Step 1
Choose a high-value source set
Start with a bounded collection such as onboarding docs, support policy pages, or one product area. Importing everything at once makes normalization harder and reuse decisions weaker.
Guide
Import a knowledgebase without carrying over old drift.
Importing is not just moving files. It is the process of normalizing existing content into pages and reusable snippets so the new system can actually improve maintenance, impact analysis, and publishing.
Goal
What a good import looks like
A strong import preserves meaning, identifies duplicate content, and converts repeated sections into snippets early. The outcome should be a clean knowledgebase that is easier to edit and publish than the source material, not a one-to-one archive of the old mess.
Workflow
Recommended import sequence
Step 2
Map source documents to page outcomes
Before importing, decide which final pages should exist in the knowledgebase. This prevents the new structure from inheriting arbitrary legacy organization.
Step 3
Extract repeated content into snippets
As you move content in, isolate anything that appears across multiple pages: disclaimers, configuration steps, escalation rules, definitions, or AI-policy constraints.
Step 4
Review composition page by page
After snippet extraction, inspect each page in its rendered form. The page should still read as a coherent document, not a stack of disconnected modules.
Step 5
Run a low-risk release first
Publish through a controlled destination or branch, inspect the generated output and manifest, and only then expand import scope.
Heuristics
What should become a snippet during import
Good snippet candidates are content blocks with stable meaning and repeated use: account verification rules, incident severity criteria, onboarding prerequisites, plan limitations, or style-governed answer templates.
Bad snippet candidates are paragraphs that only make sense in one page’s local narrative or content that changes so often and so uniquely that reuse adds overhead instead of leverage.
Next move
Import with structure, not with nostalgia for the old file tree.
The purpose of migration is to improve the operating model. If repeated text stays duplicated after import, the most important benefit of SnippetGraph is still unrealized.