In my last article, I described how to create a strong data foundation for your organization. As a brief review, I discussed the importance of clean data and setting up processes to ensure that the information that you capture about your members is standardized and consistent. In this article, I will describe how you can utilize clean and standardized data to improve your direct marketing efforts.
Clean Data to Improve Direct Marketing
Most of you probably utilize direct marketing in order to find potential members for your credit union. In a typical process you would work with a data provider to select the types of prospective members that you think would be interested in your product. Through trial and error you might make changes in your selection criteria to see if you can get any improvement in response rates over time. This is a time-consuming and laborious process, which sometimes works, but most of the time only provides small incremental improvement.
As people respond to your solicitation, you capture their information-but do you know which particular solicitation they came from (this might be difficult to identify if you do a lot of direct marketing)? Do you know which change you might have made in selection, creative, etc. might have driven this member to respond? Are you tracking your return on your marketing investment by data source?
In many cases, companies view the direct marketing process as two distinct events. First is the marketing event, creating your marketing piece. Then you send out the solicitation and wait. Like magic, calls start coming in. In the second event, you capture the customer information. But there is one problem. All the information you carefully used to select your customers is not available to capture when the customer responds. If you set up any tests with your creative or add copy you may not know which version drove a particular response. You might be able to produce aggregate level reports, but you cannot tell what particular change might have influenced an individual to respond to your offer.
There are two ways to mitigate this problem; setting up source codes and matching mail to responder files. The first method utilizes a tracking code that you place on the marketing piece. When the customer responds, they provide this number back to you. It has advantages in providing a real-time view of your campaign and an indication of which particular tests (if you set up any) are working best. The main drawback of this method is that it only provides a segmented view and does not provide information at an individual level. The second method involves matching the solicitation file to the responder file. This method is considered the best way to fully analyze a particular marketing campaign because it matches information available in the mail process to the responder information at an individual level. With this individual level matching, you can produce any type of aggregated level reporting including the reporting available using source codes. It also provides the data necessary for doing advanced segmentation and statistical analysis. The main drawback to this method is that it requires someone with a strong knowledge of data processing who can perform file matches. It is also more time consuming to match files and produce reports on a regular basis.
Rule No. 3: A critical part of improving your marketing efforts is the establishment of tracking methodologies. At a minimum, these methodologies need to allow tracking of any tests performed as well as determining the return on the marketing dollars spent.
A Closer Look at Source Codes
Think about the most important things you would like to know about any particular marketing campaign. Some examples might be:
* Data Provider: Which vendor did I use to obtain this name to mail?
* State: What state did I send this offer to?
* FICO Band: Is this a sub-prime, near-prime, or prime responder?
* Creative and copy: Which combination of creative and copy did I mail to this person?
Now you need to construct your source code. To be the most useful, the source code should be constructed in a way that each portion of the code has positional intelligence. Using the above data elements we would construct a source code that contained information about the Data Provider, State, FICO Band, and Creative/Copy. This could be done as follows:
* Data Provider: In our example, we don't expect to ever have more than nine data providers. Today we have three. The first digit in our source code will be for the data provider and have the values: Provider 1=1. Provider 2=2. Provider 3=3.
* State: Today this hypothetical company only operates in three states. However we can anticipate that they will eventually operate in more than nine states, so we need two digits per state. In our source code digits 2 and 3 represent states for this company as follows: California=01. Nevada=02. Oregon=03.
* FICO Band: We want to track three values, Sub-Prime, Near-Prime and Prime. The forth digit of the source code would take the following values: Sub-Prime = S. Near-Prime = N. Prime = P.
* Creative/Copy: One could contemplate many different creative/copy combinations for a particular marketing campaign. For this example only two are represented, but two digits are used for the source code in case more creative/copy testing would be done in the future. The fifth and sixth digit of the source code would have the following values Creative/copy test 1 = 01. Creative/copy test 2 = 02.
Reporting Becomes Easy
Once you have constructed your intelligent source code, reporting becomes easy. It also becomes easier to identify information about your customer real-time. Again, this is best illustrated by example:
Let's say a customer calls in. After capturing her data you ask for her solicitation number. She gives you the following number: 202N02. You know right way that her name came from data provider 2, she lives in Nevada, she is a near-prime customer, and she responded to creative/copy test No. 2. The fact that she is near prime would automatically suggest certain products to offer to her.
Because the source code was set up with positional intelligence you can easily set up reporting by looking at only part of the source code.
Rule No. 3A: For source codes to be a useful tool, they need to be set up carefully anticipating the reporting needs of the business. They also need to be set up with positional intelligence.
Matching Files: Mailed to Responder
To really take things to the next level, you need to have the ability to match files. This is the reason why you need a solid data foundation, because now we are going to build some framework. Building this framework does require some data expertise which you might consider either bringing in-house or contracting out. However, I will outline some of the key components for the file matching process.
The key advantage to this matching is that all information available about a responder can be reviewed at an individual level.
Unlike using the source codes, you do not have to predetermine what data you want to track. To match two files together requires a match key, which must adhere to the following: The key must be unique for an individual, and it must be present on both the mailed and responder files and have the same formats.
There are many software packages and vendors for this, but you can create a simple match key by combining elements of name, address, and SSN if available. One example of a simple match key would be: Last name, first letter of first name, Zip Code.
Once the key is created, both the mailed and response files would be sorted by this key and matched together using a variety of software packages.
Rule No. 3B: Matching files requires the use of a match key that is unique for each individual and is present in the same format on all files to be matched.
You've built Your Framework!
There are now all kinds of things you can do with this framework to add value to your business. I will describe some of these in future articles. It is important for you or your vendor to save the files that are created for your direct mail programs. You should also save your customer information. Consider the following for retention:
* For mail files, try to save at least 6 months worth of campaigns. If feasible, consider saving these files for up to one year. This will give you the ability to examine trends (for example seasonality) and provide the capability to do segmentation and data mining (discussed in a future article).
* Save your customer information for as long as feasible. This is especially important in industries with long cycles between purchases, such as real estate.
Robert Zelickson is co-founder of Zelcom Group. He can be reached at www.zelcomgroup.com.