Since selling a product to an existing customer is ten times easier than selling it to a new one, many businesses focus on strategies like service analysis, cross-selling and up-selling to target already existing customers.
Service Analysis is an important procedure that needs to be followed by any organization to understand the standard of service that is being provided. It is about checking what a customer may be interested in, depending on the services or products he had already purchased.
On the other hand, while cross-selling is about selling items in other categories the customer might be interested in, up-selling is about selling additional items that complement the ones purchased. The easiest way to explain cross-selling and up-selling is the one with MacDonalds, the godfather of such strategies. “Would you like fries with your burger” is cross-selling and “would you like to upgrade your fries and soft drink to the large size” is up-selling.
But, what is the role of ML?
Having the data collected and analyzed, a business can identify what items (A & B) are usually purchased together, to offer item B to the customer who is purchasing the item A. That’s what Amazon, Spotify, and most businesses are doing today. Predicting someone’s shopping list leads to selling him more items, thus more profit.
Suggesting a bag for a customer purchasing a new laptop is something every salesman knows and predicts, but working with millions of services or products and millions of customers with different shopping taste is not as easy to understand, as suggesting a bag for a customer who’s buying a laptop. Therefore, instead of manually trying a couple of discounts and suggestions, ML allows testing thousands of different options leading to the best possible one.
Moreover, ML never stops learning from new data, thus adapts to the modifications in customers’ purchasing behavior, in addition to dynamic pricing allowing to modify the price of an item depending on different criteria.
It is all about what (product), whom (customers), when (timing), and how (channel).
Finding the best items to suggest:
As already mentioned, it is all about finding the best product or service to suggest to the customer. Therefore, starting with grouping the customers by demographics, gender, interest, etc. is a good idea. Knowing what customers in a specific group purchase with a TV gives a vision about what to suggest to a customer belonging to this same group and purchasing a TV. Not a single business would like to end up suggesting an old lady new headphones.
Purchases of each group of customers are analyzed to get the probability of each item being bought with another one. For the group of customers C1, if item A is bought 10 times and 8 of which item B existed in the same shopping list, the probability of buying item B when buying item A is 80% for this group of customers. The highest the probability between items A and B, the more it is worth suggesting item B to a customer purchasing item A, the more the customer is likely to purchase.
Computer Engineer • Entrepreneur • Blogger