The Continuity Audit: A Facility Manager's Framework for Measuring and Closing Located-Knowledge Risk Before Retirements, Contract Changes and Reorganizations

Every facility team runs on a category of knowledge that has no good name in most organizations, though everyone recognizes it the moment it is described: which breaker panel feeds the east wing and which one merely looks like it does; the valve that isolates the third-floor riser and the trick to closing it without hammering the line; why the air handler on Building 4’s roof was piped backwards in 1998 and what happens if anyone “corrects” it. Levaru’s earlier research called this located knowledge — real, operational, tied to specific places, assets and relationships, and almost never written down, because prose is a poor medium for it and because, for as long as its holder is one radio call away, it never needs to be. It is not the same as skill: a licensed electrician carries their trade from building to building, but thirty years of one building’s decisions, exceptions and workarounds cannot be carried in. And most of it is not ineffable — someone standing in the room can point and say “that one, not the one with the newer label.” It goes unrecorded, Hanov argues, not because it cannot be captured but because no one has ever been assigned to capture it.

The loss arrives through three foreseeable doors. A retirement or resignation removes a holder outright. A contract change — an outsourcing transition, an insourcing decision, a re-tender — can remove a dozen holders on a single changeover date. A reorganization removes holders quietly: still employed, now two buildings away, the fifteen-minute answer decaying into a favor, then a bottleneck, then silence. All three come with dates known well in advance, yet the standard responses — the exit interview, the farewell-quarter transition binder — consistently arrive too late and capture too little, because they are improvised at the moment of departure. The Continuity Audit proposes the practice that should exist instead: a governed, repeatable assessment that inventories the located knowledge an operation depends on, scores each item for operational criticality and documentation coverage, and converts the result into an owned, prioritized closure plan — re-run on a defined cadence and re-triggered by the events that put knowledge at risk. It sits deliberately between the exit interview (cheap, familiar, almost always too late) and the enterprise knowledge-management program (systematic but heavy, designed for office work, and prone to stalling at exactly the plant-room layer that matters most) — and the paper frames it not as a human-resources nicety but as an operational control, as fundamental as the preventive-maintenance schedule and considerably cheaper than the failures it prevents.

Why now

Hanov frames the moment with three converging pressures, each supported by published figures the paper cites rather than generates. The first is workforce exit at scale: a study cited in IFMA’s Facility Management Journal found 31 percent of facility management employees older than 55 and eligible to retire between 2022 and 2037, and at the University of Tennessee, Knoxville, a facilities organization of roughly 600 staff maintaining more than 250 buildings recorded 97 retirements between 2020 and 2025 — one departing veteran roughly every three weeks, sustained for half a decade, in a single portfolio. The second is the measured cost of knowledge that lives in people: the Panopto/YouGov Workplace Knowledge and Productivity Report, surveying more than 1,000 United States workers, estimated that 42 percent of institutional knowledge is unique to the individual who holds it and put the friction cost at 5.3 hours per employee per week spent waiting for or recreating information — consistent with McKinsey’s estimate that interaction workers spend roughly 19 percent of the workweek searching for information and a 2023 Quickbase survey in which 53 percent of respondents chased information more than 10 hours a week. The third is that facility work increasingly changes hands by contract: Grand View Research valued the global facility-management services market at approximately US$1.75 trillion in 2024, with outsourced delivery accounting for 61.5 percent of revenue — meaning that in a majority of the market, the people holding a building’s located knowledge work for a service provider, and service contracts end. Deferred-maintenance pressure sharpens the stakes (Gordian puts deferred capital renewal on North American campuses at US$156 per gross square foot). Characteristically, the paper adds that no team need adopt these numbers as its own — its measurement chapter insists on the team’s own baseline.

An audit, not a survey

Five properties distinguish the audit from the skills matrices, brain-dump sessions and staff surveys that have gestured at this problem before. It is a scored inventory, not a list: every item receives two independent judgments — operational criticality and documentation coverage — because a list flatters everyone and prioritizes nothing, and the isolation sequence for the data-center cooling loop and the location of the holiday decorations are both “things Maria knows.” It has named ownership and sign-off, for the audit itself and for every closure item — an inspection program without an owner is a clipboard in a drawer. Coverage is mapped against systems of record: not just whether documentation exists but where a stranger would find it, which exposes the difference between documented and findable and yields the audit’s quietly valuable by-product, a gap map showing which systems the team’s knowledge actually lives in and which are ceremonial. It runs on a defined cadence with event triggers — annual refresh for most portfolios, re-run in targeted form on a resignation notice, a contract entering its final year, a reorganization entering planning. And it operates under governed handling of people-adjacent data, with rules written before the first interview. The definition is completed by exclusions stated plainly: the audit is not a performance review (used as one, it will never again receive an honest answer), not an exit interview, not a platform purchase, not a succession plan or retention strategy, and not a promise that nothing will be lost — judgment does not transfer by documentation; the audit’s defensible claim is the locatable layer.

