How is search result relevance determined?

{{variable.Company Name}} uses Intelligent Search to deliver smarter, more accurate results. This combines traditional keyword-based ranking with AI-powered vector search, allowing the system to better understand search intent and context — especially for natural language queries.

Search relevance is influenced by:

Keyword matches in titles, summaries, and tags
Vector search, which analyses the meaning of your query
User interaction data, including completions, likes, and dismissals (especially from users within your reporting line)

Short queries (3 words or fewer) use both keyword and vector search. Longer, natural-language queries use vector search alone for more precise intent matching.

What fields are searched for each {{glossary.Content}} type?

Content Type	Search Fields
File	Title, Summary, Tags
Link	Title, Summary, Tags
Article	Title, Body, Tags
Question	Question, Summary, Tags
Pathway	Title, Summary, Tags
E-learning	Title, Summary, Tags
Quiz	Title, Summary, Tags
Broadcast	Title, Tags
Event	Title, Summary, Tags

With Intelligent Search, content stored in documents (e.g. PDFs) is also indexed, improving discoverability across your full content library.

How are different search result components weighted?

Search results are ranked using the following priority:

Title
Summary or Body
Tags
Other metadata (e.g. file type, creation date)

The Intelligent Search model further enhances this by analysing semantic relevance, so content with a strong conceptual match to the search query may be ranked higher, even if exact keywords aren't present.

Does {{variable.Company Name}} support fuzzy searching?

Yes. {{variable.Company Name}}’s search includes fuzzy matching to account for typos and partial words. For example, searching for “proj” will still return “project management” and other relevant results containing that fragment in the title, summary, or tags.

How does phrase searching work?

When you search using a phrase, {{variable.Company Name}} prioritises results where the words appear together in order. If an exact match isn't found, it will fall back on results containing individual words.

{{variable.Company Name}} uses multiple matching techniques for phrase detection, including:

Exact phrase match in title and description
Partial phrase match across tags and other metadata
Fallback to keyword matches when phrases aren’t found

Intelligent Search enhances this by analysing meaning — not just word order — so even loosely phrased searches can return helpful, accurate results.

Do quotation marks affect search results, like in Google?

Using quotation marks for a single word will return only exact matches of that word (e.g. "Leader" will exclude variations like "Leadership").
For multi-word phrases, quotation marks do not currently force exact phrase matching. However, {{variable.Company Name}}’s Intelligent Search model still works to find contextually relevant matches, even without relying on exact keyword phrasing.

What is the current behaviour of content search?

Single-word searches: Return a wide range of results by matching titles, summaries, and tags.
Two-word searches: Narrow results slightly, with preference given to items where both words appear together.
Three or more words: Trigger AI-powered vector search, which interprets the full query’s meaning to return contextually relevant content — even if exact terms aren’t included.

How many tiles can appear on the Explore page?

Level 1 and Level 2 of Explore (typically categories or collections) have no item limit.
Level 3 (content items within a collection) displays 50 items per page, sorted by most recent first. Users can page through to view more content if a collection exceeds this limit.

Search and Discovery