Understanding shopper behaviour at the retail store is the key to succeeding in retail. While shopper behaviour is uni-dimensional and straight forward in the cases of high value goods such as automobiles or electronics, it is not the same in case of low value goods such as FMCG. A visit to the store is mostly exclusive in case of high value goods and understanding the motivations of the shopper on an exclusive purchase visit is not very complex. But understanding the motivations of shoppers in consolidated shopping trips such as those for FMCG is quite complicated. This is because the shopping basket is a collection of many products and each customer has a more or less unique shopping basket.
Shopping Basket Analysis in Indian FMCG context is further complicated due to low accessibility of data. Modern Trade stores are fairly ahead in terms of this kind of analysis but brands and companies do not have the access to that data. And general trade, which still contributes an overwhelming majority of the business, is notoriously laggard when it comes to technology adoption and modification of traditional business practices.
We have managed to build a network of FMCG general trade retailers and are collecting their sales to customer data. After thorough cleaning up and organization of this data we were able to run shopping basket analysis on top of it to generate some key insights.
The insights generated are association rules which have three basic metrics.
Support: For products say A and B, support is defined as the percentage of total shopping baskets which have both A and B products in them
Confidence: For products A and B, confidence is defined as ratio of shopping baskets containing A and B to shopping baskets containing A
Lift: For products A and B, Lift is defined as ratio of shopping baskets containing A and B to product of shopping baskets containing A and shopping baskets containing B
The way to interpret these metrics is that support and confidence are used as hygiene factors and a certain cutoff level for both is determined initially. Lift is the actual increase in probability of purchasing B if A is present in the shopping basket.
Let us take the case of the subcategory of ‘Air Freshener’. Above a predefined support and confidence levels, the following subcategories show high lift due to air freshener.
The chart shows that if Air Freshener is already present in a shopping basket, the chances of Cleaner being present go up by 5.6 times, and so on. The top two subcategories showing the highest lift both belong to Home Care category. Air Freshener is a subcategory under Home Care category as well. This phenomenon of showing the highest association with subcategories in the same category is observed almost universally. This means that in general trade FMCG context, if a customer purchases products from a category, her chances of picking up other subcategories of that category go up. This is the biggest rationale for stocking products of same category together. Something that the liquid business acumen passed down over generations in FMCG trade communities agrees with.
But if you take a look at the other categories which show high lift, then a picture of the shoppers of Air Fresheners begin to emerge. Tooth Paste, Fabric Wash powder, Soap, etc are all products which do not show an explicit association to Air Fresheners. But the data shows that consumers of air fresheners are almost 3 times more likely than others in purchasing these products.
How can this data be used? It can surely be used in store layout design and merchandising strategies. For a new Air Freshener brand, capturing merchandising or stocking space near high association products such as cleaner and insect repellent might not produce results as it can get lost in the clutter of other air fresheners present there. But if it can manage to get shelving space next to tooth paste, it would truly be a clutter breaking display of products which would attract customer attention. It would also not have to compete with other air fresheners brands there.
Shopping Basket Analysis can also be done at individual brand level. These would provide insights in terms of contrasts among various brands in the same subcategory. Do all brands show similar behaviour or are there brands out there which show different basket dynamics when compared to other in the same subcategory? What are they doing differently to reach the customer in a unique way? These are the questions that need to be analysed by all the brand managers. One example of such analysis done for Colgate tooth Paste is as follows.
The above chart shows a clear picture of what other products do customers of Colgate Tooth Paste purchase. Such data across categories can be used to identify if a brand is present in shopping basket along with premium bands or low priced bands. This can be an important metric to analyse whether a brand is able to attract the right and intended TG or not.