Search Quality Checks for a Growing Wiki

A knowledge base gets less useful if readers cannot find the right page. Search quality checks should grow with the wiki, especially after daily article batches add new categories and overlapping topics.

Build a query set

• Setup: install, provider, model, profile.
• Operations: cron, background task, deployment, verification.
• Security: secrets, approvals, private data, screenshots.
• Content: wiki article, public page, sitemap, visual QA.
• Troubleshooting: failed deploy, broken image, console error, fallback.

Check ranking, not just result count

• The top result should match the intent.
• A specific query should beat a generic page.
• Snippets should help the reader decide where to click.
• Category filters should not hide obviously relevant results without a clear reason.

Keep metadata useful

• Titles should use the words readers search for.
• Summaries should be specific, not marketing fluff.
• Tags should include synonyms and workflow names.
• Categories should stay balanced enough to browse.

Pitfalls

• Adding many near-duplicate pages.
• Letting old pages outrank newer, more specific guides because their body happens to contain a keyword.
• Forgetting to recompute category counts.
• Shipping new pages that are missing from the sitemap.

Verification steps

• Run deterministic local search checks.
• Open /wiki?q=... for representative queries.
• Verify category pages on desktop and mobile.
• Fetch the live sitemap and check new URLs.
• Review search logs or feedback when available.