Performance Snapshot
Key Metrics at a Glance
Total Chip Sales
$1.80M
Jul 2018 — Jun 2019
Unique Customers
71,185
loyalty card holders
Raw Transactions
264,836
after cleaning: 245,302
Best Trial Uplift
+33.5%
Store 77 — sales uplift
Stores Analysed
264
across trial & control
Top Segment Sales
$147K
Mainstream Young Singles
Analysis
What the Data Reveals
Customer Segments
Mainstream Young Singles Drive Volume — Families Drive Value
Mainstream Young Singles/Couples generated $147K in total sales through volume — 7,908 customers spending just $18.62 on average. Older and Young Families spend $32–$35 per head with 4.4–4.7 transactions each — the highest value per customer.
01
Brand & Pack Preference
Kettle Dominates — 175g is the Universal Pack Size
Kettle is the dominant brand across all major customer segments. The 175g pack size is the most purchased across the category, representing the strongest repeat purchase signal for planogram optimisation.
02
Trial Store 77
Store 77 — Strongest and Most Reliable Uplift Signal
Store 77 (matched to control Store 233, similarity score 0.97) delivered +33.5% sales uplift and +29.1% customer uplift, with 2 of 3 trial months exceeding the 95th percentile confidence band — the most statistically robust result across all three trials.
03
Trial Store 86
Store 86 — Moderate Uplift, Stronger Customer Response
Store 86 (matched to control Store 155, score 0.91) showed +8.5% sales uplift and +12.7% customer uplift. Customer growth outpaced revenue growth, suggesting the trial attracted more buyers but at lower average spend.
04
Trial Store 88
Store 88 — Positive Uplift but Weaker Control Match
Store 88 (matched to control Store 237, score 0.77) delivered +13.2% sales uplift and +6.6% customer uplift. Results should be interpreted with caution — the weaker pre-trial match reduces confidence in the uplift attribution.
05
Data Engineering
19,534 Non-Chip Rows Removed — 20 Brands Standardised
Removed 19,532 salsa dip rows and 1 bulk-buyer outlier (200-unit purchases). Standardised 20 brand name variants (e.g. RRD → Red Rock Deli) and engineered PACK_SIZE and BRAND features from product names using regex.
06
Trial Store Results
Control Store Matching & Uplift Summary
| Trial Store | Control Store | Similarity Score | Sales Uplift | Customer Uplift | Significant Months | Confidence |
|---|---|---|---|---|---|---|
| Store 77 | Store 233 | 0.97 | +33.5% | +29.1% | 2 / 3 (sales & customers) | High |
| Store 86 | Store 155 | 0.91 | +8.5% | +12.7% | 1 / 3 sales · 2 / 3 customers | Moderate |
| Store 88 | Store 237 | 0.77 | +13.2% | +6.6% | 1 / 3 (sales & customers) | Lower — weaker match |
Methodology Note
Control store selection used a composite similarity scoring function combining Pearson correlation and normalised magnitude distance across total monthly sales and customer counts during the pre-trial period. Confidence bands were derived from pre-trial standard deviation (5th–95th percentile). Trial months breaching the upper band are classified as statistically significant uplift.
Business Insights
Structured Analytical Insights
01
Segment Strategy
Volume vs Value — Two Distinct Segment Plays
The chip category is driven by two structurally different customer groups. Mainstream Young Singles/Couples generate $147K in total sales through sheer volume — nearly 8,000 customers making frequent, low-spend purchases. In contrast, Older and Young Families deliver the highest spend per customer at $32–$35 with 4.4–4.7 transactions each. Category strategy should address both groups differently: volume promotions and impulse placements for Young Singles, multi-pack and value-bundle positioning for Families.
02
Trial Recommendation
Store 77 Layout Merits Broader Rollout
Store 77's trial produced the most statistically reliable result — +33.5% sales uplift and +29.1% customer uplift against a control match scored at 0.97. Two of three trial months exceeded the 95th percentile confidence band for both metrics, confirming the uplift is not attributable to normal baseline variance. This provides sufficient evidence to recommend the Store 77 layout for broader rollout, subject to cost validation and store format compatibility assessment.
03
Planogram
Kettle and 175g Should Anchor Category Facings
Kettle's dominance across all major customer segments — paired with 175g as the most purchased pack size — indicates these should receive primary shelf placement and maximum facings in planogram design. Secondary brands and pack sizes can fill adjacency slots without displacing the core repeat-purchase drivers. Removing or de-prioritising Kettle would likely create measurable sales leakage.
04
Data Quality
Bulk-Buyer Outlier Removed to Protect Segment Accuracy
A single customer account purchasing in 200-unit increments was identified and excluded from the analysis. Retaining this record would have materially distorted average spend and transaction frequency metrics across affected segments, producing misleading category benchmarks. This exclusion was validated against segment-level transaction distributions before finalising the cleaned dataset of 245,302 rows.
Methodology
3-Task Analytical Workflow
Task 1 — Phase 1
Data Loading & Type Fixing
Loaded transaction data (264,836 rows) and customer behaviour data (72,637 rows). Fixed Excel serial date format, set correct column types, validated row counts.
Task 1 — Phase 2
Data Cleaning
Removed 19,532 non-chip rows (salsa dips), excluded 1 bulk-buyer outlier (200-unit purchases), cleaned product name spacing inconsistencies.
Task 1 — Phase 3
Feature Engineering
Extracted PACK_SIZE and BRAND from product names using regex. Standardised 20 brand name variants (e.g. RRD → Red Rock Deli, GrnWves → Grain Waves).
Task 1 — Phase 4
Dataset Merge
Merged transaction and customer data on loyalty card number. Zero data loss confirmed — final merged dataset: 245,302 rows, 12 columns.
Task 1 — Phase 5
Segment Analysis
Defined a 10-KPI framework. Analysed total sales, customer counts, avg spend, and transaction frequency by lifestage and premium spending tier.
Task 2 — Phase 1
Store-Month Aggregation
Aggregated transaction data to store-month level — 3,165 records across 264 stores. Separated pre-trial and trial period data for all three trial stores.
Task 2 — Phase 2
Control Store Selection
Built composite similarity scoring using Pearson correlation + normalised magnitude distance across total sales and customer counts. Selected controls: 233, 155, 237.
Task 2 — Phase 3
Uplift Testing
Applied pre-trial scaling factors to control store data. Plotted 5th–95th percentile confidence bands from pre-trial standard deviation to assess statistical significance.
Task 3
Client Presentation
Prepared a 13-slide PowerPoint for Julia (Category Manager) using the Pyramid Principles framework — covering segment findings, trial results, and strategic recommendations.
Task 3 Deliverable
Quantium Category Strategy Report
13-slide client-facing presentation submitted to Quantium's Category Manager (Julia), structured using the Pyramid Principles framework. Covers chip segment analysis, control store methodology, trial store uplift results, and strategic recommendations for the upcoming half-year plan.
View Presentation →