Three People Followed My Dad Around for Two Months. All Three Quit.
My dad is retiring this year after forty years as an engineer!
For the last stretch of his career, his employer has been trying to solve a problem they saw coming. They assigned people to shadow him, follow him around, watch what he does, & write it down.
Each one watched for about two months, and one after another, each one transferred to a different position.
I laughed when he told me, because capturing what those three people failed to capture is what I do for a living. I have watched this same thing play out for client after client, and the ending is always more expensive than anyone budgeted for.
Nobody at that company is asleep at the wheel. They knew forty years of knowledge was walking out to door any day and they spent real money on the problem. It would seem they just spent it on the wrong thing, and now they will likely absorb his work into the other engineers on the team and hope for the best.
Those engineers are good at their jobs, but none of them have forty years of context on why the process works the way it does.
What actually leaves when someone leaves (Studies/Metrics)
Panopto and YouGov surveyed 1,001 US employees and found that 42% of institutional knowledge is unique to the individual who holds it. Not written down or shared and not recoverable from a file share.
When that person walks out, their colleagues cannot perform 42% of their job, and the replacement has to rebuild it from scratch. The same study found new hires spend roughly 28 unproductive hours per month for six and a half months getting up to speed, and that a 1,000-person organization loses about $2.7 million a year to knowledge that never got shared.
Now consider what happens when the person leaving has four decades of it.
Leadership knows. That is the strange part.
APQC surveyed 1,000 organizations in 2025. Among senior leaders, 84% said they were concerned about knowledge loss from retiring employees.
8% of organizations consistently capture that knowledge before those employees leave.
In the manufacturing cut of the same research, 85% of C-suite leaders called knowledge loss at least a moderate concern. 41% of their organizations rarely or never attempt to capture it. Those same organizations expect an average of 51% of their workforce to retire or leave within five years.
So this is not a blind spot. Executives see it clearly and describe it accurately. What is missing is an owner. No job title contains “make sure Guy’s forty years gets written down,” so it stays on the someday list until Guy gives notice, and then it becomes a two-week scramble that produces a folder nobody opens again.
The bill arrives later, on a different line item
The reason this stays unfunded is that the cost never shows up as knowledge loss. It shows up as something else.
A contract slips six months because the transition took longer than the one-month ramp somebody sketched on a slide. A product ships with a defect that the retired engineer would have caught by looking at it. A line goes down and takes four hours to diagnose instead of forty minutes, because the person who knew that machine’s specific personality retired in March.
Siemens put a number on that last category. In its True Cost of Downtime 2024 report, unplanned downtime cost the world’s 500 largest manufacturers roughly $1.4 trillion a year, about 11% of their combined revenue, up from 8% five years earlier. In automotive, an idle production line runs up to $2.3 million per hour.
Not all of that traces to knowledge loss, but there is a strong argument that a meaningful share of it does. you can expect that none of it gets coded that way in the general ledger.
“We will just backfill the role”
This is the assumption that does the most damage, and it is worth pressure-testing with actual research rather than optimism.
Matthew Bidwell’s work in Administrative Science Quarterly found that external hires are paid 18 to 20% more than internal promotions for comparable roles, even after controlling for their higher experience and education. They also take roughly two years to match the performance of someone promoted from within.
Two years… Not the one month of ramp-up in the transition plan.
And that assumes you backfill at all. In plenty of cases the role quietly disappears and the work gets distributed across a team that absorbs it on top of everything else. That team will get most of it right. The remaining percentage is where your delays and defects come from.
What it costs to fix versus what it costs to find out
Documenting a critical role properly is not cheap. For a complex technical position with decades of accumulated judgment behind it, real capture work runs into six figures and takes months, because it requires watching the work happen, not interviewing someone about it.
That is the part the three replacements got wrong. They were asked to observe and take notes. No body told them what to look for or how they’d know when they were done. They were smart people handed an impossible assignment, so they did what smart people do and found a better assignment.
Structured capture produces something different: procedures somebody can actually follow, the decision rules behind them, the failure modes, and the reasoning that explains why the process exists in its current form. It survives the person retiring and it survives the next person too.
Run your own numbers
Rather than tell you what your return would be, here is the math. Use your figures, not mine.
Step 1. What is exposed?
Exposure = (Revenue or contract value that depends on this role)
x (Probability of disruption during transition)If a $10 million contract depends on one person’s expertise and you estimate a 30% chance of a meaningful delay during handoff, your exposure is $3 million before anyone argues about assumptions.
Step 2. What does the delay actually cost?
Delay cost = (Days of delay) x (Daily cost of delay)Daily cost of delay includes idle labor, penalty clauses, deferred revenue recognition, and the cost of pulling other engineers off their own work to firefight. Most organizations underestimate the last one because it never gets invoiced.
Step 3. What does the slow ramp cost?
Ramp cost = (Months to full productivity)
x (Monthly fully-loaded cost of the role)
x (Percentage of productivity not yet reached)A $180,000 role at 50% productivity for twelve months is $90,000 in output you paid for and did not receive. Apply that to every person absorbing part of the departing role, not only to a single replacement.
Step 4. Compare.
ROI = ((Avoided cost - Investment) / Investment) x 100Where avoided cost is the portion of steps 1 through 3 you can reasonably prevent. Be conservative. Claim half. The math usually still works, and a defensible number you can bring to your CFO beats an impressive one you cannot.
If your answer says the investment does not make sense, that is genuinely useful information and you should not spend the money. In most conversations we have, the number lands somewhere the client did not expect, because they had never added up steps 2 and 3 in the same place.
Why so few organizations can prove any of this
ATD and the ROI Institute studied what executives actually receive from their learning and development functions. 94% of organizations track inputs. 4% measure ROI and only 8% measure impact. Meanwhile 96% of business stakeholders believe impact should be measured, and 74% expect an ROI analysis.
That gap explains a lot about why knowledge capture never gets funded. The people asking for the budget cannot prove the return, because nobody built the measurement infrastructure alongside the work.
We build both. The documentation and the baseline that proves what it changed.
Where this ends
My dad will retire. His employer will get through it, because organizations survive things like this. They will absorb some avoidable cost, blame it on something else, and eventually stop noticing what they used to know.
Somebody there could have made a different call two years ago for a fraction of what the next two years will cost them.
If you have a person on your team whose departure would genuinely hurt, run the four steps above this week. Do it on paper, alone, before you talk to anyone about budget. Then decide.
If the number surprises you, we should talk.
Instructive Edge documents the processes and expertise your organization cannot afford to lose. Contact us and we will start with a conversation about which roles carry the most exposure.
Sources: APQC and eGain, “The Great Retirement: Knowledge Loss, AI and the Workforce Shift,” 2025 (n=1,000 organizations). Panopto and YouGov, “Workplace Knowledge and Productivity Report” (n=1,001 US employees). Siemens, “The True Cost of Downtime 2024.” Matthew Bidwell, Administrative Science Quarterly. ATD and the ROI Institute, “Measuring for Success: What CEOs Really Think About Learning Investments.”
