Debunk 6 Direct Marketing Myths

Introduction
Achieve high-precision marketing with analytics
Today, direct marketing is as much of a science as it is an art. While marketers have been using technology to increase
campaign effectiveness, past options have been difficult to use, time-intensive and expensive. Because there haven’t
been cost-effective and easy-to-use direct marketing tools, marketers often focused on optimizing one or two aspects
of a campaign – such as list selection or creative testing – rather than the entire campaign.
Like many direct marketers, you may be using some sort of software to maximize the value of your customer and prospect data. But your challenge is finding software that meets your specific needs so that all aspects of your campaigns demonstrate a
quantifiable, positive impact on your organization. Spreadsheets, for example, are easily available, but offer simple analyses
and a very high propensity for errors1 – errors that can affect results and put your organization at risk. Database programs are flexible and built to handle large quantities of customer data, yet they can be difficult to use without advanced knowledge. And statistical packages, while powerful, typically require a deep background in statistical analysis to avoid making errors when
interpreting the results. Furthermore, very few statistical tools are built with the “point-and-click” ease of use that a busy
marketer needs.
This paper focuses on six myths that surround direct marketing best practices. It also discusses how your organization can use specific analytical techniques and tools to bust these myths and maximize your bottom line – so you won’t just survive, you’ll thrive.
Myth #1
Simple segmentation tells everything you need to know about customers and prospects
As a direct marketer, it is critical to your success to understand the characteristics of your customers and prospects. One way to do this is by pinpointing which factors, such as age, marital status, job function and location, identify and differentiate groups or “clusters” of prospects or customers. After considering their identifying characteristics, certain “personas” can be assigned to each cluster.
Cluster analysis, however, takes time and expertise. If you don’t have an analyst on staff, you may rely on simple filtering
or segmentation instead. Based on that simpler analysis, you may feel you already know which characteristics identify your typical prospects and customers, and yet …
Myth buster #1: A deeper dive can find surprisingly profitable new clusters and personas.
To take your campaigns to the next level, think about going beyond the basic identifying characteristics of age, marital
status and other obvious characteristics and relationships. True clustering requires more than “eyeballing” your data. Instead, consider using a software solution to help reveal factors that identify hidden segments that may be used to create predictive personas. These hidden personas remain invisible until you run a true cluster analysis of your current database, and they can be the key to new sales and new customers.

For example, the owner of an automobile repair shop wants to learn if there are any specific clusters among the car owners who regularly schedule oil changes. Running a true cluster analysis provides insight into a new customer persona: stay-at-home parents who reside within two miles of the shop are more likely to schedule appointments for the middle of the week and are also likely to purchase premium products and services. As a result, the shop owner is able to customize a set of direct marketing offers to appeal to this particular cluster, whom he might name “Close-by Parents with Money,” and plan future campaigns to include this high-value group.
Myth #2
A high RFM score is all a marketer needs to identify the best prospects for a campaign
You may already be using recency, frequency and monetary value (RFM) analysis on your transactional data to classify your customers according to those who have purchased most recently, those who have purchased most frequently and those who have spent the most.
The reasoning behind RFM analysis is simple: the more recently or frequently someone makes a purchase, or the more they spend, the more likely they are to purchase again. By easily identifying the customers who are most likely to respond to your campaigns, you can decrease creative costs, trim your mailing lists and stay within budget. You can also ensure the loyalty of your best customers by creating promotions that anticipate their needs. However …
Myth buster #2: A high RFM score by itself may not always help you target the right
customer for a specific campaign.
RFM analysis should be implemented carefully with clearly stated goals that tie back to your business objectives. When using RFM analysis, customers are assigned a bin or grouping (for example, 1, 2, 3, 4 or 5) for each RFM parameter. The total of these provides a figure referred to as an RFM rank or score. Higher scores pinpoint the customers who have spent the most, most recently and most often, while lower scores identify customers who have spent the least, purchased rarely or made infrequent purchases. For example, an RFM score of “555” indicates a customer who has recently purchased from you, purchases from you regularly and has spent a lot with you. In contrast, an RFM score of “115” indicates a customer has spent a lot with you but does not do so with any recency or frequency.
RFM scores should be sorted and compared to response rates because the highest response rates to a particular campaign may NOT correlate to the highest RFM scores.
A large retail store, for instance, has departments with low-to-moderately priced merchandise, such as children’s clothing, as well as departments that sell appliances or furniture at a much higher price point. So a customer in bin “5” for both recency and monetary value might suggest a purchase history of a large number of lower-cost items or the recent purchase of a particular item with an above-average expense. Depending on the item and its cost, this customer may or may not be the best target for a campaign designed to promote high-priced items like plasma televisions or appliances.

