 |
Click for a Printer-friendly Version
- Adobe PDF
Evaluating Merge/Purge Systems: Part Five
By Jim Wheaton and Cynthia Baughan Wheaton
Principals, Wheaton Group
Original version of an article appeared in the November
1987 issue of "Direct Magazine"
[Note: Despite dramatic increases in raw computing power
and a proliferation of end-user software tools since the publication
of this series of six articles, virtually all of the content remains
highly relevant. The occasional obsolete point is highlighted.]
Statement of Purpose
In a series of six articles, we explain
a number of the key concepts that mailers should understand about
merge/purge, as well as reviewing (in the first article) a methodology
that could be helpful in evaluating the effectiveness of either
present or prospective merge/purge systems. While our comments
are primarily addressed to mailers, merge/purge vendors can benefit
by measuring themselves against the criteria that we have identified
as important.
Our objective is to describe new and specific tools that can be
used to evaluate and improve the performance of the merge/purge
process. Through commentary and examples, we will attempt
to translate into layman's terms the technical jargon that baffles
many mailers. In the process, practical applications should
become apparent.
This month's article, Part Five, focuses on run-time, ZIP correction,
output reports, vendor service, and client responsibilities.
Run-Time: A Hidden Cost
- An important issue for large volume mailers is merge/purge
run-time, which can impact turnaround on major jobs. The
best software in the world is of limited value if deadlines cannot
be met. It is important to recognize, however, that there
is often a trade-off between processing speed and quality.
- Some of the least sophisticated systems on the market, those
that rely on match-codes, are also some of the swiftest.
This is because they use simple matching procedures as discussed
in the second part of this series (Direct Marketing, July 1987).
If all other factors are equal, it simply takes less time to analyze
just part of a record.
- Some leading vendors have developed single-pass software for
jobs that have previously required two passes, which is often
the case with business-to-business unduplication. This obviously
cuts run-time significantly, and counterbalances the additional
time required by sophisticated systems to process the entire record.
Also, as you will see, the ZIP correction method that is chosen
can reduce run-time.
ZIP Correction
ZIP correction is another important element
of the merge/purge process. Software has been developed by
a number of vendors to ensure that the ZIP code is consistent with
the remaining elements of the address. When an inconsistency
is perceived to exist, corrective action is taken.
A problem with some ZIP correction software is a tendency to render
undeliverable a technically incorrect but still-deliverable address.
This happens when only the street, city and state elements of an
address are evaluated. For single-ZIP cities, however, only
the city and state, and not the street, are evaluated.
The following example shows how, in a single-ZIP situation, focusing
only on the city and state can lead to undeliverability:
- In record number one, Brad Kraft really lives in Sandy
Hill, Pennsylvania, not Sand Hill, Pennsylvania.
- Due to an input error, the "y" was left off the city name.
- Even though these two cities have similar names, the ZIP Codes
are quite different:
- Sandy Hill is 19401.
- Sand Hill is 17042.*
- Record #1 is technically an incorrect address because it contains
a misspelled city name. It is, however, deliverable.
This is because the postal carrier within ZIP 19401 is familiar
with the Brad Kraft household. He may even know him personally,
and have been delivering his mail for years. The postal
carrier will get that package to Brad Kraft!
- ZIP correction software, however, does not know Brad Kraft.
All it knows is to compare the city name in the record to a look-up
table of valid cities within Pennsylvania. When a match
is found, the software will ensure that the record contains the
corresponding ZIP Code. The implicit assumption is that
the city in the record, not the ZIP, is correct.
- In our example, the software will change the ZIP to 17042,
the one that corresponds to Sand Hill, as we can see in record
number two. Because no one in Sand Hill has ever heard of
Brad Kraft, he will never receive his package. The record
is now undeliverable.
This is how some ZIP correction software can take a technically
incorrect, but deliverable, record and render it undeliverable.
Just to give you some perspective, the following are other examples
of similar-sounding single ZIP cities with very different ZIP Codes.
All are within Pennsylvania and begin with the letter "S":
We are not suggesting that slight input errors would cause all
of these to be incorrectly handled by some of the ZIP correction
software that is currently available. Instead, they are meant
to give you some perspective of the potential magnitude of the problem.
