The Problem
- Business: 7 disconnected data tools created duplicate licenses ($1.2M/year), swivel-chair workflows, and inconsistent reporting across departments.
- Users: Analysts exported CSVs from 3+ systems and stitched them in Excel; 68% of data requests took 24+ hours to fulfill.
- Constraints: Each tool had embedded business logic; migration had to preserve historical data fidelity and comply with SOC 2 Type II.
The Solution
- Strategy: Unified query layer with role-based access; phased migration by department to de-risk the cutover.
- Design: Single-pane search, federated query builder, saved views with sharing, and lineage visualization for trust.
- Validation: 22 contextual inquiry sessions; 6-week pilot with 180 analysts; error-rate monitoring post-launch.
Executive Summary
Enterprise analysts across sales, finance, and operations were spending hours daily copying data between 7 disconnected tools — Snowflake, Tableau, internal REST APIs, Excel add-ins, and legacy Oracle reports. The result was stale data, conflicting metrics, and a $1.2M annual tool sprawl.
I led the product design for a unified data platform that exposed a single search-and-query interface on top of federated sources. The platform included role-based access, data lineage visualization, and collaborative saved views — turning 24-hour requests into self-service queries.
Business Problem
- $1.2M/year in overlapping tool licenses across 7 disconnected data products
- 68% of ad-hoc data requests took 24+ hours due to cross-tool stitching
- 12% of executive dashboards showed conflicting KPIs because source definitions diverged
- Compliance audits revealed 4 tools with incomplete data lineage documentation
- Sales and finance teams maintained parallel spreadsheets — creating a shadow data culture
User Research
Conducted 22 contextual inquiry sessions across sales ops, financial planning, and IT operations. Shadowed analysts for 2-hour blocks to observe real querying, exporting, and stitching workflows.
Insight 01
73% of analyst time was spent locating the right tool and exporting data — not analyzing it.
Insight 02
Power users built personal bookmark libraries and shared them via Slack — an organic signal of unmet need.
Insight 03
Executives didn't trust self-serve numbers because they couldn't trace lineage or verify freshness.
Insight 04
Compliance officers needed automated lineage trails; manual documentation was always 2–3 releases behind.
Discovery Insights
- Users didn't want "one more dashboard" — they wanted one place to ask questions and get trusted answers
- Data lineage visualization was the trust signal executives needed to replace spreadsheet shadow systems
- Role-based query templates would cover 80% of routine requests without writing SQL
- Shared views reduced duplicate work — the same report was rebuilt independently by 3 teams
Product Strategy
Partnered with data engineering and product leadership on a phased migration:
- Phase 1: Unified search and query interface on top of existing warehouses — no data migration
- Phase 2: Role-based templates, saved views, and sharing to replace recurring report requests
- Phase 3: Data lineage visualization and automated compliance documentation
Defined success metrics: query-to-answer time, daily active users, integration error rate, and compliance audit findings.
Design Exploration
Explored 3 concept directions for the unified interface: a traditional BI dashboard approach, a Slack-like conversational query, and a single-pane search with federated results. Testing favored the search-first approach — it matched mental models from consumer tools and required minimal training.
Designed lineage visualization as an expandable trace showing source system, transformation logic, and last refresh — directly addressing the trust barrier that kept executives in spreadsheets.
Validation
- 22 contextual inquiry sessions with sales, finance, and IT analysts
- 6-week pilot with 180 analysts — tracked query time, error rate, and NPS
- Usability testing on search and lineage flows with 14 participants
- Executive stakeholder review with live demo using real production data
Final Solution
Delivered a unified data platform with natural-language search, federated query builder, role-based templates, collaborative saved views, and automated data lineage — all backed by existing warehouse infrastructure to minimize migration risk.