When Systems Don't Talk, Your People Pay the Price
The hidden cost of manual work and why the most expensive part isn't on the balance sheet.
I've watched this scene play out over and over again in my last organization.
Let's take onboarding for example. A new hire joins the team. All their personal and professional details are manually entered into the recruitment system. Then a capable, experienced employee enters that same information into the onboarding system. Then another employee enters the same data again into the workforce management system. And finally, into the payroll and benefits systems.
The same data. Four systems. Often four separate moments of typing, by the same person or different ones. Each platform was bought at some point to solve a real problem, and each one does. But none of them were designed to talk to the others so a person, usually one of the strongest people on the team, becomes the bridge between them.
This is hours of their week. Sometimes hours of their day. And nobody, including them, quite registers how much of their working life this is consuming.
When I'm brought in to look at why an organization isn't performing the way leadership thinks it should, this is almost always part of the answer. Not the part anyone names. But underneath the symptoms; missed deadlines, declining morale, "we just need more headcount," the team that used to be your strongest now somehow underperforming, there's usually a quiet, structural fact: the systems that run the business don't talk to each other.
And the people are being asked to do the talking instead.
The cost is bigger than anyone admits
The numbers on this are striking once you actually look at them.
Gartner's research puts the average annual cost of poor data quality at roughly $12.9 million per organization. A widely cited estimate from MIT Sloan Management Review and Cork University Business School concludes that companies lose somewhere between 15 and 25 percent of revenue annually because of bad data, much of it introduced when humans move information between systems by hand. A 2024 survey from Parseur calculated that manual data entry costs U.S. companies an average of $28,500 per employee, per year.
McKinsey's analysis suggests that employees across many roles spend close to 45 percent of their working time on tasks that could be automated. A separate Zapier survey of office workers found that 76 percent spend up to three hours every single day on data entry alone.
You can quibble with any single figure, but the direction is unambiguous: the work of moving information from one system to another, by hand, is one of the largest hidden expenses on a modern P&L. And it's almost never broken out as a line item which is exactly why it persists.
Where the breakage actually happens
The errors compound. There's a well-known principle in data quality work, sometimes called the 1-10-100 rule, that captures the dynamic well: catching a data error at the point of entry costs roughly 1x to fix. Catching it after it's traveled through your systems costs around 10x. Catching it after a decision has already been made on it costs 100x.
I see this constantly. A clinician, or a finance analyst, or an operations manager (the role doesn't really matter) types a number wrong because they're tired, or rushed, or doing the same task for the eighth time that morning. The number lands in the wrong system. By the time anyone notices, three downstream reports have been built on top of it. A leadership decision has been made. Maybe a customer has been billed incorrectly. Maybe a patient's record reflects something that isn't accurate.
Then the team spends the next two weeks unwinding it. A separate project, with its own cost, that nobody wanted.
The part most leadership teams underestimate is how often this happens quietly. Most data errors don't surface as dramatic incidents. They surface as a slow drift between what your dashboards say and what's actually true on the ground. By the time the gap is wide enough to notice, no one can reconstruct exactly when or where it started.
The part nobody puts on the slide deck
Here's what I find most striking after twenty-five years of doing this work: the operational cost of disconnected systems is almost always less destructive than the human cost.
Recent research bears this out. A 2025 study found that 85 percent of employees identified repetitive tasks as a leading cause of burnout. The Parseur survey reported that 56 percent of employees experience burnout specifically from repetitive data tasks. The Zapier study found that 38 percent of U.S. office workers cite burnout, not workload, not management, not compensation, as the single biggest barrier to their productivity.
What this looks like inside a real organization is this: your most capable people, the ones you hired because they could think, are spending their best hours doing work that doesn't require them. The work that does require them, the strategic thinking, the judgment calls, the actual mission of the organization, gets crammed into whatever's left over. Which usually isn't much.
Then one of two things happens.
They quit. Often quietly, often citing reasons that don't quite explain the depth of their disengagement. They go somewhere else, where leadership has thought more carefully about how the work is structured.
Or they stay, and slowly stop bringing their best to the work, because there's no room for it anyway. They check in. They check out. The mission of the organization, which is the reason any of you are doing this, recedes into the background.
Either outcome is expensive. Both are usually invisible until someone tries to fix the symptom of "we have a retention problem," "morale is low", without addressing the structural cause underneath.
This is a leadership problem, not an IT problem
The instinct, when this finally surfaces at the leadership level, is to treat it as a technology problem. Buy a new platform. Hire an integrations vendor. Stand up an automation project. Sometimes that's part of the answer. But in my experience, it's almost never the whole answer, and it's frequently the wrong place to start.
Disconnected systems are a symptom of how the work was designed or, more often, how it accumulated over years of well-intentioned point solutions without anyone holding the whole picture. Software A was bought to solve a problem in 2018. Software B was bought to solve a different problem in 2021. Nobody was responsible for the seam between them, so a person became the seam. Then another person. Then a whole team whose actual job description has very little to do with what they spend their days on.
You can't solve that with software alone. You solve it by stepping back, mapping how the work actually moves, not how the org chart says it moves, finding the manual seams, and making clear, deliberate decisions about which ones get automated, which ones get redesigned out of existence, and which ones genuinely need a human in the middle.
That's a leadership exercise, not an IT exercise. And it requires someone willing to look at the whole picture and name what's actually underneath.
What it looks like when it's working
In organizations that get this right, the most senior people, the ones whose judgment you're paying for, spend their time on judgment calls. Systems hand information back and forth without asking a person to act as a courier between them. Errors get caught at the point of entry, not three reports later. The team has the bandwidth to pursue the actual mission of the organization, which is presumably why everyone signed up in the first place.
This isn't a perfectly automated, no-humans-in-the-loop fantasy. It's just a workplace where the work has been thought about its become where someone has owned the question of how the parts connect, and where the connections make sense.
If this sounds familiar
If you're reading this and recognizing your organization, you're not alone, and the pattern is more fixable than it usually feels from the inside. The first step is almost never buying something. It's understanding the actual shape of the work, including the parts no one talks about because they've become normal.
That's the work I do at Keller Coast Consulting. If you'd like a conversation about what's getting in the way at your organization, you can grab time directly on my calendar.
Sources
Gartner research on the average annual cost of poor data quality (~$12.9M per organization).
MIT Sloan Management Review and Cork University Business School joint research on revenue loss from bad data (15–25% of revenue).
IBM Institute for Business Value, 2025 report on data quality and AI readiness — finding that 43% of chief operations officers name data quality as their most significant data priority.
McKinsey analysis on the percentage of working time spent on automatable tasks (~45%).
Parseur, 2024 Manual Data Entry Survey — $28,500 per-employee annual cost; 56% of employees report burnout from repetitive data tasks.
Zapier, State of Workplace Productivity Report — 76% of office workers spend up to three hours per day on data entry; 38% identify burnout as their top productivity barrier.
Moody at Work, 2025 Workplace Burnout Study — 85% of employees identified repetitive tasks as a leading cause of burnout.