What goes wrong when you ignore that the organization's knowledge graph?

The most common reason fails: Confusing documented knowledge with operational knowledge. An organization can have extensive documentation — wikis, runbooks, architecture diagrams — and still have a fragile knowledge graph if no one has internalized the documented knowledge well enough to act on it under pressure. Documentation is a node in the knowledge graph, not a substitute for the graph itself. The knowledge graph includes tacit knowledge (how the system actually behaves, not just how it is supposed to behave), relationship knowledge (who to call when the documented process does not work), and contextual knowledge (why the system was designed this way, what alternatives were considered and rejected). Documentation captures explicit knowledge. The knowledge graph includes explicit, tacit, and relational knowledge.

The fix: Create a knowledge map for your team. List the five to ten most critical knowledge domains for your team's work. For each domain, list every team member and rate their knowledge level: 'can teach' (4), 'independent' (3), 'with documentation' (2), 'no knowledge' (1). Sum each domain's scores and divide by the number of team members to get an average knowledge density. Flag any domain where only one person scores 3 or above — these are your fragile knowledge nodes, where a single departure would create a critical gap. Flag any domain where no one scores 4 — these are domains where the team lacks deep expertise. Use the map to plan knowledge distribution activities for the next quarter.

The underlying principle is straightforward: Every organization has a knowledge graph — a network of expertise, institutional memory, relationships, and documented information that its schemas operate on. Mapping this graph reveals where knowledge is concentrated, where it is fragile (held by a single person), where it is redundant, and where critical gaps exist. The knowledge graph is to the organization what working memory is to the individual: the substrate that schemas operate on.