How it works?

Blotout Cloud follows a clear path from infrastructure to insight: deploy → collect → process → analyze → activate → ask with AI. This page walks through each stage.

1. Deploy single-tenant infrastructure

Blotout Cloud is provisioned with Terraform into a private AWS VPC. You choose one path:

Path	What happens
Bring Your Own Cloud	Resources are created in your AWS account using credentials you supply to Terraform Cloud.
Blotout Hosted	Blotout creates and manages a dedicated AWS account under its organization for you.

Typical outputs after deploy:

• Web application URL — https://<org>-ui-<env>.blotout.io
• Event collector (SDK) URL — https://<org>-<env>.blotout.io

Inside the VPC you get EKS (web app, Airflow, Airbyte, services), RDS for metadata, and S3 buckets for landing, processed, staging, outbound, and logs. See Architecture for diagrams and Installation for the deploy steps.

2. Organize work with Tags and Applications

Everything in the product is scoped by a Tag — an EdgeTag environment, usually a customer domain or brand. Under each tag you create Applications (web, Android, and so on). Each application receives:

• App ID
• API token
• EdgeTag URL

Teams select Tag (and Team) in the web app top bar; Applications, Pipelines, Audiences, and Analytics filter to that context. See Tags and Application.

3. Collect first-party events via EdgeTag

1. Sign in to Blotout Cloud and complete User Onboarding.
2. Create or select a Tag and Application.
3. In EdgeTag, add a provider with the SDK URL (event collector) and app-id from Cloud.

The SDK on your domain captures clickstream and commerce events. Events hit the collector and land in the landing S3 bucket as Parquet, partitioned by event_date — without embedding long-lived secrets in browser code.

4. Process into a semantic lakehouse

Airflow DAGs (for example events_incremental, id_stitching_incremental) flatten, deduplicate, and enrich raw events into the processed bucket and Iceberg tables in Amazon Athena.

dbt models build the silver/gold semantic layer used everywhere downstream:

• view_core_events
• view_users
• view_id_graph
• unified_events

Schema details live in Data Schema; bucket and naming rules in Data Storage Standards.

5. Connect marketing and commerce pipelines

From Platform → Data Pipelines, create Airbyte connectors under a Tag (Shopify, Facebook Marketing, Google Ads, TikTok, Recharge, Amazon, and more). Data lands in S3, organization DAGs process it, and Athena schemas (typically <SOURCE>_<ENV>) become available for dashboards and MCP.

See Data Pipelines.

6. Analyze in the web app

Area	What you get
Analytics	Pre-built dashboards for Shopify, Facebook, TikTok, clickstream, and attribution
Data stream	Live Event Stream explorer for debugging and inspection
MCP Insights	MCP session activity, errors, and Athena query cost

All of these read from the same lake and semantic views.

7. Build audiences and activate

1. Define segments from unified events (templates such as Suppression or Welcome Back, plus custom rules).
2. Schedule sync via Airflow.
3. Activate audiences to Meta, Klaviyo, TikTok, and other EdgeTag-connected channels using the outbound path.

Collection and activation share the same EdgeTag-connected world — measure and act on the same identities.

8. Query with the Cloud MCP Server

Install the MCP server in an AI client, log in with OAuth via EdgeTag, then ask natural-language questions. The server:

1. Selects a tool (@Tool on Athena, ads, commerce, dashboards, email)
2. Builds SQL via report generators when needed
3. Runs queries in a dedicated Athena workgroup (cost isolation)
4. Returns results — optionally as HTML dashboards emailed to your team

Prebuilt report types are listed under Reports Available Out of the Box; stack details under MCP Architecture.

End-to-end flow

Deploy (VPC / Terraform)
  → Tag + Application
  → EdgeTag provider (SDK URL + app-id)
  → Events → S3 landing
  → Airflow + dbt → Athena semantic layer
  → Pipelines (Airbyte) into the lake
  → Analytics · Data stream · Segments
  → Activation (Meta / Klaviyo / TikTok / …)
  → MCP natural-language queries on the same lake

Next

Start with Onboarding, or dig deeper into Architecture.