This is a discussion specifically about actions that Glean users can take to improve Glean Search results quality. Please ask questions and offer observations based on your experiences.
A common question type that comes up is how to improve the quality of Glean Search results. The question is usually something like:
- "How do I get a document to return first in the search results?"
- "How do we tell Glean to favor data source A over data sources B and C?"
- "Why don't I see documents XYZ in the search results?"
Before we start talking about direct responses to those questions, let's provide a little background. Glean uses machine learning to train relevancy models based on the company's enterprise data. There are many features that it trains to optimize the quality of search results to provide a personalized experience that is most relevant to the user executing the search. This is the foundation that the Glean experience is built on. Glean doesn't offer direct control over the relevancy model parameters because that can have surprising, unintended, and far-reaching consequences. However, there are actions that users can take to improve search result quality.
Here are some of the most effective approaches:
- Glean Browser Extension. Especially helpful if used by all users.
- Glean Verification feature. Verify known-good documents. Deprecate old documents.
- Glean Pins feature. Pin search queries to specific search results.
Have these worked for you? What other approaches have you tried?
