# Introduction

> How to deploy Centinel Analytica on your site and customize protection settings.

Source: https://docs.centinelanalytica.com/

## Before you begin

* Create your account in the [Centinel Analytica dashboard](https://dash.centinelanalytica.com).
* You'll need:
  * your **site key** (public, used by the browser script)
  * your **secret key** (server-only, used for `/validate`)

## How it works

Centinel has three parts:

* **Server validation** (required): your backend, or a platform integration, calls the validation API for each protected request and acts on the decision (`allow`, `block`, or `redirect`). The validator issues the `_centinel` session cookie, and your response sets it.
* **Policy rules**: rules select bot detection and actions for matching requests. The dashboard's default rule enforces nothing, because the validator skips it. Add and verify an explicit matching rule before you rely on a policy action.
* **Browser script** (optional): collects browser and network signals that server-side data alone cannot see, and mirrors the `_centinel` cookie. Detection works without it, with a smaller signal set.

## Next steps

1. Copy your keys from the dashboard. Review the default policy. Add and order path-specific rules where you need different behavior.

   Go to: [Dashboard](https://docs.centinelanalytica.com/install/dashboard.md) and [Policy Rules](https://docs.centinelanalytica.com/admin/policy.md)

2. Wire up validation: use a platform integration, or call `/validate` from your backend.

   Go to: [Platform integrations](https://docs.centinelanalytica.com/platforms.md) or [Validation](https://docs.centinelanalytica.com/install/validation.md)

3. Add the browser script for stronger detection on pages visitors load.

   Go to: [Scripts](https://docs.centinelanalytica.com/install/scripts.md)

New to the decisions and block reasons in the dashboard? The [Glossary](https://docs.centinelanalytica.com/concepts/glossary.md)
defines them.

## Getting help

Contact support with:

* the URL you're protecting
* a sample `/validate` request payload (redact secrets)
* the decision you received and any `crawler` metadata (if enabled)
