost of the opportunity it displaces.

The second is error and rework cost. Manual processes introduce variability, and variability produces errors: an onboarding sequence that depends on staff remembering which documents to request will occasionally result in incomplete files, and a reporting process that involves copying data between systems will produce discrepancies. Each error costs time to identify and correct, and some carry consequences beyond time in delayed settlements, failed audits, and client dissatisfaction. The cost is real yet largely invisible because businesses rarely track it systematically, presenting instead as the general sense that operations are slower and more difficult than they should be.

The third is the capacity cost of operational drag, which shows up not as an expense but as slower growth, longer decision cycles, and the persistent sense that the leadership team is running at full capacity without getting ahead. Every process that consumes management attention is a process that prevents that attention from going somewhere more valuable, and when the leadership team is spending meaningful time each week resolving operational problems or being pulled into work that falls to them because there is no infrastructure below to catch it, the business is paying for that drag with its strategic capacity.

What the research shows

Asana's Anatomy of Work Index, drawn from a survey of more than 10,000 knowledge workers globally, found that 60% of working time is spent on what they describe as work about work: chasing status updates, attending unnecessary meetings, switching between tools, and completing duplicative tasks, accumulating 209 hours per year per person on duplicative work alone.

The Thomson Reuters Institute has documented the same pattern in professional services environments, where time spent on administrative and process work consistently displaces the advisory and analytical capacity that clients are paying for.

These are not edge cases but the default state of most businesses that have grown without systematically examining where senior time is actually going, and the cost is chronic in a way that makes it harder to see precisely because it never reaches the level of urgency that demands immediate action.

Where the highest-value costs tend to concentrate

Across the businesses we work with, operational cost tends to concentrate in a small number of areas.

Client intake and onboarding is consistently one of the highest. The transition from signed agreement to active engagement is, in many businesses, manual, inconsistent, and dependent on institutional knowledge about how things are supposed to work, meaning that when onboarding is slow or incomplete it damages the client relationship before it has properly begun and consumes staff time that should be going into delivery.

Reporting and financial reconciliation is another. The monthly or quarterly effort to compile operational and financial data into a format that leadership can act on is, in most businesses, significantly more labour-intensive than it needs to be: the data exists, but the effort of assembling it into something readable is what takes the time, and that effort rarely produces insight that a well-structured system could not produce automatically.

Compliance and document management sits alongside these. In regulated industries in particular, the volume of documentation that must be collected, verified, stored, and retrieved is substantial, and when that process is manual it is both expensive and fragile, with consequences when it fails that extend well beyond the time required to correct it.

The conversation that changes the picture

When we sit down with a new client, the first question we ask is not which tools they are using or what their technology budget looks like. We ask: what is this actually costing you.

The question is concrete, covering specific processes, specific people, specific amounts of time, and specific consequences when things go wrong. The answer almost always reveals a gap between what the business believes its operations cost and what they cost when the full picture is assembled honestly, and that gap is where the work begins, where quantifying the cost is the first step toward closing it and closing it is what makes the investment in a well-designed Intelligent Operating System a financial decision with a clear return rather than a technology decision with a speculative one.

Every month of continued operational drag represents a real and calculable expense compounding forward against the cost of fixing it once, which means the cost of doing nothing is not zero: it is a number worth knowing.

If you have tried to quantify the operational cost in your business, what surprised you most about where the time was actually going?

Karst is an operations and technology firm that designs and implements Intelligent Operating Systems to deliver commercial outcomes. Built by operators. Powered by AI. Measured in outcomes.

The businesses most affected by key-person dependency rarely experience it as a crisis. They experience it as friction: decisions that slow down because one person is at capacity, opportunities that stall because one relationship is unavailable, growth that requires scaling something the business has never been able to scale. The key person keeps showing up, so the cost of depending on them never quite becomes urgent enough to address. Meanwhile, the ceiling on what the business can do is set not by its market, its capital, or its strategy, but by the bandwidth of one or two individuals who have become structurally irreplaceable.

That ceiling is expensive. It limits how fast the business can take on new work, how confidently it can commit to growth targets, and how attractive it looks to anyone evaluating it from the outside. Key-person dependency is one of the most widely acknowledged risks in business and one of the least addressed, because the business keeps functioning well enough to defer the conversation. The cost of deferring it, however, compounds quietly in the background every year that nothing structural changes.

Mercer and Marsh's 2024 People Risk Report, which surveyed over a thousand risk and HR professionals, found that 34 percent of organisations cited dependency on key individuals and inadequate succession planning as a primary concern. The concern is real and widely felt. What has changed is that it is now structurally solvable in a way it has not been before.

The cost the P&L never shows

The visible cost of key-person dependency is the disruption that follows when the key person is unavailable. The invisible cost is larger and more persistent: the ceiling it places on the business every day that person is present.

