A shared data catalogue lets executives ask questions in plain language and gives business teams the definitions they need to build their own dashboards.
- Analyst wait eliminated
- 4-7 days
- To executive data visibility
- ~1 month
- To build the full data catalogue
- ~2 weeks
The challenge
Every new analysis request at Licious meant four to seven days of back and forth with an analyst. Executives needed a way to answer questions and create dashboards themselves.
What we built
The work started with a small set of trusted tables, then expanded across the wider data estate once the approach had proved itself.
Parallel Loop connected the data and standardised the definitions behind each answer, so teams could ask questions using the same business language.
- Connected seven Tableau gold tables for natural-language queries and dashboard creation.
- Built a full catalogue across the bronze, silver and gold tables in Licious's Databricks data lake on Azure, alongside AppsFlyer and Trino.
- Defined metrics with their units, synonyms, governance owners and underlying queries. AI drafted the definitions and people reviewed them to standardise downstream queries.
- Added a request flow in chat for missing metrics. Users can nominate a metric for an analyst to review and add to the catalogue.
The impact
Executives gained data visibility in about a month, with the full catalogue built in roughly two weeks. The four-to-seven-day analyst turnaround was eliminated, and business teams across Licious now create their own dashboards.
- Executives ask the data directly instead of waiting for an analyst.
- Teams across functions build dashboards independently.
- Once the catalogue stabilised, category teams began gross margin analysis, customer quality feedback analysis and price parity monitoring.