Naming Conventions That Survive

Naming conventions are the single most underinvested component of knowledge architecture. Organizations spend weeks designing taxonomy structures and zero time establishing the rules for how individual documents, categories, and metadata fields will be named. Then they spend years dealing with the consequences: a knowledge base where some articles are titled 'Q3 Account Planning Process', others are 'Account Planning Q3 Guide', others are 'Guide: How to Do Account Planning (Updated)', and the search index has to work around inconsistent labeling that makes every query a coin flip.

A naming convention is a set of rules that makes knowledge artifact titles predictable, consistent, and searchable across all contributors and all time periods. The rules must be simple enough that anyone can follow them without consulting a style guide, specific enough that they eliminate the ambiguity that produces inconsistency, and durable enough that they remain valid as the organization grows and changes.

1. Content Type First Start every title with the content type: Policy, Procedure, Reference, Runbook, Decision, Guide, Template. This ensures that a search for 'contract review' surfaces all knowledge about that topic, categorized by what kind of artifact it is. Users can immediately distinguish between 'Policy: Contract Review Requirements' and 'Runbook: Contract Review Process' and 'Reference: Contract Review Checklist' — three different things that serve different purposes.
2. Consistent Domain Language Establish a controlled vocabulary for domain terms and enforce it in titles. If the organization calls it 'account planning' in official communications, every knowledge artifact about that topic must use 'account planning' in the title — not 'territory planning', 'strategic planning', 'account strategy', or any other synonym that feels equivalent. Synonym proliferation is the primary cause of retrieval failure in knowledge bases with large corpora.
3. No Dates or Version Numbers in Titles The single most common naming convention mistake: titles like 'Sales Process v2.3 Updated April 2024'. Within a year the document is still titled 'Updated April 2024' and is now badly out of date. Use the title for identity — what the document IS — and use metadata fields for version and date. A document titled 'Procedure: Sales Discovery Process' is always findable regardless of when it was last updated.
4. Sentence Case, Not Title Case This sounds trivial and matters significantly at scale. 'Customer Onboarding Runbook' and 'Customer onboarding runbook' are different search results in some systems and confuse contributors about the 'right' format. Pick one. Enforce it. The choice matters less than the consistency.

Naming conventions must be enforced at the point of creation, not retroactively. A knowledge governance process that allows anyone to publish a document with any title and then tasks a KM team with normalizing titles after the fact will always be losing ground. The enforcement mechanism is the contribution workflow: a knowledge artifact that does not follow the naming convention is not published until it does. This requires a governance process with a clear owner and enough organizational authority to enforce the standard — which is why naming conventions are a governance problem as much as an architecture problem.