Skip to content

Analyse

Upsell & Cross-sell Product Mix Strategy

Analyzed all Benelux store receipts using Market Basket Analysis, product clustering, correlation heatmaps and Python predictive modeling to drive upsell/cross-sell and design a launch plan for a new product bundle.

PythonMarket Basket AnalysisClusteringRetailUpsell & Cross-sellProduct Strategy

Presentation

Brand affinity & K-Beauty deep dive

Brand affinity & K-Beauty deep dive — 1/10Brand affinity & K-Beauty deep dive — 2/10Brand affinity & K-Beauty deep dive — 10/10

01 / 10

Interactive demo

Explore the brand affinities

The lift matrix from our Market Basket Analysis, made explorable. Hover the heatmap, build a basket to see which brands to cross-sell next, or open the K-Beauty focus to see which Korean brands to bring in first.

Simulated in your browser · illustrative data

Loading the demo…

The challenge

Carried out during a three-week Business Deep Dive with Sephora, this project leveraged transaction-level data across the Benelux region to uncover purchasing patterns and build a data-driven upsell and cross-sell strategy.

The question

Which product associations in Benelux store receipts offer the highest upsell and cross-sell potential, and how should Sephora structure the launch of a new combined product line?

What I did

Performed Market Basket Analysis and association-rule mining in Python, built correlation matrices and heatmaps, and applied product clustering to isolate high-affinity product groups. Developed a Python predictive model to score upsell and cross-sell opportunities and translated the findings into a go-to-market launch plan.

Backstage · Data sources

Complete transaction dataset covering all Sephora store receipts across the Benelux region (baskets, SKUs, quantities, categories and pricing).

The result

Identified the highest-converting product combinations across Benelux and delivered a complete launch plan for a new cross-product bundle designed to lift average basket value through targeted upsell and cross-sell.