Analytics Warehouse

The Analytics Warehouse is the new generation of Treble’s data platform. It gives you direct SQL access to your operational data with near-real-time freshness, full history, and several datasets that were never available before — including the complete message trace of every conversation and the node-by-node journey of every flow.

The Analytics Warehouse replaces the previous Data Warehouse (client_analytics), which is now deprecated. Both remain available during the migration period. If you are migrating existing queries, start with the Migration Guide.

What’s new

	Previous warehouse	Analytics Warehouse
Freshness	Up to 3 hours	~15 minutes
History	Last 3 months	Full history
Conversation messages	Not available	Full trace, message by message
Flow journey	Not available	Step by step, AI steps included
Session variables	Not available	Available
Ads attribution	Not available	Click-to-WhatsApp ad → session
Incremental sync	Not supported	`synced_at` watermark on every row

Connection

Once access is enabled, you will receive credentials to connect:

Parameter	Value
Host	Provided by Treble
Port	`8443` (HTTPS) / `9440` (Native TLS)
Database	`treble_client_analytics`
User	Provided by Treble
Password	Provided by Treble

You can use any ClickHouse-compatible client: DBeaver, DataGrip, clickhouse-client, or connect directly from Metabase, Looker, Power BI, Google Sheets, or another BI tool.

Access is available for companies that request onboarding. Contact your Account Manager or support to start the process.

The data contract

Three promises define how the warehouse behaves. Understanding them will save you time:

Data is near real time. New sessions, messages, and campaign sends appear within minutes of happening.
Today is provisional; history is final. Operational data keeps changing while it is in flight — a campaign send that is “in process” at 10:00 may be “delivered” at 11:00, in Treble and in your warehouse alike. Figures for any given day consolidate fully within 48 hours, and the warehouse self-corrects continuously (every 30 minutes) during that window. Historical data beyond that window is immutable.
Every row tells you when it landed. All tables expose a synced_at column: the moment the row (or its latest correction) was written. This powers reliable incremental synchronization into your own systems.

Naming glossary

The warehouse uses one consistent vocabulary across every table:

Term	Meaning
`poll_id` / `poll_name`	A conversational flow (what you build in the Treble flow editor).
`session_id`	One execution of a flow by one user — a conversation.
`direction`	`INBOUND` if the user started the conversation, `OUTBOUND` if a campaign did.
`agent_conversation_id`	A conversation handled in the agent workspace (Sales).
`treble_id`	The contact’s phone identity: country code + cellphone.
`synced_at`	When this row was last written or corrected in the warehouse.

Security and limits

Row-level security: your credentials can only ever see your company’s rows — enforced by the database on every table, on every query.
Read-only: your user cannot modify, create, or delete anything.
Protection limits: queries have a 30-second execution cap and row/memory limits that keep the platform stable for everyone. See Query Optimization for how to stay well inside them.

Available tables

Conversation Flows

Everything that happens inside your automated flows.

Table	Grain	What it holds
`fact_treble_sessions`	1 row per conversation flow	Every flow execution, inbound and outbound
`fact_treble_session_messages`	1 row per message	The full conversation trace
`fact_treble_session_nodes`	1 row per interactive step	The journey through the flow — questions, AI steps, WhatsApp Flows
`fact_treble_session_variables`	1 row per variable	Variables captured during each flow

Campaigns

Outbound sends and how users engage with them.

Table	Grain	What it holds
`fact_campaign_sends`	1 row per send	Every outbound campaign message, full delivery detail
`fact_campaign_daily`	1 row per campaign per day	Daily delivery rollup
`fact_hsm_responses`	1 row per response	Answers given to HSM templates

Agent Workspace

Human-agent operations: conversations, messages, transfers, availability, productivity.

Table	Grain	What it holds
`fact_agent_conversations`	1 row per agent conversation	Conversations handled in the agent workspace
`fact_agent_conversation_messages`	1 row per message	Messages within agent conversations
`fact_agent_conversation_transfers`	1 row per transfer	Transfers between agents
`fact_agent_status_changes`	1 row per change	Agent availability history
`fact_agent_daily`	1 row per agent per day	Daily agent productivity

Conversions & Attribution

Where conversations come from and what they achieve.

Table	Grain	What it holds
`fact_target_events`	1 row per event	Conversion/target events fired in flows
`fact_ad_sessions`	1 row per ad-attributed session	Click-to-WhatsApp ad attribution
`fact_whatsapp_link_events`	1 row per event	WhatsApp link clicks and conversation starts

Dimensions — catalogs for enrichment

Agents, tags, teams, HSM templates, agent-tag assignments, polls, WhatsApp channels, and flow nodes (including AI agent configuration). See Dimensions.

Get Started

Meta

Build with Treble

Main Platform

Agent Platform

Integrations