> For the complete documentation index, see [llms.txt](https://docs.sumble.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.sumble.com/system-setup-and-configuration/data-destinations/blob-store.md).

# Blob store (GCS / S3)

Sumble can exchange data with you as files in object storage — Google Cloud Storage or Amazon S3. This is the simplest integration option: no API connections, no database credentials. You send us your account list as a CSV, and we write enriched data back as Parquet files that you can process however you want.

Both directions are optional and independent. Use one, the other, or both.

## What we deliver

Each pipeline run writes a set of Parquet files to your bucket:

* **Enriched organizations** — your accounts with all configured enrichment columns
* **Alerts feed** — intent signals generated for your accounts
* Additional tables as configured for your account

Files are standard Apache Parquet, readable by any tool that supports it (BigQuery, Athena, Snowflake, Spark, pandas, DuckDB, ClickHouse, etc.). Each run replaces the previous set of files.

## What you send us

To have Sumble enrich your accounts, put a CSV in the bucket at a fixed path and refresh it on your own schedule. We read the same path on every run.

* One row per account, with a header row
* Include the [account fields](https://docs.sumble.com/system-setup-and-configuration/data-fields.md#accounts) Sumble needs. Matching quality depends most on the website
* Either plain `.csv` or gzipped `.csv.gz`; we detect which
* Any extra columns you include are preserved and available for filtering

You can send a [people list](https://docs.sumble.com/system-setup-and-configuration/data-fields.md#people), or a [users list](https://docs.sumble.com/system-setup-and-configuration/data-fields.md#users), the same way.

## Update cadence

Files are updated daily, after the Sumble data pipeline completes.

## Setup

The blob store integration requires an enterprise plan.

### Google Cloud Storage

1. Create a GCS bucket (or designate an existing one).
2. Grant Sumble's service account access to it — we'll give you the address.
3. Tell us the bucket name, and the path of your input CSV if you're sending one.

### Amazon S3

1. Create an S3 bucket (or designate an existing one).
2. Grant Sumble's IAM principal access to it with a bucket policy — we'll give you the ARN to use. Read access is enough if you only send us data; add write access for us to deliver enriched files.
3. Tell us the bucket name, its region, and the `s3://` URI of your input CSV if you're sending one.

If your security policy requires cross-account role assumption rather than a bucket policy, create an IAM role that trusts our principal, and give us the role ARN (and an external ID if you use one) instead.

### Amazon S3, Sumble-hosted

If you would rather not create a bucket, we deliver to one of ours: `s3://sumble-customer-data-shares/<your prefix>/` in `us-east-2`.

1. Tell us the AWS principal that will read the files: the ARN of the IAM role your warehouse or loader uses, or your account id. We grant it `s3:ListBucket` and `s3:GetObject` on your prefix only. Nothing else in the bucket is visible to it.
2. If you send us an account list, also tell us the principal that uploads it. We grant that one `s3:PutObject` under `<your prefix>/input/`. It can be the same principal as the reader.
3. Read with any S3 client as the reader principal; cross-account access needs no bucket of your own. Snowflake, Athena and Databricks external stages work the same way.
4. Upload your account list to `s3://sumble-customer-data-shares/<your prefix>/input/accounts.csv` and tell us the path. Overwrite the same key when it changes.

Files we deliver expire after 30 days, so load each delivery within that window. Files you upload under `input/` are kept until you replace them.

<a href="https://sumble.com/book-a-demo" class="button primary">Book a time to chat with us</a>
