Problem
Massmart needs clearer visibility into which basket compositions and store formats create stronger value and repeat behaviour across groceries, liquor, and general merchandise.
Focus
I built a synthetic basket-level retail dataset across Makro and Game, then ranked basket types, category combinations, promotions, and shopper signals.
Outcome
The case shows where higher-value retail behaviour actually sits and where Massmart should focus its next basket-building moves.
// analyst.signals
Basket behaviour modelling
Store-format comparison
Decision-oriented retail analysis
Overview
This case study looks at how basket composition and store format shape higher-value shopping behaviour across Makro and Game in Gauteng.
Business Context
Retail growth is not just a traffic problem. It depends on whether shoppers build baskets that are larger, broader, and more likely to repeat. For Massmart, that means understanding which store formats and category combinations create stronger basket value and cleaner commercial upside across groceries, liquor, and general merchandise.
Problem Statement
The business needed a clearer view of which basket types, category combinations, and store formats were driving the strongest commercial behaviour, rather than relying on headline transaction counts alone.
Analyst Objective
Build a decision-oriented retail intelligence case study that identifies the basket patterns, promotion combinations, and shopper signals most closely associated with higher-value behaviour in Gauteng.
Stakeholders
- Commercial and category teams deciding which basket missions to prioritise.
- Trading teams reviewing where category combinations should be promoted together.
- Store-format and growth teams comparing where Makro and Game are strongest.
- Leadership teams needing a cleaner view of what higher-value shoppers actually look like.
Key Questions
- Which basket types create the strongest basket value and repeat behaviour?
- Which category combinations contribute the most value and penetration?
- Where is Makro outperforming Game, and where is Game outperforming Makro?
- Which promotion combinations create the strongest uplift, and where is the margin risk highest?
- Which shopper signals are most useful when identifying higher-value customers?
Analysis Logic
- Basket value must be derived from category mix rather than assigned independently.
- Store format matters because Makro and Game are built for different basket missions.
- Repeat behaviour should correlate with basket quality, but still include noise so it behaves like retail rather than a scoring formula.
- Promotions are only useful when they are read together with uplift, margin pressure, and basket composition.
KPI Framework
- Basket metrics: basket value, basket size, category diversity, private-label share.
- Behaviour metrics: repeat shopper rate, basket archetype distribution, high-value shopper share.
- Category metrics: combination penetration, value contribution, mix effect.
- Commercial metrics: promo uplift, margin compression risk, store-format ranking.
These metrics mattered because the project is meant to support commercial decisions, not just describe transaction activity.
Approach
- Designed a synthetic basket-level retail dataset with four basket archetypes: Quick Trip, Stock-Up, Liquor-Led, and Mixed Basket.
- Modeled category value across groceries, liquor, and general merchandise so basket totals behave realistically.
- Built ranked outputs for basket types, category combinations, store formats, and promotion impact.
- Added a decision layer that answers what Massmart should prioritise next.
Insights
- Mixed Basket is the strongest value and repeat-purchase archetype in the portfolio.
- Makro outperforms on basket size and category breadth.
- Game remains better aligned to smaller, more accessible shopping missions.
- Category combinations matter more than isolated single-category pushes.
- Promotions create real uplift, but not without visible margin pressure.
Deliverables
- Synthetic basket-level retail dataset across Makro and Game
- Ranked basket-type, category-combination, store-format, and promotion outputs
- Notebook and exported visuals for stakeholder storytelling
- Published standalone project plus embedded portfolio case file
Results
- Overall repeat-shopping rate landed at 49.3%.
- Quick Trip remained the dominant basket type by count.
- Makro led on average basket size and repeat behaviour in the final model.
- Mixed Basket delivered the highest average basket value.
Recommendations
- Prioritise Mixed Basket growth because it combines the strongest value and repeat behaviour.
- Promote category combinations rather than single-category offers wherever basket crossover is already visible.
- Use Makro to deepen stock-up and mixed-basket missions.
- Use Game to expand accessible trips and household-led basket growth.
- Treat the highest-uplift promotions carefully when margin compression is visible.
Embedded Project
Links
- Live project: Massmart Gauteng Basket & Store Format Intelligence
- GitHub repo: massmart-gauteng-basket-intelligence