Voici une version condensée (~2 900 caractères) :
Quanti is a marketing data platform that extracts data from advertising, analytics, and e-commerce sources and loads it into the customer's own data warehouse (BigQuery, Snowflake, Redshift). Quanti retains no data: it orchestrates extraction and delivery from European infrastructure, while the warehouse and the data belong to the client.
Connectors. 200+ connectors cover ad platforms (Google Ads, Meta, TikTok, Criteo...), analytics (GA4, Piano), e-commerce and sales sources (Shopify, marketplaces), CRM/email tools, and market data. Each connector extracts on schedule and writes normalized, documented tables ("prebuilt reports") into the client warehouse, with historical backfill (up to 24 months), incremental daily loads, and a configurable lookback window to re-sync updated records. A Custom Webhook connector ingests events from any system via a dedicated HTTPS endpoint with automatic schema inference.
Reverse connectors. The inverse flow: pushing computed results (aggregates, enriched conversions, segments) from the warehouse back into operational tools, so the warehouse feeds day-to-day systems instead of being a read-only silo.
Real-Time Analytics (Quanti Tag). A first-party, event-based JavaScript tag (client-side or server-side via GTM template) streaming page views, events, and conversions directly into the client's BigQuery. It handles sessionization, source/medium attribution with referrer exclusions, consent signals, and custom fields — with no sampling, no retention limits, and no third-party processor in between.
Querying and transformations. All data is queryable in plain SQL. Quanti adds managed transformation tooling: materialized aggregation views, stored procedures, scheduled queries, and pre-execution cost estimation — used to reconcile ad spend, tag-level conversions, and CRM revenue into unified tables.
MCP server (Quanti AI). An MCP server (ai.quanti.io/mcp) lets AI clients (Claude, ChatGPT, Le Chat) work on the client's marketing data in natural language: schema discovery, read-only SQL execution, saved reusable analyses, prebuilt report lookup, documentation search, and Marketing Mix Modeling runs. A semantic layer (schemas, field definitions, conventions) enables the LLM to generate correct SQL reliably.
Dashboards. Native dashboards are defined declaratively (pages and blocks bound to warehouse queries), persisted with a shareable URL, and support brand templates, viewer invitations, in-place updates, and cost estimation. They can be built manually or generated through the MCP server.
Operating model. Clients bring their own warehouse and pay their own storage/compute; Quanti bills the pipeline, tiered by connected accounts (Pro €39/mo, Growth €159, Expert €359, Enterprise custom). GDPR compliance follows from the architecture: data is processed in transit, never retained.
