Modern search experiences need to handle more than exact keyword matches.
Users should be able to write natural language queries, use synonyms, express themselves differently from the available content, or make spelling errors.
SitecoreAI Search Experiences includes several capabilities designed to address these scenarios, including semantic reranking, fuzzy search, and relevance rules such as boost, bury, and pin.
This article explains what these features do, how they are configured, and how they interact.
What Is a Search Experience in SitecoreAI?
Sitecore defines a Search Experience as the section of a web page that allows site visitors to search within a predefined content source and view the results with a rich search experience.
SitecoreAI currently supports different types of search sources:
- Content sources, which index structured content from the Content Editor (Sitecore)
- Site sources, which discover and index website content through web crawling.
- Push sources, which receive documents directly from external systems through the Ingestion Service API.
A search source also stores the configuration that controls how its content is indexed, filtered, and ranked.
What Is Semantic Reranking?
Sitecore defines semantic reranking as an advanced search setting that reranks the initial keyword results by using AI to better understand the meaning, intent, and context behind a search query.
According to Sitecore, semantic reranking can help return relevant results when users search using
- Synonyms
- Natural-language phrases
- Different wording from what appears in the content.
A query such as:
| Search Query | Search Results |
|---|---|
| How do I improve team leadership skills? | – Guide to becoming a better manager – Leadership development training resources |
Even though the wording in the query does not exactly match the wording in the indexed content.
Semantic Reranking Does Not Replace Keyword Search
Sitecore’s documentation specifically states that semantic reranking works with initial keyword results.
Semantic reranking should therefore not be described as a replacement for keyword-based search.
Field configuration within the search source remains important.
Sitecore lets content authors configure how each field behaves within a source. Fields can be configured as:
- Included
- Searchable
- Filterable
- Sortable
A field marked as Searchable can participate in searches performed against that source.
For example, a source might contain fields such as:
- Title
- Description
- Category
- Author
Depending on the requirements of the search experience, only some of those fields might be configured as searchable.
How to Enable Semantic Reranking
Semantic reranking is configured as an Advanced Setting of Sitecore Search
For a Site Source, Sitecore documents a configuration process in which you create the source, define indexing behavior, configure the fields, and then enable any required Advanced Settings before saving the source.
The Advanced Settings currently documented by Sitecore are:
- Semantic reranking
- Fuzzy search
Sitecore also states that feature availability is part of a phased rollout:
“Feature availability is part of a phased rollout. Your organization may not see this functionality yet. It will become available when your environment is included in the rollout.”
As a result, some organizations may not yet see these capabilities in their SitecoreAI environments.
Sitecore also announced Semantic Search for Sitecore Search as an Early Access capability for Search customers with a Sitecore Stream license.
For the Early Access version of Sitecore Search Semantic Search, implementation guidance published by Sitecore community experts describes an enablement step before the semantic search API can be used: customers request activation through Sitecore Support and specify which fields should participate in semantic search.
Semantic Reranking vs. Fuzzy Search
Semantic reranking and fuzzy search address different problems within a search experience.
Semantic Reranking
Semantic reranking does not replace keyword-based search.
According to Sitecore’s documentation, the search first produces initial keyword-based results. Semantic reranking then uses AI to reorder those results based on the meaning, intent, and context of the query.
This can improve result ordering when the user’s query and the indexed content use different wording but express related concepts.
The key point is that semantic reranking works on top of the initial keyword search results and changes their order. Sitecore does not document it as a mechanism that completely replaces textual search.
Fuzzy Search
Fuzzy search addresses a different problem.
Its purpose is to tolerate spelling mistakes and minor variations in the text entered by the user.
For example:
| Search Query | Search Results |
|---|---|
| prodcut managr | – Product manager job description |
Sitecore states that fuzzy search can handle characters that are:
- missing
- added
- out of order
When semantic reranking and fuzzy search are enabled at the same time, Sitecore documents that fuzzy matching only activates after a search returns no results.
Boost, Bury, and Pin
In addition to semantic reranking and fuzzy search, SitecoreAI lets you configure rules that directly influence the order of search results.
These rules make it possible to adjust the relevance of specific content without relying only on the search engine’s automatic ranking behavior.
Boost increases the relevance of an item or group of items, making them more likely to appear higher in the search results.
For example, it can be used to give greater visibility to content that the organization considers especially important for specific searches.
Bury does the opposite.
It decreases the relevance of selected items, causing them to appear lower in the search results.
This can be useful when content is still valid and searchable but should not receive as much prominence.
Pin lets you place a specific item in a defined position within the search results.
Unlike boost and bury, which modify relevance, pin directly controls where a result appears.
For example, a piece of content can be configured to always appear in the first position when the conditions defined by the rule are met.
How Rules Interact with Semantic Reranking
Sitecore allows semantic reranking to be combined with relevance rules.
- Boost and Bury can change result ordering even when semantic reranking is enabled.
- Pin keeps a result in its configured position regardless of its semantic relevance.
This allows semantic reranking to be used without losing control over content that requires specific prioritization.
What Types of Content Can SitecoreAI Search?
SitecoreAI Search Experiences can work with different source types depending on where the content is stored.
A Content Source indexes published structured content from the Content Editor.
It is useful when the content you want to search already exists in Sitecore, such as:
- articles
- events
- products
- other structured content types
SitecoreAI ingests published items based on the selected template.
A Site Source indexes content directly from a website.
SitecoreAI discovers pages through sitemaps, crawls them, extracts the configured fields, and stores that information in a search index.
This source type can also be used with websites whose content does not come from the Content Editor.
A Push Source allows content to be sent directly to SitecoreAI through an API.
It is intended for content managed in external systems, such as:
- product catalogs
- knowledge bases
- custom applications
- content that cannot be crawled
Documents are sent through the Ingestion Service API, which also gives the external system control over when the index is updated.
In summary, SitecoreAI can build Search Experiences using content managed in Sitecore, content discovered through crawling, or information pushed directly from external systems.
Conclusion
SitecoreAI Search Experiences extends traditional search with capabilities designed to improve relevance and handle different types of user queries.
Semantic reranking works on top of the initial keyword-based results and reorders them using meaning, intent, and context. Fuzzy search helps with spelling mistakes and character variations, while boost, bury, and pin provide explicit control over result ordering.
These capabilities do not replace traditional search configuration. They complement it.
The real value of semantic reranking should be validated with representative content and queries from each implementation by comparing how results change before and after enabling it.
References
Sitecore Documentation – Advanced settings
Sitecore Documentation – Search experiences
Sitecore Documentation – Rules
Sitecore Developer Portal – Advanced ranking rules and settings in Search experiences