The four instruments

The Handoff Test, promoted from a heuristic in the series’ first paper to the audit’s headline measure: if the most experienced person in a role were unreachable for 90 days — no calls, no texts — what percentage of the located knowledge the role depends on could a competent stranger recover from the organization’s systems alone? As an instrument it is computed as the coverage-weighted share of the role’s inventory items recoverable from named systems, optionally weighted again by criticality; it can and occasionally should be run live — give a competent outsider a routine task in an unfamiliar building, forbid them from asking people, and watch where the systems fail them. Thirty minutes of observed struggle, the paper notes, is worth a hundred self-assessments.

The Located-Knowledge Inventory is the evidentiary base, designed against the blank page — located knowledge is invisible to its holder the way water is invisible to fish, so elicitation works from prompts: the walk-through interview (sixty to ninety minutes, on foot, through the interviewee’s actual spaces, because located knowledge is spatially indexed in memory); work-order mining (orders reassigned repeatedly until they found the one person who could do them, recurring faults whose fix is never written the same way twice); a two-week tally of the questions reaching the most-consulted staff by radio, phone and doorway; and the incident record, read for what the response depended on knowing. Items enter in a deliberately small schema — a portfolio-scale inventory of a few hundred items fits in a spreadsheet, and should.

The Criticality–Coverage Matrix converts the inventory into decisions using deliberately coarse high/low judgments on both axes. Each quadrant carries a standing instruction: the red quadrant (high criticality, low coverage) gets first claim on closure effort, with named owners and dates; well-covered critical items get verification, not congratulation — documentation believed current has a way of describing the building two renovations ago; low-criticality gaps get accepted and recorded, the written acceptance itself a continuity artifact; the rest is simply maintained.

The closure-mechanisms menu is deliberately plural and vendor-neutral: annotated 360-degree capture (the natural closure for spatial items), short runbooks, narrated walkdown videos, structured shadowing, CMMS/IWMS record enrichment, who-to-call relationship maps, and vendor documentation clauses. Two principles govern the matching: the record must land in a system that survives its author — a beautifully annotated capture in a personal drive fails the Handoff Test as completely as no capture at all — and prefer the mechanism the holder will actually complete: a veteran who will never write a procedure will happily narrate one on camera while pointing at the equipment.

Where it pays back first

Retirements. The canonical case, and a familiar genre: the date known a year out, “knowledge transfer sessions” remembered six weeks from the end, a binder organized by whatever occurred to people in the room, and the first winter’s phone call to a person now fishing. With an operating audit, a retirement declaration does not begin a discovery exercise; it filters one — the departing person’s red-quadrant items become the transfer plan, already scored, with the remaining months allocated to what matters most rather than what is easiest to remember. The same machinery serves the arrival that follows: a new hire inherits a syllabus — the role’s items, where each is documented, the accepted gaps, and the veteran time best spent on judgment rather than addresses. The Panopto study’s finding that new hires receive roughly 2.5 months of formal training yet take up to six months to reach full productivity suggests how much of a first year is spent rediscovering exactly the layer the audit exists to have captured.

Contract changes. The industrial-scale case: a changeover replaces a shift’s worth of holders on a single date, with incentives misaligned at the worst moment — the outgoing provider demobilizing with no commercial reason to fund careful transfer, the incoming provider not yet knowing what to ask, the owner discovering the gap in the first winter. The audit gives the owner three levers matching the contract’s phases. At procurement, the inventory becomes contract language: not “provider shall cooperate in orderly handover” but a schedule of named knowledge items whose documented, verified closure is a condition of final payment — extending the structured-deliverable logic the construction-handover world encoded in COBie from commissioning data to operating knowledge. During the contract, cadence keeps the coverage map current — and knowledge documented in the owner’s systems throughout the term is knowledge the owner actually owns, whatever happens commercially. At transition, the inventory is the checklist, and the live Handoff Test becomes an acceptance procedure for the handover.

Reorganizations and unplanned absence. The quiet pillar: no one leaves the payroll when a portfolio is consolidated or a hands-on supervisor is promoted into planning, so no alarm fires — but holding is not availability, and within a year the knowledge is operationally gone, its holder alive, employed and two buildings away. Reorganizations also break the relationships layer inventories persistently underrate: who at the utility answers after hours, which tenant contact can authorize access. The audit treats a reorganization as a trigger firing a targeted refresh during planning, while moves can still be sequenced — organizations running it through a restructuring often find it altering the restructuring in small, cheap ways: a move delayed a quarter to complete a closure sprint, dual coverage added for a critical building. The limiting case is the absence nobody schedules — illness, injury, resignation-with-two-weeks — which produces without warning exactly the situation the Handoff Test simulates.

