The best way to start down the path of expanding your customer base or choosing a site for building a new location is to gain a deeper understanding of your existing customer base. In other words, you need to answer the question: who buys from me?

The first step in the modeling process is often starts with market segmentation. This involves dividing your existing customers into groups or segments by using identifiable attributes, such as age, income, behavioral or psychographic features. Groups are determined and ranked based upon their appropriateness as targets for particular marketing efforts.

Unfortunately, many business owners approach the market segmentation process based upon educated guesses regarding which variables are most predictive of marketing success or failure.

By contrast, predictive modeling involves testing multiple segmentation variables in sophisticated ways that results in a model of which variables and to what degree should be used to build a segmentation model. Predictive modeling leverages various techniques to combine the given variables.

Depending upon the nature of your existing customer list and your marketing goals, typical data sources that we use during the predictive modeling process include:

• existing consumer segmentation data
• demographic data
• geographic data
• behavioral data, such as how often a person buys from you, which products or services they buy and how much they spend

We combine all of this data using statistical and in some cases machine learning techniques that help us find meaningful patterns in your customer and sales data.

The result: a detailed, quantitative model of your best customers. Most of our models also include qualitative descriptions of “who” your customers are, yielding insights into:

B2C:

• Which segmentation clusters represent your historically-best customers
• Which media they buy
• Psychographic information, such as how they perceive certain types of advertising and what motivates them to make a purchase
• Demographic information, such as average incomes, ages, home values, and more
• Geographically where you can find more prospects just like your best customers either by market, city, zip code, or even by household
• Specific prospect lists

B2B:

• Which types of companies (as classified by SIC or NAICS code) tend to be your best customers
• Where you can find more companies that resemble your best customers
• Specific prospective company lists

Product and Upsell Opportunity Modeling

Our data modeling services do not stop there. When analyzed properly, your customer address data and purchase history data can reveal important patterns, such as:

1. which products tend to be purchased in tandem, indicating possible product bundling opportunities
2. which products, when purchased now, are likely to be associated with future purchases of related products
3. which types of products are under-represented by your product line
4. which types of products are typically running out of stock and therefore represent lost sales opportunities
5. which customers are at the highest risk for churning (i.e., to stop buying from you)
6. which customers represent upsell opportunities

To round out all of this, we can design a specific marketing campaign to help you reach out to your best prospects using virtually any type of media – all with the right message sent at the right time.

Find out more about key aspects of predictive modeling, including statistical modeling, artificial neural network modeling and scoring & ranking

Contact MindEcology today to get started.