AI in marketing & sales: Propensity to buy

"Propensity to buy" is a value which represents how likely a customer is to purchase a particular product. Successful propensity to buy models give crucial insight into how to design and distribute marketing material as well as allocate sales staff time. This allows companies to be significantly more efficient with their marketing, leading to increased sales without any increased costs.

To buy, or not to buy, that is the question.

Want to replicate this use case in your business? Get in touch

02/ What can be achieved?

A model like this one can classify potential customers by their likelihood to purchase a particular product. This can be integrated into marketing and sales strategies.

03/ The problem

A lack of knowledge about a customer’s propensity to buy leads to one-size-fits-all marketing, where time and resources are blindly allocated to all potential customers.

04/ Opportunity for deep learning

Deep learning has the ability to find patterns in large datasets with complicated data-types. The model could for example use a combination of semantic analysis of text written by the customer, demographic information, purchase history as well as information about how they navigate the website to make a prediction for that customer’s propensity to buy.

05/ How is the AI model implemented?

E-commerce is an example where the vendor typically has easy access to information about their customers. Once a propensity to buy score is established, this could be used to redistribute how discounts are handed out to customers. Customers with high propensity to buy require smaller discounts in order to make the purchase than customers with a low propensity to buy. This insight leads to higher sales and retention without any increased costs.

Another example of the benefit of a propensity to buy model is in business marketing. Business sales typically have longer decision cycles, higher average order values and larger influence of interactions with a sales team. Propensity to buy scoring will allow salespeople to allocate their time more effectively, again increasing sales and revenue without any additional costs.

06/ Data requirements

A model like this would need historical data of demographics and pre-purchase behaviour of customers linked to if a purchase was made.

Follow our tutorial to learn how to build an AI model for this problem on the Peltarion Platform.

07/ Where to learn more?

To learn how to build a propensity to buy model using a banking advertising dataset, have a look at our tutorials on building and improving AI models for tabulated data.