Bulk exports

Crunchbase provides downloadable data exports for customers who want full collection snapshots outside of the REST API.

Exports are available in two formats:

• NDJSON (Recommended) – preserves nested and complex JSON structures
• CSV (Legacy Flat Format) – flattened tabular structure for spreadsheet-based workflows

Both formats are delivered as compressed .zip files and are available per collection.

Updated files are generated approximately once daily for endpoint collections such as companies, people, funding rounds, acquisitions, and IPOs. There is no guaranteed refresh time.

⭐ NDJSON Export (Recommended)

NDJSON (Newline Delimited JSON) preserves nested and complex JSON structures without flattening relationships or losing data fidelity.

Each export is delivered as a compressed .zip file containing a .ndjson file. Each line in the file represents one valid JSON object (one record per line).

Why use NDJSON?

• Supports nested objects and arrays
• No artificial flattening of relationships
• Better for data lakes and warehouses
• Ideal for Spark, Snowflake, BigQuery, Databricks, etc.
• Future-proof schema evolution

NDJSON URL Format

https://api.crunchbase.com/v4/data/static_exports/<collection>.ndjson.zip?user_key=<user_key>

Accessing the CSV Export (Flat Format)

CSV exports are available for customers who prefer a flat, tabular structure.

These files are:

• Flattened representations of the API data
• Delivered as compressed .zip files
• Structured for compatibility with Excel and traditional BI tools

CSV URL Format

https://api.crunchbase.com/v4/data/static_exports/<collection>.zip?user_key=<user_key>

Accessing the CSV/NDJSON endpoints

Locate your User Key. The user key is a 32 character string that you should have received after signing up for Crunchbase data. If you have not received your user key, please contact support.

Check an export before downloading

Use an HTTP HEAD request to confirm that an export is available and inspect its metadata without downloading the archive. HEAD requests are free and do not count toward usage. The following example checks the Organizations CSV export and supplies the user key in a request header:

printf "Crunchbase user key: "
read -s CRUNCHBASE_USER_KEY
printf "\n"
curl --head --show-error \
  --header "X-cb-user-key: ${CRUNCHBASE_USER_KEY}" \
  "https://api.crunchbase.com/v4/data/static_exports/organizations.zip"
unset CRUNCHBASE_USER_KEY

A successful request returns HTTP 200 OK and headers such as:

• Content-Type: should be application/zip
• Content-Length: archive size in bytes
• Last-Modified: when the current export was generated, expressed in GMT (UTC)
• ETag: an identifier for the current version of this file; save the complete value, including its quotation marks, to check for changes
• x-amz-meta-filename: the generated archive filename
• Accept-Ranges: whether the server supports partial downloads

If the response is 401 Unauthorized, confirm that the user key is present and valid. A HEAD request checks availability and metadata only; it does not download or validate the contents of the ZIP file.

Check whether a downloaded file has changed

After successfully downloading a file, save the ETag from that download's response headers. To check whether your local copy is still current, send a HEAD request to the same URL with If-None-Match set to the saved ETag. Replace the example value below with the complete ETag, including its double quotes:

printf "Crunchbase user key: "
read -s CRUNCHBASE_USER_KEY
printf "\n"
curl --head --show-error \
  --header "X-cb-user-key: ${CRUNCHBASE_USER_KEY}" \
  --header 'If-None-Match: "saved-etag-value"' \
  "https://api.crunchbase.com/v4/data/static_exports/organizations.zip"
unset CRUNCHBASE_USER_KEY

• 304 Not Modified: the file has not changed. Keep your existing complete local copy.
• 200 OK: the file has changed. Download that file again and save the new download's ETag. The HEAD request itself does not download the file.
• Other responses: resolve the error before treating the local file as current.

You can also send If-None-Match with a GET request: an unchanged file returns 304 without an archive body; a changed file returns 200 with the archive. Keep your existing file when the response is 304, and replace it only after a 200 download completes successfully. An unchanged ETag does not mean an interrupted local download is complete.

Check a set of files across a refresh

Files refresh approximately once daily, without a guaranteed refresh time. Files within the same format are usually updated in tandem, but separate downloads spanning a refresh can capture different export generations.

1. Use one format throughout: CSV or, preferably, NDJSON. The two formats can refresh at different times.
2. Check Last-Modified across the required files to compare refresh timing. Timestamps may differ slightly and do not need to be identical.
3. Download the files and save each successful download's ETag and Last-Modified response headers.
4. After the set finishes downloading, recheck every file with a conditional HEAD request using its saved ETag. Download only files that return 200, then recheck the set until all return 304.

ETags identify individual file versions, not a shared version across collections. These checks detect file changes during downloading; they do not establish that every collection represents the same underlying database snapshot.

Download an export

Once you have identified a file to download, choose its URL from the tables below. Use NDJSON (recommended) or CSV consistently across your set of files.

To download in your browser, append your user key after user_key= in the selected URL, then open it. The browser downloads the ZIP archive. For programmatic downloads, send a GET request to the same endpoint and supply your key in the X-cb-user-key header, as in the examples above.

Download URLs (NDJSON ⭐ Recommended)

Download URLs (CSV Option)

Unzip the file. You will have downloaded a .zip file containing the NDJSON/CSV you've requested. Unzip the file to begin using the file.

Using the Data Exports

We encourage you to leverage the Data Export for your internal business and research needs. You are also welcome to publish analysis and aggregate statistics derived from the Crunchbase dataset. Unless otherwise noted in your license or Data Access Terms, you may not license, sublicense, sell, offer to sell, distribute or otherwise provide any of the Crunchbase data to any third parties.

Attribution Requirements

When sharing information about the data, follow these attribution requirements.

• Attribution must be accompanied by a hyperlink to Crunchbase
• If content being attributed is primarily about one entity, link must point to that entity on Crunchbase.
• Link must be plainly visible to end user
• Link must be in close proximity to the attributed data
• Link must be visible to Internet spiders (e.g. Googlebot)
• Link cannot include the “nofollow” tag"