$31.5B
Annual losses across the Fortune 500 attributable to forgotten organizational knowledge.
— International Data Corp. (IDC)
Every year, organizations forget how to do the very things they exist to do — at an estimated cost of $50 billion. The Organizational Amnesia Project is a long-running study of that forgetting: its causes, its symptoms, and the institutions it hollows out.
Organizational amnesia is the gradual, often silent loss of an organization's collective memory — the know-how, judgment, and context that lived in the people, processes, and pages that made the place work.
It is not a data breach. Nothing is stolen; nothing is even, in the strict sense, deleted. A senior engineer takes a new job and the only person who understood the failover script is gone. A reorg buries the postmortem from the last outage two folders deeper than anyone will ever look. A migration drops the metadata that explained why a clause exists in the contract template. Multiply by ten thousand small acts of forgetting. The organization, in aggregate, has stopped knowing what it knew.
Researchers describe the phenomenon as accidental, unintentional evaporation of knowledge that accumulates from learning and from individual and collective actions. We use the term amnesia deliberately: like the human kind, it is rarely total, often unnoticed, and almost always more expensive than the institution suspects.
In a longitudinal Project Management Institute study of a century-old firm, researchers found the institution lost its bearings roughly every 20 years — one full leadership generation. We reproduce the pattern here.
The most studied cause. Voluntary and involuntary turnover; retirement, layoff, restructure. Tacit knowledge — the kind that resists documentation — leaves with the body that held it. Above 10% annual turnover, productivity begins to degrade in measurable ways.
Knowledge that exists but is not used drifts toward the unusable. Old runbooks rot in old wikis. Process documents go stale. The information is technically present; nobody trusts it enough to consult it.
The same fact lives in seven systems and three Slack threads, each authoritative, none of them final. Disconnected platforms create silos that trap explicit knowledge. The knowledge is everywhere and so, effectively, nowhere.
Selective recall. The team remembers the wins and forgets, with surprising thoroughness, the failures it would most benefit from learning from. PMI calls this giving lip service to lessons learned; it is the most common documented form of organizational amnesia.
Generational handover. Reorganization. Acquisition. Each is a moment at which the implicit "how we do things" can fail to make the jump from the old constellation of people to the new one. PMI's longitudinal work finds an institution loses its bearings about once a generation.
Forgetting is, methodologically, a problem. The thing you're trying to study has already left the building. Our approach combines four instruments.
A structured inventory of explicit, implicit, tacit, and collective knowledge across a representative sample of teams. Conducted in partnership with the firm's people-ops function.
Long-form qualitative interviews with employees within 60 days of leaving. Captures what they know they knew and, more importantly, what nobody asked them about before they left.
Anonymized, aggregate metrics on how long it takes employees to find information; how often the search ends in a colleague rather than a document; how often the document is wrong.
For each material project failure in the study period, we trace whether the contributing knowledge had previously existed in the organization — and if so, where it went.
A 12-question diagnostic, derived from the project's field instrument, that estimates the memory health of your organization in roughly four minutes. Scored on a 0–100 scale across four bands: Intact, Eroding, Amnesiac, Acute.
Begin the diagnostic→An interactive estimator that converts headcount, turnover, salary, industry, and tenure into a defensible per-employee and total-firm dollar cost of organizational amnesia. Built from Panopto, McKinsey, IDC, and Gartner figures.
Open the calculator→The field-tested symptoms — drawn from the project's longitudinal sample — that indicate institutional knowledge is being lost in measurable ways.
Read the field guide →Why the AI-memory problem and the organizational-memory problem are isomorphic — and what RAG, agents, and long-context models do (and don't) fix about institutional forgetting.
Read the essay →Books, peer-reviewed papers, industry reports, and lectures on organizational memory, knowledge management, and AI memory — annotated for relevance, updated quarterly.
Open the reading list →Closely related but not identical. Knowledge loss is the broader phenomenon — anything that removes institutional know-how, including deliberate decisions to stop tracking something. Organizational amnesia is specifically the unintentional, accidental subset: the knowledge the organization didn't mean to lose.
Documentation helps with explicit knowledge — the policies, runbooks, and step-by-step procedures that already exist in writeable form. The harder problem is tacit knowledge: the judgment, relationships, and instinct that don't reduce well to bullet points. The literature is consistent that "write it down" is necessary but never sufficient.
The most-cited figure is IDC's estimate of $31.5B in annual losses across the Fortune 500 from forgotten organizational knowledge. Kransdorff has put the upper bound, including "experiential non-learning," at up to 9.7% of GDP in developed economies. We aggregate these and project upward of $50B in direct annual losses across the global enterprise base.
The project is operated as a research initiative of Reattend, with academic partners noted on the Lab page. All findings are released openly under CC BY 4.0; participating firms are anonymized.
Yes — intake for the 2026–2027 cohort is open. Email lab@reattend.com with a one-paragraph description of your firm and what you'd like to learn.
Reattend builds tools to mitigate organizational amnesia in practice. The project is upstream of that work — we'd publish the same findings if no product existed. Reattend.com is the product side of the same problem.
Closely. The retrieval, ranking, and freshness problems that organizations have always had with their wikis, runbooks, and decision records are the same problems AI engineers are now solving with RAG, vector stores, and agent memory. We argue in a dedicated essay that AI memory and organizational memory are isomorphic — and that the older knowledge-management literature has 30 years of warnings the AI-memory community is now beginning to rediscover.
Departure-driven knowledge loss is one of the five mechanisms (we call it departure); employee turnover is its main driver. But organizational amnesia is broader — knowledge can also evaporate via decay, dispersal, defensiveness, and discontinuity, even in firms with zero turnover. A team that never loses a person can still lose its institutional memory through disuse, fragmentation across systems, and reorgs. See the five mechanisms.
If your organization, today, lost the five people who tacitly know how everything works — could it remember itself by Monday?
Most organizations cannot. Knowing where you stand on that is the first useful thing.