Knowledge base search fails on content quality, not search settings

Teams adjust search settings while the underlying library holds three versions of the same policy, two of them out of date. The search layer finds the wrong source because the library itself is unclear.

Fix the corpus first

• Remove superseded versions rather than keeping them for reference
• Record effective dates so recency can be used at search time
• Chunk on meaning boundaries, not on a fixed character count
• Attach metadata that filters by product, region and audience

Build a judgement set

Collect real questions from users and mark which passage contains the correct answer. Without this, every search change is guesswork and regressions go unnoticed.

Measure search separately

Score whether the right passage was found before scoring the final answer. Mixing the two hides whether the problem is in search or in the response layer.

Practical checklist

• Define the acceptance rule before any volume starts
• Review a small pilot before committing the full budget
• Track errors by category, language and reviewer
• Keep consent, source notes and version history with the files

Before you ask for a quote

A clear brief saves days. Share a sample file, target language or region, expected volume, deadline, quality threshold and any privacy restrictions. A supplier can then price the work on real effort rather than assumptions.

• Which languages, markets or user groups must be represented?
• What format does the final file need to arrive in?
• Who will approve ambiguous cases during the pilot?