Lead Data Analyst · Global Sales Strategy & Operations · Gurgaon
Numbers are the easy part. Knowing which number the business should act on is the job.
I build the reporting layer that commercial teams actually run on: SQL warehouses, tested models, and reports that survive contact with a leadership review. My day job is forecasting and funnel performance for a global marketing organisation - which mostly means telling people the number is down before they find out themselves. 📉
The projects below are self-directed builds on synthetic data. Each one takes messy source extracts through a layered SQL warehouse into a tested reporting model and a finished report pack. They demonstrate method, not results - the findings are planted in the generators and recovered by the analysis, which is rather the point.
| Project | What it demonstrates |
|---|---|
| marketing-funnel-performance-report | 14-page demand-gen report: MQL to SAL to DSO to $NB, three years, like-for-like YoY on a partial year |
| manufacturing-network-analytics | OEE decomposed to availability, performance and quality; downtime Pareto; supplier traceability |
| revops-funnel-analytics | Cohorted funnel conversion, channel economics, pipeline coverage, new-business growth |
| people-analytics | Headcount flows, attrition, tenure survival handled for right-censoring, hiring funnel |
| lead-enrichment-quality | Vendor economics on cost per usable record, identity resolution, field completeness |
Bad source rows are quarantined and counted, never silently dropped, and the exception counts are published on the same page as the KPIs. A metric nobody can audit is a metric nobody trusts.
- Conversion should be cohorted on creation date, not stage date. Otherwise you are celebrating a mix shift.
- Report the median cycle time. The mean is wrong in the direction of optimism, always.
- You cannot average a ratio across rows of different size and call it an average.
- "The rate grew 8%" means nothing. "+5 pts" means something.
- Year to date against a full prior year is how a growth number quietly lies to a board.
SQL · Python · Power BI · DuckDB · Databricks · Snowflake · Azure · Excel
Open to data and analytics roles across Europe and Ireland. LinkedIn