1. Background
An FMCG company, a market leader with nationwide coverage, servicing approximately 43,000 retail stores (points of sale or POS) with an average sales frequency of once per week per store.
The company had already achieved maximum numerical distribution (quantitative growth):
However, the challenge of driving further sales growth became a pressing issue, as the current strategies and resources no longer delivered the desired results.
2. Our approach
It became clear that to achieve further growth, we needed to shift focus from quantitative to qualitative expansion - specifically, increasing sales in existing stores.
To achieve this, we developed an analytical system designed to track key store performance by lost stores, new stores, and stable (repeat) stores. This allowed us to concentrate efforts on retention and growth within our existing base.
In March-April, the store base was categorised and analysed as follows:
We also developed a ranking system for each store using a modified ABC analysis, based on the Pareto principle (20/80). Our approach included an additional level, i.e. ABCD - where stores were classified as follows:
Rank A: High-value major stores (50% of total turnover);
Rank B: Good-value major stores (30% of total turnover);
Rank C: Minor stores with growth potential (15% of total turnover);
Rank D: Minor stores without potential (5% of total turnover).
Next, we needed to build an analysis to monitor this qualitative distribution over time. The key idea was to apply the ABC analysis dynamically. This led us to develop what we called the "distribution matrix." It may sound complex, but the concept is actually straightforward.
3. What is the Distribution Matrix?
The distribution matrix is a dynamic analysis tool that tracks how stores move between different performance ranks over time, comparing past and current periods. It allows us to see how stores migrate from one rank (A, B, C, or D) to another.
Each store was assigned two letters: one representing its rank in the previous period and the other reflecting its rank in current period:
Thus, we created a 5x5 matrix, including lost and new stores, for clear and comprehensive visualization of store performance and rank transitions:
In the matrix, rows represent store ranks from the previous period, while columns show ranks in the current period. For instance, a store ranked as Ca indicates it was a high-value client last month but has experienced a significant decline in sales this period. However, there is potential to address and rectify the situation.
But let's look at the real data, which can reveal a lot of insights.
So, to recap,
March: 45,060 stores;
April: 43,392 stores.
4. What insights can be drawn now?
At first glance, losing more stores than we gained (5,936 lost vs. 4,268 new) seems negative. However, a deeper analysis shows that 3,911 of the lost stores were rank:d, contributing only 5% of turnover. These were low-value stores, so losing them has minimal impact.
On the positive side, we attracted more new rank:A and rank:B stores than we lost, meaning we acquired higher-quality customers.
We saw 686 stores move from rank:b to rank:A, while only 295 stores dropped to rank:B, highlighting a positive trend in upgrading store performance.
The main focus for the sales team should be on the rank:a - clients that have seen a significant drop in sales. Specifically, 26 clients have fallen to rank:C and 13 - to rank:D. Addressing the reasons for these declines is crucial to stabilizing our key stores.
For faster and more intuitive visual analysis, we introduced color tuning to the matrix. By comparing pairwise symmetrical values (e.g., Ab vs. Ba), we highlighted the larger square with a more saturated color. This quick visual cue made it easy to grasp overall trends within seconds, enabling faster decision-making at a glance:
In conclusion
While we didn't invent anything new, we adapted the classic ABC analysis by making it dynamic and visually presenting it in a matrix format.
This approach enabled the sales team to focus on key clients with growth potential, avoiding the clutter of unnecessary data that often overwhelms. The sales team - which was my target audience for this analytics - quickly grasped how to use the matrix for efficient store analysis.
The result? Despite stagnant numeric distribution growth, we achieved overall sales growth by focusing on and better managing high-value clients.
Comentários