---
title: "Data storage and privacy"
---

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

# Data storage and privacy

Evidential is built with **privacy by design** — it stores **only what’s essential** to make experiments work, mostly aggregated, while raw data always stays in your own environment.

What we store:

## Aggregated Data

We mostly store only **anonymized and aggregated** information in a **PostgreSQL application database**, such as:

- **Experiment specifications** — name, hypothesis, start/end dates, configuration details
- **Experiment results** — regular snapshots of aggregated experiment metrics

## Granular Data

There are two types of user-led actions that can expose granular data containing PII:

- **Participant IDs** — used to track assignment and outcomes.
- **Unique values of columns used as filters** — stored to enable autocomplete. Avoid using columns with PII (e.g., names, emails) as filters.

:::caution[Keep personal data out of participant IDs and filter columns]
Evidential stores the participant IDs used in experiments and the unique values of columns used as filters. Make sure
IDs contain no raw personal information, or hash them before use, and don't filter on columns with PII such as names
or emails.
:::

## What We Don’t Store

Evidential **doesn't store or copy raw-level data**. All analysis queries run **directly in your database**.

## Learn More

Read our full [Privacy Policy](https://docs.evidential.dev/legal/privacy/) for details on data handling and protection.

Source: https://docs.evidential.dev/concepts/data-storage-and-privacy/index.mdx
