Text vs Keyword in Elasticsearch: Key Differences and Usage

In Elasticsearch, text fields are analyzed and broken into tokens for full-text search, while keyword fields are not analyzed and are used for exact matching and sorting. Use text for searching words inside content and keyword for filtering or aggregating exact values.

Quick Comparison

This table summarizes the main differences between text and keyword fields in Elasticsearch.

Aspect	Text Field	Keyword Field
Purpose	Full-text search with analysis	Exact value matching and sorting
Analyzed	Yes, broken into tokens	No, stored as is
Use Case	Search inside large text	Filter, sort, and aggregate exact values
Storage	Takes more space due to analysis	Compact, stores exact string
Performance	Slower for exact matches	Faster for exact matches
Example Query	Match query	Term query

Key Differences

Text fields in Elasticsearch are designed for full-text search. They are analyzed, meaning the text is broken down into smaller pieces called tokens (like words), and these tokens are indexed. This allows Elasticsearch to find documents that contain the search terms anywhere inside the text, supporting features like stemming, stop words, and relevance scoring.

On the other hand, keyword fields are not analyzed. They store the entire string exactly as it is. This makes them ideal for filtering, sorting, and aggregations where you want to match the exact value, such as tags, IDs, or categories. Since they are not tokenized, they cannot be used for full-text search.

Choosing between text and keyword depends on your use case: use text when you want to search inside the content, and keyword when you want to filter or sort by exact values.

Code Comparison

Here is an example of defining and querying a text field for full-text search.

json

{
  "mappings": {
    "properties": {
      "description": {
        "type": "text"
      }
    }
  }
}

POST /my_index/_search
{
  "query": {
    "match": {
      "description": "quick brown fox"
    }
  }
}

Output

{
  "hits": {
    "total": {"value": 2},
    "hits": [
      {"_source": {"description": "The quick brown fox jumps."}},
      {"_source": {"description": "A quick fox is clever."}}
    ]
  }
}

Keyword Equivalent

Here is how you define and query a keyword field for exact matching.

json

{
  "mappings": {
    "properties": {
      "status": {
        "type": "keyword"
      }
    }
  }
}

POST /my_index/_search
{
  "query": {
    "term": {
      "status": "active"
    }
  }
}

Output

{
  "hits": {
    "total": {"value": 3},
    "hits": [
      {"_source": {"status": "active"}},
      {"_source": {"status": "active"}},
      {"_source": {"status": "active"}}
    ]
  }
}

When to Use Which

Choose text fields when you need to search inside large bodies of text, like articles, descriptions, or comments, where partial matches and relevance matter. Choose keyword fields when you need to filter, sort, or aggregate by exact values, such as tags, IDs, or status codes. Mixing both in your index is common to support both search and filtering efficiently.

Key Takeaways

Use text fields for full-text search with analyzed content.

Use keyword fields for exact matching, filtering, and sorting.

Text fields break content into tokens; keyword fields store exact strings.

Mix both field types in your index to support versatile search and filtering.

Queries differ: use match for text and term for keyword fields.