Marketing Intelligence
Built on years of turning market data into clear strategy and measurable results since 2016
What the client needed
A collectibles retailer struggled with email because the list had no usable segmentation by product interest, behavior, or location. The catalog had 100,000+ SKUs across 7 macro categories, each with 3–7 subcategories, but that taxonomy was never tied to customer profiles. Most buyers were repeat customers, yet shipments, category affinity, and basic demographics were not tracked. Past emails were one-size-fits-all blasts and underperformed.
Without real clusters, offers could not match collector type or rarity. There was no radius targeting or store-level attribution. Omnichannel was weak: after buying online, customers rarely went to stores because follow-ups were generic, local inventory and events were not highlighted, and coupons lacked store codes or scannable IDs for POS attribution.
My part of the project
I started by cleaning and standardizing all customer and purchase data. The old files had inconsistent product codes and customer IDs, so I built a single structure that linked transactions, demographics, and location. Then, using R, I modeled purchase clusters and grouped customers by similar behavior and product preference, which revealed clear collector profiles for each category.
With that foundation, I created segmented campaigns tested with control groups to measure real impact. Each cluster received targeted content such as exclusive launches, restock alerts, or specific product lines instead of generic offers. The campaigns were tracked end to end, from send rate to conversion uplift.
To connect online and physical channels, I designed omnichannel campaigns that allowed online purchases with in-store pickup. QR codes and store-specific coupons enabled POS attribution, while activation schedules ensured stores adjusted displays and engagement for each audience segment.
