Predictive Marketing Blog

How to Find Your Best Prospects Using Analytics – Part 1

Many Marketing departments start and stop developing their new prospect profile by using market research data from a new customer survey or analyzing their new customers with appended data to uncover buyer demographics. Advertising and marketing dollars are then aimed at the profile that isolated the variables and values containing the largest proportions of their new customers. They conclude: “here are the key characteristics that identify most of our newest customers, and we should create marketing campaigns to target advertising to these buyer characteristics.”

This market research approach in uncovering the profile of new prospects for targeting advertising by looking at current customers in a vacuum can be quite faulty and lead marketers astray and waste valuable advertising and marketing dollars.

The market research customer profile

Let’s suppose you are selling a product directly to consumers and perform a market research study using your newest customers to uncover their demographics for targeting the next campaign (actual data was used for this illustration).

Figure #1: Approximately 25 demographic characteristics were appended to a 17,000 record sample of new customers. The 3 highest percentage characteristics are shown here.

Three demographic characteristics stand out and identify 50% (or more) of these new customers: Variable (1) Occupation = Employed (50%); variable (2) Years at Address = 3-5 years (68%) and variable (3) Household status = Not a Senior Household (61%).




This marketing research profile identifying “most” customers can be stated as: customers are mostly employed, living at their address 3-5 years, and are under 65. Therefore it looks like Marketing should target advertising dollars to this customer type to grow sales.

But this would be an incorrect conclusion and wasteful of marketing resources. Why?

Using additional analysis, the 17,000 sales generated by this marketing program and a random 200,000 sample of names marketed to, we can prove that this derived target is wrong.

Those selected as meeting the market research demographic characteristics (employed, under 65, and living at their address 3-5 years) actually underperformed when compared to sales from the random. Clearly most of the time, selecting prospects based solely on the highest percent demographics of new customers is incorrect. Of course if 100% of your customers are female, 100% are single and 100% are renters, then, of course you can use customer demographics in this very rare, special case to select new ones. But, 99.99% of the time, profiles are not that clearly defined.

Figure #2: When we compare the sales rate from those matching the high percentage target to a random sample to determine if the high percentage names outperform random names, we found:


What is happening here?

The market research target built by combining the highest percentage new buyer characteristics had a sales rate of 1.2%. The random sample had a 1.5% sales rate meaning the approach of using the market research new buyer characteristics to define a target market underperformed random names.

Basing targeting on the market research characteristics of new customers without comparing to the “world around you” will not identify the true characteristics of prospects that are most likely to buy. Knowing the characteristics of your customers can help in developing advertising and new cross-sell opportunities, but are not useful in identifying who to target as your next new consumer. Truly accurate targeting analytics is all about finding differences or characteristics that discriminates your customers from your marketing population.

In this example and in most real world cases, best new likely buyers to target for your next campaign come from different sets or combinations of characteristics than those encompassing the highest percentage of new buyers. The discrimination that we seek in using one characteristic at a time is greatly extended by database analytics, which combines multiple customer characteristics in a way that goes beyond intuitive understanding and finds those characteristics that separate your new customers from your marketing universe.

Here’s an example: let’s look closer at employment. While 50% of new customers are employed, we found in the random market 60% are employed. When you compare 50% of new customers identified as employed to the 60% of the market being employed, the reality of being employed is actually negative indicator of who has the best buyer characteristics. This is true for both of the other top variables. The random market is more also highly concentrated (than customers are) in those ages under 65 and those living in their house 3-5 years. Thus, each of these is also negative indicators for identifying highly likely to buy prospects and they are not useful for targeting new customers.


Too many marketers fall into the trap of surveying their new customers and then using those characteristics to define their target market. Figure #2 shows us that such an approach identified a target that is 22% less likely to buy than randomly selecting names. In part 2 of this series, we’ll discuss who marketing should be targeting and how to easily identify your best new customer prospects.