Modeling & Predictive Analytics

Develop a more effective customer acquisition plan

What are they and how they can help you develop a more effective customer acquisition plan

Acquiring new customers is critical to the success of your business. But how do you know which prospects to target? In this edition, Data expert, Steve Quast shares how customer acquisition models help brands answer this question and develop more effective customer acquisition strategies.

Meet the expert
Steven Quast
VP, Analytics & Insights

As Vice President in Data Axle’s Analytics & Insights organization, Steve has extensive experience providing consultative analytic support to marketers. He manages a team responsible for the development, testing, and implementation of analytical solutions including revenue attribution, predictive modeling, customer segmentation, and custom analysis applications. Steve has a BA in Mathematics (Statistics) and Economics from St. Olaf College in Northfield, MN. He works out of the Minneapolis office.


What are customer acquisition models and what value do they provide?

Customer acquisition models help marketers maximize acquisition budgets by focusing efforts on prospects that have higher conversion rates and are less costly to acquire. These models support all direct-to-consumer marketing channels, including direct mail, email, display and social. They are customizable to attract prospects who meet specific criteria once they convert, such as likelihood to:


Are there different types of customer acquisition models, and what makes them unique?

Below are the three most common customer acquisition models:

  • LOOK-ALIKE MODELS (also known as Clone models)
    Look-alike models most commonly identify prospects with demographic and psychographic characteristics (age, gender, income, hobbies, interests) similar to those of your best customers. Prospects who resemble your best customers generally make the best targets for new customer acquisition because they convert at a higher rate.

Why It Works: We developed a look-alike model for a direct mail campaign to attract prospects that look like high value customers. We discovered that prospects who closely resembled the existing high value customers had 23% higher response rate than the rest.

    They identify prospects most likely to respond to specific marketing initiatives.  Often “response” is defined as a purchase, but response can also mean an email open, a display ad click, a ‘follow’ on social or a lead gen action such as a form submit or a call for more information.  To ensure the response model works well across marketing channels, it should include all cross-channel responses.

Why It Works: When we built a response model for a large retailer, we discovered that the top 20% of prospects ranked by the model had 46% higher than average response rate while the bottom 20% had 32% lower-than-average response rate.


    These models predict the dollar value of prospects after they become customers. The models utilize a forward-looking timeframe, such as 12 or 36 months, in order to inform marketers how to shift their budget and help them prioritize initiatives that target prospects with the largest potential value.

Why It Works: Prospects with the highest predicted future value are acquired at the same rate as other customer acquisition targets. However, revenue generated by these prospects in the year following conversion is significantly higher than that of other new customer segments.


What data sources are required to build a robust customer acquisition model?

Customer Acquisition models are largely dependent on 3rd party data, including:

  • Demographic data (age, gender, presence of children in household, income, etc.)
  • Psychographic data about lifestyle and attitudes (i.e. outdoors enthusiast, sports aficionado, avid book reader)
  • Geo-location, defining weather patterns and other regional characteristics


Zero and first party data such as purchase activity, email engagement (opens, clicks), preference center data and other customer information may also be required.


How often should models be refreshed or updated?

We recommend revisiting models on a bi-annual basis to ensure they reflect current customer behaviors and attributes. The composition and behavior of a brand’s customer base changes over time due to shifting markets, business objectives and marketing strategies.

Response models are especially sensitive to changes in marketing channels or tactics (e.g. creative considerations and offers), so these models need to be re-evaluated (and potentially redeveloped) when such changes occur.


What other considerations are pertinent to customer acquisition models?

Models can and should be combined with other tools. For example, we create micro-targeting functionality by combining segmentation techniques (i.e. personas) with modeling techniques. The combination enables the marketer to target on two levels – relevancy and likelihood to respond.

A two-step process to creating accurate micro-targeting:

  1. First, develop personas.
    For example, let’s say there are three distinct segments identified within a brand’s acquisition audience: young families, families with older children and empty nesters. Marketers can use this information when developing creative and selecting images for communications targeting each of these segments. In this example, young families would see pictures of children at play, families with older children might see kids playing sports and empty nesters might see older adults traveling.
  2. Next, develop a response model to identify which prospects to target.
    The model will not weigh the personas you’ve identified in step one equally because their responsiveness will vary. The most responsive persona will comprise the largest percent of the campaign with the other two making up the balance.
Stay in touch! Subscribe to our newsletter.