Market Basket Analysis
Market basket analysis is a statistical method intended for analysing customer buying behaviour. It contributes to the growth in sales thanks to improved direct marketing, sale support, loyalty programmes, etc.
The majority of store purchases are made on an impulse and it is difficult to explain them using traditional analytical methods. Deloitte’s market basket analysis offers ways of providing the client with the right shopping impulse at the right moment.
Deloitte’s market basket analysis is a method of analysing customer shopping behaviour. Instead of relying on the costly process of acquiring new customers, retailers may use the market basket analysis to achieve revenue growth in respect of its existing customers. The using of a self-service tool in addition to the existing database of historical purchases will make it possible to identify the patterns of customer shopping behaviour in a visual and interactive way.
Benefits you will derive
The market basket analysis contributes to revenue growth as it delivers improvement in the following areas:
- Cross-selling and up-selling
- Sales support and direct marketing
- Loyalty programmes
- Product placement and store appearance
- Supplier chain
- Product bundling
The objective is to identify usual patterns of shopping behaviour using large sets of data. Given the traditional IT infrastructure capabilities, the possibility of identifying such behavioural patterns is usually limited. The market basket analysis generates a set or rules taking the following form: “if the basket contains goods X, then it also contains goods Y”. Any such rule creates a link between the two types of goods in a way that may be illustrated in a chart as an edge (arrow) and its frequency depicted using different-sized circles in the middle of the arrow. Store managers may use the chart to learn about the shopping behaviour patterns as identified in the data and apply this knowledge to product placement, among other things.
The market basket analysis tool has been designed as a self-service. This means that the user first uses the form to enter the time interval, the stores and the categories of goods in respect of which they wish to carry out the analysis. Subsequently, they will launch the calculation, its result being a comprehensive report which will be saved in the database and which may be accessed at any time in the future. The report has been designed to be visual and highly interactive as there are numerous types of goods and the user must have the option of finding interesting patters fast or focusing on a specific group of goods.
The data entering the calculation algorithm include items on receipts, which are indeed big and detailed data. To process such a large volume of data, special big data technologies are needed. We prefer using a Hadoop cluster. Another option is to use cloud services with readily available computing power or a combination of these (Hadoop in the cloud).