When decisions route through one individual, the pace of the business is governed by their capacity. When client relationships are held by one person, the business's ability to win new work, expand existing accounts, or survive that person's departure is constrained by a single point of failure. When operational knowledge lives in one person's head, the business cannot be documented, delegated, or scaled without them.

Research published in MIT Sloan Management Review in December 2024 found that organisational bottlenecks are routinely mismanaged because businesses address them piecemeal rather than systemically, fixing the most visible constraint without addressing the underlying resource and workflow architecture that generates bottlenecks in the first place. The result is that relieving pressure in one area simply shifts it elsewhere. The business feels like it is making progress. The structural dependency remains.

That dependency also has a market price. In formal business valuations, key person discounts of between 5 and 25 percent are routinely applied to the assessed value of a business where performance, relationships, or decision-making are concentrated in one individual, with higher discounts applied where the dependency is more acute or succession planning is absent. The reduction is not a reflection of poor performance. It is an acquirer or investor pricing the risk that the performance does not continue without that person present.

Where key-person dependency actually lives

The risk concentrates in different places depending on the nature and scale of the business. Recognising it requires looking beyond the obvious.

In a founder-led professional services firm with fifty people, the dependency is almost always relational. The founder holds the senior client relationships personally. New business flows through their network. Account growth depends on their involvement in the conversation. The firm can deliver excellent work, but its commercial ceiling is the founder's personal bandwidth.

In a mid-market manufacturing business with three hundred staff, the dependency tends to be operational. One production manager holds the institutional knowledge of the plant: the supplier relationships that took years to build, the workarounds that keep ageing equipment running, the scheduling logic that lives in a spreadsheet and in their head simultaneously. The business runs because they show up, but the risk profile of that business changes materially when they do not.

In a high-growth technology business, it is frequently architectural. One engineer understands the codebase in its entirety. Others work within it, but decisions about how it scales, where it is fragile, and what it can support depend on one person's comprehension of the whole. That dependency does not appear on any risk register until something breaks at the worst possible moment.

The form the dependency takes matters less than the pattern it creates: a business whose capacity, quality, or commercial performance is bounded by the continued presence of one individual in a role that was never designed to be held by systems.

The view from inside

What rarely surfaces in conversations about key-person dependency is that the key person themselves is often the most acutely aware of the problem.

Senior leaders and founders who have become operational bottlenecks did not set out to be. They became indispensable because the business needed them to be, and because nothing was ever built to reduce that need. Many carry a version of the same frustration: they can see the decisions that should not require them, the approvals that should flow without their sign-off, the work that should move faster than it does. They want to step back but the structure does not allow it.

The ask that comes up repeatedly in conversations with C-suite leaders is more immediate than succession planning. It is a system that lets the business move faster without them being in every critical path: something that holds the logic of how decisions get made, routes the right information to the right people, and handles the operational cadence that currently requires their personal involvement. The goal is not to remove them from the business. It is to free them for the work that actually requires their judgment: the strategic decisions, the relationships, the problems that genuinely need the most experienced person in the room.

Succession planning matters, and most serious leadership teams have some version of it. Finding a successor with the same capabilities, relationships, and institutional knowledge as the person they are replacing is genuinely difficult. The businesses that reduce that difficulty most effectively are the ones that have already systematised what the key person holds, so the incoming person steps into a role supported by infrastructure rather than one held together by the individual judgment of whoever occupied it last.

What changes when the infrastructure changes

An Intelligent Operating System addresses key-person dependency at the level where it actually lives: in the processes, decisions, and knowledge flows that currently require a specific person to function.

For the founder-led professional services firm, this means the client onboarding, account management cadence, and reporting processes that flow through the founder are systematised so that senior practitioners can manage them independently. For the manufacturing business, the scheduling logic, supplier communication, and production reporting the operations manager holds personally are embedded in a system that surfaces the same decisions and flags the same exceptions, regardless of who is looking at it.

In each case, the key person's judgment, relationships, and expertise remain valuable. The operational bottleneck they have become does not. A business that has systematised what its key people hold can scale what they contribute, deploy their judgment where it adds the most value, and present to any acquirer, investor, or partner as a business whose performance does not depend on the continued presence of specific individuals.

A business that has systematised what its key people hold does not just reduce its risk profile. It changes what growth means. Every new hire, every new client, every new market entered compounds through infrastructure that was built to carry the weight rather than inherited from necessity. For businesses that have been growing around their key people rather than through a system built to support that growth, the shift is less about risk reduction than about what becomes possible once the constraint is removed. 

Karst is an operations and technology firm that designs and implements Intelligent Operating Systems to deliver commercial outcomes. Built by operators. Powered by AI. Measured in outcomes.

