This isn't a technology problem. It's a knowledge architecture problem. And in 2025, it went from chronic to acute.
Federal workforce knowledge loss has been a recognized problem for decades. But in 2025, it became an emergency. Across the federal government, approximately 238,000 workers departed while only 116,912 were hired — a net loss of roughly 121,000 positions in a single year.[1]
Some agencies were hit harder than others. The Department of Education lost 42.6% of its staff. The IRS lost approximately 40% of its IT workforce. The National Science Foundation lost 40% of its PhD-holding workforce. CDC, NOAA, and EPA all experienced significant reductions.[1]
The scale of this workforce reduction is unprecedented in modern federal history. Every departure takes institutional knowledge with it — knowledge about how systems actually work, which contacts to call, what was tried before and why it failed. At this pace, agencies aren't just losing people. They're losing the ability to function.
“Strategic human capital management” has been on GAO's high-risk list since 2001 — 24 consecutive years.[2]Of the 37 items on GAO's 2025 high-risk list, 20 stem directly from agencies not having the right skills or enough people.[2] One-third of federal employees are eligible to retire within five years.[3]
When a senior analyst retires, it's not just their individual expertise that walks out the door. It's the entire map of who-knows-what that collapses. Nobody even knows what was lost.
Before 2025, this was a slow leak. Now agencies are losing decades of accumulated expertise in months. The compound effect of thousands of simultaneous departures doesn't just add up — it multiplies.
Most agencies rely on SharePoint sites, shared drives, wikis, and informal mentorship to transfer knowledge. After-action reviews — when they happen — are typically Word documents produced from manually facilitated sessions.
The Department of Defense built JLLIS (Joint Lessons Learned Information System) to capture operational lessons. GAO flagged it for “limited functionality” that “impedes information sharing.”[2] SOLLIMS (Stability Operations Lessons Learned and Information Management System) was recently sunsetted entirely, creating a gap with no replacement.
Meanwhile, 80% of the federal IT budget — over $100 billion annually — goes to maintaining legacy systems that create the very silos knowledge management is supposed to bridge.[2]
FOIA requests take 2-4 hours each with 18-month backlogs. Onboarding a replacement analyst costs roughly $80,000 in ramp-up time. Multiply by hundreds of rotations per year — that's $10-50 million annually in lost productivity per agency.
The government knowledge management software market is estimated at $2-4 billion and growing 13-18% annually[4]— a measure of how much agencies are already spending to solve this problem, mostly with tools that weren't designed for it.
| Before | After | |
|---|---|---|
| Time per FOIA request | 2-4 hours (manual review) | 5-15 minutes (structured triage) |
| Request backlog | 18-month average | 3-6 months |
| Analyst expertise needed | Senior (knows what to redact from memory) | Any (system identifies patterns) |
| Knowledge when analyst leaves | Gone | Captured in the system |
| After-action reviews | Word docs from facilitated sessions | Structured, searchable, linked to outcomes |
| Onboarding cost | ~$80,000 per replacement | Dramatically reduced |
| Cross-department visibility | Siloed — each team on its own | Connected — links surfaced automatically |
You can't measure what you don't know you lost. Unlike a server outage or a budget shortfall, knowledge loss doesn't trigger any alarm. That's why it stayed on GAO's high-risk list for 24 years without resolution.
No single departure used to feel catastrophic. The compound effect of dozens of departures over years was devastating but never urgent on any given day. In 2025, agencies lost 40% of entire workforce segments in months.
SharePoint stores files. JLLIS was flagged for limited functionality. SOLLIMS was sunsetted. Static documents don't self-update, don't flag when they're stale, and don't connect across departments.
Databases store data. Knowledge management requires architecture — a system that understands what's important, what's decaying, and what connects to what. That's a design problem, not a procurement problem.
Let me show you exactly where — and what it's costing you.
Request a Demo