The question most businesses ask when they decide to address their operations is a reasonable one: where do we begin. The answer that seems obvious, start with whatever is most painful, is also in most cases wrong.
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.
