Even though cross-selling differs from up-selling they use very similar analytical approaches and methods. Cross-selling is selling additional products to existing customers, and its attractiveness owes to the several facts; firstly it increases revenue per customer, secondly it helps with retaining existing customer since customer with more than one product are less likely to switch to the competitor, and thirdly - it is cheaper way of getting the revenue from existing customer than from new customer given high acquisition costs. Unlike the acquiring of new business, cross-selling involves an element of risk that existing relationships with the client could be disrupted. For that reason, it is important to ensure that the additional product or service being sold to the client or clients enhances their value to the organization. There are three basic ways of cross-selling: 1.) while servicing an account, the product or service provider may hear of an additional need, unrelated to the first, that the client has and offer to meet it, 2.)selling add-on services is another form of cross-selling. That happens when a supplier shows a customer how value can be enhanced if buying another products or service by supplier, and thirdly, it can be done through “solution” selling where product is sold within package with installation services for example, or insurance package, or even delivery.
So, when you cross-sell, you always offer the customer a product or service related to whatever they are already buying, and therefore many cross-selling opportunities arise naturally. (Note of caution - not to overload the customer otherwise you may push him away). Customers always appreciate recommendations, especially if that recommendation doesn’t always lead to the most expensive product in a shop. That build the trust and customer is more likely to act on your recommendation. Up-selling is somewhat different since it positions higher priced products in a good/better/best progression. It is sales technique geared toward existing client to induce him to purchase more expensive items, take-up upgrades, or other add-ons in an attempt to make a more profitable sale. Upselling usually involves marketing more profitable services or products but can be simply exposing the client to other options that were perhaps not considered. In practice, large businesses usually combine upselling and cross-selling techniques to enhance the value that the client or clients gets from the organization in addition to maximizing the business's profit. In doing so, the organization must ensure that the relationship with the consumer is not disrupted. So if you suggest to the customer to go for your premium brand, buys extended service contract, buy larger computer disk, opt for “super-size” meal or go for more extensive insurance coverage – you doing up-selling!
Up-selling is somewhat different since it positions higher priced products in a good/better/best progression. It is sales technique geared toward existing customer to induce him to purchase more expensive items, take-up upgrades, or other add-ons in an attempt to make a more profitable sale. Upselling usually involves marketing more profitable services or products, but can be simply exposing the customer to other options that were perhaps not considered. In practice, large businesses usually combine upselling and cross-selling techniques to enhance the value that the client or clients gets from the organization in addition to maximizing the business's profit. In doing so, the organization must ensure that the relationship with the customer is not disrupted. So if you suggest to the customer to go for your premium brand, buys extended service contract, buy larger computer disk, opt for “super-size” meal or go for more extensive insurance coverage – you doing up-selling!
So, how can analytics help organizations in cross-selling and up-selling? There are several different ways. First one is through association, or market-basket analysis. This technique produces set of rules that identifies association between certain entities in transactional data-set such as products, services, etc. Rules such as “If product A then product B” can be ranked based on metrics such as “confidence” which express conditional probability that if product “A” is in a basket, product “B” will also be present. Another metric confidence derived metric is a lift that, expresses the measure of pull that one product has on the other product. Lift of 2 mean that product “B” is twice as likely to be in a basket if product “A” is there. Last metric is a measure of support that denotes how often this association occurs. It is important when evaluating these rules to focus on rules that have high confidence, high lift and relatively high support (associations that occur very infrequently have little practical value). Also, it is important to focus on potentially useful and non-trivial associations and ignore associations that are rare and inexplicable. So, how do these useful and non-trivial associations can be used to cross-sell and up-sell? If there is natural “pull” from product “A” to product “B” - which product you should try to cross-sell? Certainly, it is product B, however if the customer has this product already but doesn't have the product A – then product A should be the one to sell. This type of associative product knowledge opens many marketing opportunities, such as having product in association closer in the store, or placing slow moving product in between them, giving discount on one while raising price on the other and never advertising or giving discounts on both in a same time, etc.
Another analytical method for cross-selling is product clustering which answer a similar question as association analysis– what is the typical product portfolio that is purchased together. Main difference here is that this method is done on known client data records as oppose to anonymous cash till slip records for association analysis. That means that product holding clusters could be overlaid with demographic clusters providing specific customer profile that buys specific product portfolio. Cross-selling actions are clear – promote the customer whatever he is missing from the portfolio, and promote it in a way that is in sync with its profile.
Predictive modeling approach works well for high value less granular items. Here, there would be a target variable that indicates purchase of a specific item in a period of time, or purchase volumes or frequencies above a specific level. Input variables would be behavioural and demographic variables and output of such model would be the probability of purchase of such model. When this model is deployed on a new data it is likely to give high probability to customers who don't have this product but in some ways they look like customers who do. Assumption here is that those customers are more likely to purchase such product if given right marketing stimuli than randomly selected customers. After new data is scored customers ranked above some probability cut-off company would some promotional material related to this product.