End-of-season pricing, calculated rather than guessed
Senior Data Analyst · since Jan. 2024
The context
At end of season, every markdown was decided by hand: thousands of references reviewed one by one, gut-feel choices, hours of data entry, and a margin quietly slipping away.
My intervention
I built a dynamic discount engine (Hex + Python): from stock, sales and product attributes, it computes the optimal discount rate for each reference. The business adjusts its strategy with a few sliders, re-runs the calculation and exports a ready-to-load file. Upstream, 80% of the SQL/API flows are automated (Airflow) for reliable, up-to-date data.
The result
No more manual entry or random choices: thousands of markdowns computed in seconds, consistent and traceable. The result: +10% margin vs. the previous season, a business team autonomous on the tool, and end-of-season sales finally steered on data rather than intuition.