Convea AI

Introduction

Convea centralizes eCommerce data (Shopify, Klaviyo, Google Ads, Meta Ads, and more), normalizes it in ClickHouse, and models it with DBT to expose actionable insights powered by AI. The goal: move from manual analysis to answers in <3 s with p99 < 200 ms across 100M+ daily events and 200+ integrations—without relying on SQL or analysts.

The Challenge

Heterogeneous schemas, API limits, and rising volumes required balancing freshness and cost. We also had to standardize definitions across sources (attribution, campaigns, audiences) and make every metric traceable for non-technical users—without losing analytical rigor.

Solution & Approach

We designed a batch-first platform with a semantic layer: unified ingestion, incremental models in DBT and materializations for hot paths; canonical, domain-based metrics; and a Next.js frontend that applies AI to generate personalized insights and natural language queries.

Ingestion & Data Modeling

• Connectors for Shopify, Klaviyo, Google Ads, Meta Ads, and more
• ClickHouse for fast, large-scale queries
• Incremental DBT models and materializations to reduce latency and cost
• Semantic layer unifying and standardizing key metrics
• Quality controls and traceability by metric and domain

AI Analytics

• Interpretation of data with business and campaign context
• Automatic generation of insights and personalized recommendations
• Natural language queries and on-demand reporting
• Contextualized visualization in the dashboard for faster decisions

Results & Impact

Convea cuts analysis time from hours to seconds and standardizes performance interpretation across channels. It delivers insights in <3 s with p99 < 200 ms, on top of 100M+ daily events and 200+ integrations—accelerating decision‑making without increasing operating costs.

Platform Achievements

• p99 < 200 ms on key views and queries
• 200+ active integrations across eCommerce and Ads platforms
• Insights in <3 s over 100M+ events/day
• Shared metric standards across Marketing and Growth

Technical Contribution

• Incremental modeling in DBT and materializations for hot paths
• ClickHouse optimization (partitioning, sorting, indexes) for high throughput
• Domain-based semantic layer (campaigns, audiences, attribution)
• Pipeline orchestration and observability with SLO-driven alerting
• AI modules enabling actionable insights and NLQ

Convea is evolving into the trusted intelligence layer for eCommerce: it turns scattered data into operational knowledge, shortens time-to-insight, and standardizes how Marketing and Growth make decisions at scale.