She wasn’t being replaced. But the repetitive part of her job — pulling data, slicing dashboards, writing SQL to answer the same "what’s driving this?" question — was.
Feeld is a fast-growing dating platform operating across multiple geographies, platforms, and user segments. As the business scaled, leadership needed a faster and more rigorous way to understand what was driving performance across the full user lifecycle — from acquisition to engagement to monetization.
The team had already bought into metric trees as a framework, designing them in Miro and implementing DoubleLoop as a canvas-style interface. It was a strong foundation — but it stopped at visualization. The trees showed what the business looked like; they didn’t tell you what had changed or why.
Answering "Why did ARPU drop last week?" or "Is the lift in active users coming from new, retained, or reactivated cohorts?" still meant hours of manual SQL. That ceiling showed up everywhere: executive reviews leaned on high-level narratives rather than rigorous decomposition, and segment-level questions often didn’t get asked at all — because the cost of answering them was too high.
Trace kept the metric tree paradigm Feeld already believed in and added the analytical engine underneath — attribution calculations, segment-aware traversal, anomaly detection. The business is now represented as three connected trees.
01 Engagement
02 Active users
03 Monetization
Applies to all 3 trees
Because Feeld had already adopted the metric tree framework, onboarding moved fast. Trace connected directly to Feeld’s Snowflake warehouse, reused existing metric definitions, and slotted into the team’s workflow across Finance, Data, and Growth.
Deep dives that took hours of SQL and dashboard archaeology now take minutes. More importantly, lowering the cost of investigation changed the team’s behavior: when asking another question became cheap, they started asking more of them
→ A single connected model linking growth → engagement → actives → monetization, rather than fragmented dashboards by function.
→ Executive visibility into revenue drivers that holds up to scrutiny.
→ Analysts spending more time testing hypotheses and shaping recommendations.
The 5-minute deep dive was only the first step. Trace’s AI analyst now runs on top of Feeld’s metric trees — proactively surfacing material shifts, answering “why did X move?” in plain language, and generating executive-ready narratives without requiring someone to drive every investigation.
The metric tree provides the business context. The AI analyst does the analysis.