Myth #3
When testing campaigns, the one with higher response rate is the winner
Sending a new campaign into production costs time and money. In order to launch the most efficient and effective campaigns, you probably test existing (control) campaigns against new (test) campaigns. Before deciding that the campaign with the highest response rate is the clear winner, you need to know that …
Myth buster #3: A high response rate doesn’t always guarantee a better campaign,
so statistical validity must be checked.
While your test campaign may yield a higher response rate, it doesn’t guarantee that it’s the winning campaign. When
comparing the response rates of your campaigns, check whether the higher response rate is statistically significant. For
example, if your control package receives a response rate of five percent and your test package has a response rate of seven
percent, the natural reaction is to declare the higher response rate the winning package. However, if a statistical analysis shows there is a significant probability that the higher response rate can be attributed to chance, then a re-test may be
called for before making any decision about replacing a control package.
By calculating the statistical significance of response rates, your team can make smarter decisions about which test campaign is truly the winner.
Myth #4
If a campaign produces a below-acceptable response, it is a losing campaign
After conducting a test campaign it’s helpful to identify prospect profiles that include the age, marital status, job functions and other identifying characteristics of the respondents, even if overall the campaign performed below expectation. With
prospect profiling (not to be confused with customer clustering), you can group potential respondents into profile tiers,
ranked from the highest response to the lowest response.
To truly improve campaign effectiveness, however, you need a technology that can help discriminate among the response rate for each tier versus the overall response rate. Only then can you understand which profiles to select to achieve your target response, break even or return on marketing investment (ROMI) because …
Myth buster #4: Even within a “failed” campaign that comes in below targets, there likely are
identifiable winning segments that can be profiled for future successful campaigns.
For example, an independent bookseller conducts a campaign to her customer list. Overall, the list comes in below break even. However, after running a prospect profile, the bookseller discovers two profiles – women over age 50 who bought a hardback in the past 12 months and men under age 40 who bought “How To” books – that perform well above the rest. Armed with those profiles, the bookseller can rent lists with similar profiles and target customers with those profiles for future campaigns to improve ROMI. The results could include increased response rates, decreased costs in marketing to current customers
and improved targeting for acquiring new customers.
Both customer clustering and prospect profiling play a key role in list selection. Once you know which characteristics predict the best responses, you can contact a list broker and provide the profiles of responsive customers. In turn, the list broker can compile a mailing list of customers and prospects with similar characteristics – increasing the likelihood that they will respond to your organization’s campaign.

Myth #5
Once the obvious characteristics are identified, a propensity to purchase score can be locked in
Marketers must decide which of their campaigns and offers will provide the highest ROMI, and this is often accomplished through propensity to purchase scoring. This enables you to analyze the attributes of those who have already responded
to a test mailing and quickly identify who is most likely to purchase. Once you have a prediction of which customers are most likely to respond to your offer, you can eliminate the customers who are least likely to respond from your mailing lists and
stay within your budget. However, it is essential that the method used for scoring be validated. Finding attributes that are
predictive can require trial and error because the factors that are predictive may not be obvious. As a result …
Myth buster #5: Propensity to purchase scores can and should change based on
the predictor variables that were used to create the score; therefore, they need
to be validated before being used for selections for campaigns.
Keep in mind that propensity to purchase scores are statistical predictions based on the input variables you select. For
example, propensity scores based only on a few selected variables (for example, age, income, number of transactions) will likely change if you introduce additional and potentially categorical variables (for example, gender or retired). Be certain to
use a technology that has the capability to validate the quality of the model that produces the score.
Myth #6
Postal codes with the same response rate produce equal results
For mail campaigns, you may already use zip code analysis to help discover whether there is a geographic bias in campaign response rates; that is, if certain geographic areas perform better than others. When you do so, you may find different
geographies have the same raw response rate, and conclude that they are performing equivalently.
The catch is that all response rates are not created equal. In other words …
Myth buster #6: Three percent doesn’t always equal three percent.
In order to mail where it matters most, use technology that can determine when a response rate is statistically significant.
Consider, for instance, postal code A, which shows a response rate of three percent for 10,000 recipients. Compare this to postal code B, which has a response rate of three percent for 100 recipients. Although the response rates for both postal codes seem equal, the small number of recipients for postal code B likely makes its response rate statistically insignificant. When calculating response rates for each postal code, your software should weigh response rates based on the number of recipients – so you can select codes with statistically accurate response rates.
Once you’ve identified the geographic areas that respond well statistically to your campaigns, a versatile software program can also use prospect profiling to identify the characteristics of respondents in those postal codes. Then you can tailor future offers to match their interests.

What to look for in direct marketing technology
As a direct marketer, you need statistically powerful yet easy-to-use software solutions that can help maximize the value of your campaigns. You will also benefit from affordable solutions with quick and accurate “push-button” technologies.
When searching for software, consider selecting a vendor that understands the needs of direct marketers and offers the
analytic techniques mentioned earlier. In particular, the solution selected should enable you to:
■ Immediately produce clusters of contacts not obvious to the naked eye so you can assign certain personas to each cluster
■ Conduct an RFM analysis in just a few clicks that effortlessly lets you use transactional- or customer-level data
■ Easily run a control package test and view the statistical significance of response rates to make certain that
you know when a test package has truly outperformed the control
■ Use prospect profiling to group potential respondents into tiers and discriminate among the response rates
for each tier versus the overall response rate
■ Identify propensity to purchase in your customer base, validate the scoring and export the scored file
for campaign selection
■ Quickly perform postal code analyses that produce weighted response rates and determine when response rates
are statistically significant
Conclusion
Bust the marketing myths – make your marketing programs as precise and profitable as possible by selecting a software
solution that relies on the analytical tools mentioned above. With the right software solution, you can:
■ Cross-sell and up-sell more effectively to your current customers
■ Choose the best lists
■ More easily acquire new customers
■ Select the best offer for each group
■ Test your campaigns and choose the best one
■ Increase response rates for all of your campaigns

Advertisements
  1. No trackbacks yet.

What Do You Think? Comments Welcome.

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: