> ## Documentation Index
> Fetch the complete documentation index at: https://docs.bondata.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Web Search

> Search the web for each record and store results as a new field

Performs a web search for each record using Exa and stores the results as a new field. Use it to find company websites, news articles, LinkedIn profiles, or any public information.

<Frame caption="Web Search node - dynamic query with Mentions, search type, and result configuration">
  <img src="https://mintcdn.com/bondata/DlMd45I1zc1MvbNc/images/guide/nodes/enrichment/web-search.png?fit=max&auto=format&n=DlMd45I1zc1MvbNc&q=85&s=2a0d9a95c9edc251d78b36eba57680b1" width="709" height="481" data-path="images/guide/nodes/enrichment/web-search.png" />
</Frame>

## Configuration

| Setting               | Description                                                                          |
| --------------------- | ------------------------------------------------------------------------------------ |
| **Search Query**      | The search query. Use `{{field_name}}` or **Mentions** to make it dynamic per record |
| **Search Type**       | **Auto** (high quality, \~1s latency), **Keyword**, or **Neural**                    |
| **Number of Results** | How many results to return per query (max 100)                                       |
| **Result Field Name** | The field name where search results are stored                                       |
| **Filters**           | Narrow results by domain, date, or other criteria                                    |
| **Contents**          | Control what's extracted from each result                                            |

## How It Works

<Steps>
  <Step title="Write a search query">
    Compose the query using **Mentions** to make it dynamic per record (e.g., `{{Company Name}} official website`).
  </Step>

  <Step title="Choose search type">
    Select **Auto** for best quality, **Keyword** for exact term matching, or **Neural** for semantic search.
  </Step>

  <Step title="Configure filters and output">
    Optionally filter by domain or date, choose how many results to return, and name the output field.
  </Step>
</Steps>

## Output

A new field containing the search results is added to each record. The field becomes available as a **Mention** in all downstream nodes.

## Best Practices

* Use **Mentions** to make queries dynamic per record
* Set domain filters to focus results (e.g., only `linkedin.com` for profile lookups)
* Start with a small number of results per query and increase if needed
* Use **Neural** search for conceptual queries, **Keyword** for exact terms

## Related Nodes

* **[AI Enrichment](/guide/nodes/enrichment/ai-enrichment-node)** - enrich records using LLM-generated content instead of web data
* **[HTTP Request](/guide/nodes/actions/http-request-node)** - call a specific API endpoint rather than performing a web search