Engagement snapshot
- Mandate
- Reduce knowledge fragmentation without creating a heavy documentation mandate that people would ignore.
- Timeline
- 10 weeks from audit to chosen-stack rollout guidance.
- Team shape
- Knowledge lead, IT manager, 2 solution architects, and 5 department owners.
The problem
Employees were split across wikis, drives, chat histories, and meeting transcripts, so finding current policy or client context took too long.
What we built
Selected and connected a knowledge stack with source-of-truth boundaries, search layers, retention rules, and adoption playbooks.
Operating context
The company had no single knowledge problem. It had many small competing knowledge systems that all felt partially current. Search quality and trust suffered because no one knew which sources truly mattered for which question types.
Key constraints
- Different departments had different documentation habits and compliance obligations.
- Search needed to respect retention and access boundaries.
- Adoption would fail if the new model depended on perfect documentation behavior.
What we built
Source-of-truth classification
Defined which information classes belonged in which systems and which content should never be duplicated.
Search and retrieval layer
Designed a search approach that respected access and freshness while making high-value content easier to find.
Adoption governance
Created ownership and maintenance rules so the knowledge system would stay current after rollout.
Delivery path
Knowledge audit
Reviewed where employees searched, where they found stale information, and where duplication was causing confusion.
Stack and model selection
Chose tools and a structure that fit both usability and policy constraints.
Adoption rollout
Prepared migration priorities, ownership assignments, and training patterns for business teams.
Why it mattered
The project improved knowledge flow because it attacked ambiguity about where truth lived. Search became better because the content system became clearer, not because a search box alone got smarter.
Implementation notes
- Knowledge stacks need source-of-truth boundaries before they need better search.
- Adoption depends on realistic maintenance ownership, not aspirational documentation standards.
- Reducing duplication is often the fastest way to improve confidence in knowledge systems.