23% sales growth thanks to basket analysis
We checked which products customers buy together in a local drugstore. Changing the product layout on the shelves brought an immediate effect.
Drogeria Natura Lubelska had thousands of saved receipts, but no one analyzed them. We checked what actually lands in the baskets of customers from Lublin. Thanks to this, we changed the store layout and increased profit without spending a penny on additional ads.
The challenge
The store struggled with low average receipt value, which stayed flat at 34.70 PLN for 11 months. Customers usually dropped in for one specific thing and quickly left. Attempts to manually arrange promotions at checkout did not bring an effect, because no one knew what actually encouraged people to make larger purchases.
By the way, the owner suspected that the problem was competition from large chains, but the data showed something completely different. The store had regular customers who simply didn't see complementary products because they were in different aisles. Staff also lost about 4 hours a week trying to guess how much stock to order for the following week.
Our approach
Our 3-person team from Lublin Logic Services extracted data from the cash register system for the last 412 days of operation. We analyzed over 14,000 individual transactions. We applied a simple neural network model to analyze association rules to find product pairs that most often occur together.
We focused on hard facts. Instead of guessing, we used mathematics to calculate the probability of buying a second product when choosing the first. We identified 12 key changes in product layout that had the greatest earning potential.
The solution
We implemented a new display scheme in the facial cosmetics and supplements aisles. For example, cotton pads and specific types of toners, which the algorithm linked in pairs, landed next to each other at eye level.
Additionally, we prepared a simple Excel sheet for staff connected to our model. Now every Tuesday, the system automatically suggests which 3 products are worth placing on the promotional shelf at the checkout. The selection is based on what is currently selling best in the entire Lublin area that week.
Results
After implementing changes in December 2023, the store noted an immediate increase in interest in products from higher margin groups. Numbers don't lie – total sales increased by nearly a quarter.
Timeline
-
October 2023Retrieval and cleaning of the base of 14,000 receipts from the register system.
-
November 2023Data analysis and creation of the product association model.
-
December 2023Rearranging aisles and implementing the weekly recommendation system.
-
January 2024Summary of financial results and audit of effects.
"We were skeptical about Big Data analysis in such a small store. It turned out, however, that the data from our receipts is a gold mine. The average basket jumped by over 14 PLN in less than two months."