ZIP Correction Methodologies
[Subsequent, August 5, 2003
comment: Due to changes in U.S.P.S regulations and corresponding
standardization of industry practices, this section is obsolete.]
There are two basic ZIP correction methods, both of which are
performed in the edit process. The primary difference between
the two is the percentage of total names that are ZIP corrected.
The first method is to pass all incoming records through ZIP correction.
- Proponents contend that this increases the number of matches
in the unduplication phase.
- However, care must be taken to correctly handle a deliverable
address that contains an incorrect city, as in the Brad Kraft
example above.
The second method is to ZIP correct only the input records that
cannot be Carrier Route coded, since Carrier Coding virtually ensures
that an address will be deliverable. With this method, approximately
10 percent of total names will undergo ZIP correction.
- Carrier Route coding examines the street and ZIP, but does
not force changes that can result in undeliverability.
- ZIP correction run-time can be reduced by as much as 90 percent,
and costs are lowered proportionately.
- However, match rates may be adversely affected by the retention
of incorrect city names.
Because our study did not involve a quantitative analysis of these
two ZIP correction methods, we are unable to recommend one more
than the other. You should be familiar with the two, however,
and discuss them with your vendor.
Reporting
Valuable, but frequently overlooked, byproducts
of the merge/purge process are the summary and control reports.
Analysis of these reports can result in valuable marketing insights.
Reporting quality varies significantly among vendors, ranging
from handwritten notes on computer paper (which we actually received),
to concise and understandable representations of all the key processes
that have transpired (which we also received).
- The best reports are those that are written with the user in
mind and do not require a great deal of "translation" by the vendor.
- Always review the standard reports of any vendor under evaluation
or currently in use. It is important to be able to understand
how each field is defined, as well as how different fields relate
to each other. A vendor should not inundate you with partially
organized data but rather should put that data together in a meaningful
fashion. You should be able to clearly account for each
name eliminated in both the edit and unduplication phases of the
merge/purge.
- It might be helpful to design your own "ideal" documents, at
least in very rough form, in order to evaluate how well a vendor's
standard report meet your needs.
- Be sure to include a review of the format of output reports,
such as the duplicate and final output listings. These are
the documents that need to be periodically reviewed on a "micro"
level in order to assist in monitoring the appropriateness of
the parameters applied to the mailing. Some vendors' reports
make it easier to spot overkill and underkill than others.
Information requirements can vary significantly for different mailers.
- They may be tied to senior management reporting needs, or to
the level of maturity of the direct marketing venture.
- Special reports are sometimes necessary. Some vendors
offer flexible reporting software at a reasonable cost, while
others can produce special reports only at a significant expense
to the client. You have to decide how much you are willing
to spend for the flexibility you require.
Vendor Service
Merge/purge is a complex process. Vendor
service can be a critical factor in maximizing the quality of names
available for mailing. The best vendors attempt to build a
close, long-term relationship with the client.
- Time should be spent by the vendor in translating technical
jargon into English and defining the marketing needs of the client
in systems terms.
- Parameters must be fine-tuned by the vendor over time, with
approval from the mailer, as new needs evolve.
- Jobs need to be turned around on a timely and predictable basis.
These are all criteria you should consider in evaluating a vendor.
Client Responsibilities
To be fair, the vendor is not the
only party with responsibility in this process. To maximize
the effectiveness of the merge/purge, each company should designate
an employee to be responsible for the details of execution:
- The employee would need to understand the total range of vendor
capabilities and act as the primary vendor contact. This
will minimize contradictory and poorly directed communications.
- This person would also provide comprehensive, timely, written
instructions (we cannot emphasize that enough!) for each job,
as well as analyze merge/purge output for the four types of errors
reviewed earlier.
With the client and vendor working together, the merge/purge process
can become a very powerful business tool.
Jim Wheaton and Cynthia Baughan Wheaton are Principals at Wheaton
Group, and can be reached at 919-969-8859 or jim.wheaton@wheatongroup.com.
The firm specializes in direct marketing consulting and data mining,
data quality assessment and assurance, and the delivery of cost-effective
data warehouses and marts. Jim is also a Co-Founder of Data
University www.datauniversity.org.
Top >>
|
|