A professional services firm that has grown from a boutique practice to a national operation still runs conflict checks through a shared inbox, because nobody ever had the time to replace a process that technically worked. A financial services group with hundreds of staff still produces weekly performance reports by having analysts manually extract figures from four separate systems, because the systems were acquired at different times and were never integrated. A logistics operator managing thousands of movements a month still relies on a central operations manager to manually assign jobs each morning, a process designed when the fleet was a fraction of its current size, and one that has become the single most expensive bottleneck in the business.

In each case, the process was functional once. It is now the constraint on everything else: on how fast the business can take on new work, on how much time senior people spend on administration instead of judgment, on how reliably the business delivers when it is under pressure.

The pattern holds across industries and at every scale. Processes designed for a business at one stage of growth are rarely revisited as the business evolves: they accumulate workarounds, get staffed around, and the original logic that made them sensible becomes the ceiling that limits performance.

Why it persists

Operational infrastructure does not get fixed for three reasons that reinforce each other.

The first is that the symptoms get misread. When a growing business experiences slow delivery, margin erosion, or decisions that always funnel through the same two people, the diagnosis is usually a talent problem or a management problem, leading to a better operations manager, a strategy day, clearer KPIs. The underlying process debt, the accumulation of workflows that were never updated, stays invisible because it is not where attention goes.

The second is that every fix gets treated as a tool decision. The business buys a project management platform, a CRM, an AI writing assistant, a new finance system, and the tools arrive while the processes they were supposed to replace continue running, sometimes in parallel with the new technology. Docusign's 2024 Digital Maturity Report found that 54% of staff at organisations that had increased technology investment were not using those tools to drive efficiency. The tools are not the problem; the operating model they are being asked to sit inside is.

The third is urgency. Operational reform requires time that never exists: time to map workflows end to end, find the points of genuine friction, design replacements that do not break what is already working, and implement them without stopping the business while it runs. There is always a client, a quarter, a hire, a deadline that comes first. Infrastructure investment is permanently deferred to the moment things slow down. Things do not slow down.

The ambition is not in question. The Thomson Reuters Institute's 2025 C-Suite Survey, which canvassed 200 senior executives across eight countries, found that digital transformation ranked as a high priority for 82% of respondents, with improving operational efficiency close behind at 64%. What that investment rarely includes is the work that determines whether transformation actually lands: identifying the processes creating the most friction, building systems designed to replace them, and sequencing implementation so the business realises value from day one rather than enduring months of disruption on the promise of future return.

What actually changes

The distinction that matters is between adding tools and implementing an Intelligent Operating System.

Adding tools assumes the existing operating model is sound and layers capability on top of it. An Intelligent Operating System starts with the specific processes creating the most friction, cost, or constraint, and replaces them with systems built to handle that work. The operating model stays intact. The manual infrastructure inside it does not.

In a professional services firm, this means the coordination work around scoping, tracking, and delivering client engagements is handled by the system, so that work requiring senior judgment reaches the people who can apply it, and everything else runs without intervention. In a financial services business, it means performance information is accurate, consolidated, and available in real time, rather than assembled by hand from four systems every Friday. In a large logistics operation, it means job assignment, client communication, exception handling, and reporting are handled by systems designed for the purpose, freeing senior coordinators for the decisions that actually require human judgment.

The outcome in each case is not faster execution of the same process. It is the same operating model, running without the drag that was costing it margin, capacity, and the time of its most expensive people.

The compounding argument

The case for acting is not symmetrical. Inaction does not hold a business in place; it moves it backwards, because the cost of process debt grows with every person hired into it.

A business that scales headcount without addressing its process infrastructure does not dilute the friction across more people, it multiplies it. The approval bottleneck that slowed ten people slows thirty. The manual reporting process that took one analyst three hours now takes two analysts five, with discrepancies between them that require a third to resolve. The client onboarding that depended on one person's institutional knowledge now depends on several people's competing versions of it. Each new hire either absorbs the dysfunction, making it a permanent cost, or works around it, making it a permanent source of inconsistency. 

The value of change is well documented. Asana's Anatomy of Work Index, which surveys nearly 10,000 workers across major economies, found that improved processes could recover 4.9 hours per week per person, more than six full working weeks across a year. At scale, that is not a productivity improvement. It is a structural change in what an organisation can do with the people it already has.

The businesses that move first do not simply recover margin, though they do that. They build something that compounds: each new module removes friction, each reduction in friction frees capacity, and that capacity either flows to higher-value work or reduces the cost of doing the same work at greater scale. Most leadership teams already know where the constraints are. The cost of knowing and not acting is the one that accumulates fastest.

Karst is an operations and technology firm that designs and implements Intelligent Operating Systems to deliver commercial outcomes. Built by operators. Powered by AI. Measured in outcomes.