Crawl, walk, run — and five maturity levels

Implementation is staged. Crawl is a bounded 60–90-day pilot: one team, building cluster or the three most-consulted people; a one-page governance charter — including the evaluation firewall — communicated before the first interview; metrics baselined from day one, because closure work contaminates the baseline it was supposed to be measured against; the pilot’s closure budget spent exclusively on the red quadrant; and at the end, one live Handoff Test and three numbers — the baseline, the movement, and the cost in hours. A pilot that cannot beat its baseline is treated as a finding, not a failure. Walk extends the inventory across the portfolio in criticality order, no faster than closure capacity can honestly sustain — “an inventory that grows faster than its red quadrant shrinks is accumulating anxiety, not continuity” — with cadence in writing and triggers wired to where events first become visible: resignation notices routed from HR, contract final-year flags from procurement, a seat at reorganization planning. The closure backlog is managed with the same weekly discipline as the maintenance backlog it resembles — deferred maintenance on the organization’s knowledge. Run is embedded operations: transition-out schedules in vendor contracts, an operating-knowledge handover in project closeout alongside the as-builts, the inventory syllabus in onboarding, the refresh on the annual operating calendar. At this stage the audit passes its own test — method documented, successor named — because an audit that would not survive its founder is a finding about the audit.

A five-level maturity model locates teams before they plan a route: Undocumented, Anecdotal, Inventoried, Managed, Embedded. Most organizations assessing themselves honestly land at Level 0 or 1, and the paper is explicit that this carries no embarrassment — the practice has not historically existed as a discipline, so its absence is the industry’s default. The move from 0 to 2 is the pilot and is achievable inside a quarter; the move from 2 to 3 — from having done an audit to operating one — is where most programs stall; Level 4 is less a destination than a disappearance, the audit absorbed into contracts, closeouts and calendars, which is where continuity practices go when they succeed.

Baselines before benchmarks

The measurement chapter publishes no industry benchmark for any metric, declines to invent one, and regards any program justified by someone else’s numbers as unjustified. Six metrics, each capturable with existing systems or a two-week sampling exercise: the Handoff Test score by role; the concentration index — the share of sampled located-knowledge queries resolved by the three most-consulted people, the program’s early-warning gauge, reported at team level and never as a league table; time to independent competence; rework attributable to missing located knowledge; closure-backlog burn-down; and inventory recency, the metric that keeps the others honest, reported beside every number the way a laboratory reports a calibration date. An illustrative cost model — every input a stated assumption, not a benchmark or promise — sketches a 20-person technical team across a 25-building portfolio with three departures anticipated over 24 months: roughly 300–390 hours of year-one effort against on the order of 545 hours recovered, before counting a single avoided emergency escalation, disputed contract handover or failed audit finding. The point is the shape, not the outputs: bounded, predictable effort against open-ended, compounding exposure.

Governance and honest limits

Because the audit gathers information from individuals about the organization’s dependence on them, governance is treated as a condition of getting true answers, not a compliance appendix. Five rules constitute the minimum: a firewall — audit material feeds no performance evaluation, compensation decision or workforce-reduction analysis, stated in the charter, said aloud in every interview, and kept, because “the first exception ends the program’s supply of honesty — permanently”; participation framed as legacy, not extraction; access control on outputs, which describe single points of failure by name; data-protection compliance by design, with collection minimized, retention rule-based, and works councils designed with rather than around; and dignity in the mechanics — the live test debriefed as a systems test with the role’s holder as its expert consultant, not its examinee, and closure work planned and credited as workload.

The limitations chapter is candid. The audit inventories the locatable layer, not judgment — diagnostic intuition and the feel for a chiller’s sound the week before it fails are developed, not documented, and any program promising to capture everything a veteran knows is overpromising by roughly the amount that matters most. Scoring is structured judgment, not measurement — a reason to track movement rather than worship absolute values. A lapsed inventory is worse than none in one specific way: it manufactures false confidence, and presented as current in a dispute it is a liability. One misuse of audit material in a personnel decision does not damage the program; it ends it. And the audit is not for everyone in equal measure: a five-person team with genuine knowledge redundancy may honestly conclude after a one-day exercise that its red quadrant is empty — a legitimate conclusion, reached by the audit at exactly the scale the risk deserved. The outlook — speech-to-text and language models collapsing elicitation costs, text mining over CMMS history surfacing hotspots as a standing dashboard, convergence with the building’s spatial record — makes capture cheaper without supplying cadence, ownership, honesty or a firewall: the durable investment is the practice. The paper aligns that practice with ISO 30401 (supplying a facility-native method for the knowledge-loss requirements most organizations experience as abstract), ISO 41001, the ISO 55000 series, IFMA’s competency framework and COBie handover conventions.

The closing argument is structural: the Continuity Audit is preventive maintenance — inspection on a cadence, findings in a backlog, work orders with owners and dates — applied to an asset class the CMMS has no module for and the balance sheet has no line for, which nonetheless determines how every other asset performs the morning after a veteran’s last shift. The retirements are scheduled; the contract end dates are signed; the reorganizations are already on someone’s whiteboard. The paper is vendor-neutral, naming no commercial platform; the full instruments, tables, figures and reference list are in the PDF.