Shipping a DeFi Funnel Pipeline in 90 Days

I was Growth Lead at Kyber Network from January to April 2026. The DeFi aggregator handles $200B+ in cumulative trading volume across 18 EVM-compatible chains. I joined to own growth for the core aggregator. The role scope shifted materially in Q1, so I shipped what I could and exited cleanly. This is what I built.

The Problem

Kyber had partial analytics coverage but no unified view of the user funnel. Acquisition, wallet connection, first trade, and retention all lived in separate tools. Nobody could answer the core question: where exactly were users dropping between landing page and first trade?

What I Built

A funnel analytics pipeline in Bun/TypeScript tracking 14 data sources across 6 conversion stages. The pipeline joined on-chain signals (wallet connects, token approvals, swap executions) with off-chain signals (page views, button clicks, campaign UTMs) into one view.

The 6 stages:

1. Landing (UTM-tagged traffic)
2. Wallet connection
3. Token selection
4. Quote preview
5. Swap approval
6. First completed trade

Each stage measured drop-off rate, time-to-next-stage, and source attribution.

What I Found

40% drop-off between wallet connection and first trade. Users connected their wallet but never completed a swap. The bottleneck wasn’t awareness or acquisition — it was the gap between “I’m interested” and “I’ve done the thing.”

This became the team’s top onboarding priority. The pipeline gave the product team a clear, quantified target: close the wallet-to-trade gap.

The AI Automation Layer

I also automated the GTM reporting workflow using AI agents. Analytics query generation, campaign reporting, and competitive analysis that previously took hours per cycle ran in minutes. Not a novelty — a practical time multiplier for a small growth team.

The Exit

Three months. Production analytics system shipped, top-priority growth lever identified, clean handoff. The tenure was short; the output wasn’t.