13 Apr 2018
Basket Analysis for eCommerce and Brick-and-Mortar Retail
Data Mining & Text Mining are gaining a foothold across many verticals, including eCommerce. While many retail merchants are already aware that this Data Science and NLP techniques can benefit them online, they barely have any idea about how much of a boon they could be for their brick-and-mortar business. Nevertheless, making friends with Data & Text Mining can pack a wallop for you, as a retailer, across the board. Moreover, some of the more intricate applications of Data and Text Mining are suitable for both eCommerce and brick-and-mortar stores and you can use them in parallel in order to enhance the effect.
We’ll dwell on just one such application, so that you can decide whether bringing it into play can bring you closer to your business goals.
Basket Analysis: How NLP and Data Science Can Help You Sell More
One of the awesome techniques you can apply both online and offline is Basket Analysis. Even the name of this approach suggests something you deal with day in day out in both your physical stores and on the Web: baskets or carts filled with your products by your customers. Did those items wind up in the same basket by pure chance? How often do they find themselves in the same company? How many baskets like that do you process over a meaningful period of time? Can that be considered a trend?
You will have probably deduced by now it deals with identifying a pattern in accordance with which the products you sell co-occur. Is that possible to do? With Data & Text Mining, it certainly is. You can explore the invisible relationships between the different products and product groups that make up your online and offline inventory and determine which of them “gravitate to one another”.
How can you benefit from this knowledge?
In fact, the ability to identify which of your products sell together opens up an entire host of practicable opportunities.
- If you have determined there is an A>B relationship between two products or groups of products, you can offer a complementary product B to those online shoppers who have added to cart Product A. By means of Text Mining that is designed to understand a product’s attributes from text the way a human would do, you can further tailor and enhance this offering. This can be done in several ways:
- You can display a complementary product or products based on the description and properties of Product A that has been added to cart.
- You can offer a complementary product or products based on your shopper’s previous searches for products of this category. In this case, the complementary products you can offer can feature the properties your shopper was interested in during their previous searches, which will increase the selling odds.
- Basket Analysis allows you to optimize the arrangement of your products in your eCommerce store. In addition to placing together products that have proven to be associated, you can also approach the equation from the other end and identify product categories one and the same shopper is unlikely to be interested in. Someone making a search for wholesome herbal teas is not very likely to buy strong liquor during the same visit, don’t you think so? More so, by using your Basket Analysis insights, you can eliminate any undesirable negative associations that may affect your visitor’s buying decision.
- Text Mining gives you another great ability that can further enhance the way you present your complementary products on a suitable occasion prompted by Data Mining: the ability to display relevant customer reviews. When a complementary product gets recommended, a Sentiment Analysis-driven capability of your system can identify positive feedback related to this product or products starting with the most recent, and display it to your shopper.
- With physical stores, it’s a bit more difficult to juggle things and cough up whatever can be relevant at the right instant in time. However, the insights you’ve gleaned online can be projected on to the brick-and-mortar part of your business just as well. By means of merchandising, you can still position together products that “gravitate” toward one another. You can also try to avoid provoking any negative associations.
It is beyond the scope of this article to describe some of the benefits of Basket Analysis that are not associated with influencing buying intent, but they, nevertheless, are present too. In particular, Basket Analysis can be used for product inventory-related predictions.
Sources used for this article: