If category or market share are key initiatives for your organization, it’s likely that you are trying to determine your most promising sources of new revenue.  Companies can no longer devise plans for the year (based on traditional approaches such as segmenting and clustering) and then sit back and relax. Today, businesses must seek to maximize the potential of these strategies by determining each customer’s propensity to buy – that is which customers are most likely to purchase more of the same product or another product. Because it’s always easier to sell to existing customers, it makes sense to focus your initial data analytics on building models to help maximize your share of your customers’ (share of wallet) business. Cross selling and up selling are natural applications for these kinds models which are leverage predictive analytics, because an organization generally knows far more about current customers than it could possibly find out about external prospects.


Create Marketing Models

customer-centric data purchase models analytics
Use Data to Create Customer-Centric Models

You need data to create any model.  A model is a way to organize your data and how the data elements relate to each other to help you see a pattern and understand and impact something in the “real” world. A model is only as good as the data. Models are dynamic, the more dynamic the world you’re in, the more dynamic your model needs to be in. If you’re in a business that is highly transactional with lots and lots of transactions, you’ll need update frequently.  Whatever the sales cycle is for the company is the frequency in which the model updated.

To address share of wallet, many organizations would need to know the following:
•    which customers are most likely to purchase a specific product (propensity to buy models)
•    what is the next product a customer is likely to buy (next most likely product model)
•    which customers are likely to defect or reduce expenditures (attrition predictor models)

Each of these takes data.  Data cleansing, transformation, initial and ongoing validation etc. are required before you can build any model.


Propensity-to-Purchase Models

As with all model building, the key challenges are defining a customer, extracting quality data and analytical and modeling skills.  We see the terms “propensity modeling” and “predictive modeling” used in many articles. They sound alike, but there is probably a subtle distinction between the two. Predictive modeling is a broad category – you can build predictive models for all sorts of things, such as a predictive model around customers at risk for defection. A propensity to purchase is a type of a predictive model.

The purpose of a propensity model is to understand the likelihood a customer will be predisposed to purchasing  a product based on a purchases they’ve already made at some point in time.  Traditional propensity-to-buy models score customers based on their similarity to past purchases. These models require having historical data and measuring past performance of the enterprise regarding offerings and customer activities so that you can effectively deploy cross-selling and up-selling techniques. Cross selling involves getting existing customers to purchase additional services. Up selling refers to convincing these same customers to trade up to products that are more profitable.

Who is likely to buy?
A number of data elements are essential to creating a model to identify who is likely to buy.

Propensity to purchase models need data related to these variables: who has bought, what have they bought, when did they buy it, in what order did they make the purchases, and what combination of products were bought by which set of customers. Firmographic and macroeconomic data can also be useful for model construction. The data is used to create a model that essentially scores each customer. The score represents the likelihood the customer will want to purchase. The top-scoring products for each customer become the customer’s best next offers.

However, history is not always a predictor of the future. The solution is to build propensity-to-buy models based on the way purchasers look just before they purchase the product and then leverage the analysis to predict outcomes of prospects.

Feeding real-time data to the propensity-to-buy models and injecting predictive analytics functionality, such as decision trees, are important for effectively using the model.  It is also important to have both a treatment and a control group to see what response you get. This allows you to identify the buyers in both groups and then evaluate what kind of people did not buy in the control group but did buy in the treatment group – these are those who only purchased because of the treatment. From this you can build the propensity to influence model. Need help, let’s talk about how you can tap our expertise.

Comments are closed.


“I love your articles and advice – I feel like everything you write is thought-provoking and actionable.” – Marcie, Marketing Director, Technology industry.

Join our community to gain insights into creating growth strategies and execution; and employing growth enablers, including accountability, alignment, analytics, and operational excellence.

Follow VEM on Twitter