Content area
Full text
While cynics may argue that it is in the interests of data owners to make money from duplicate records, the latest pre-merged datasets appear to contradict this
David ReedMachismo does not feature much in direct marketing. It is hard to be aggressive about email or competitive about prize- draws. Yet there are some aspects of direct marketing that allow for boasting.
In the data world, for example, you will hear rival suppliers claiming to have the best matching routines, or to remove more duplicates than anyone else. You may even hear examples of reverse macho, as a company shows you that it has the smallest file of residuals.
All of these skills are essential ways to keep the costs of direct mail down. To achieve the right level of targeting at sufficient volume, virtually every company will need to source data from multiple providers. That in turn demands that these files are merged and deduped - a cost which can still leave unmatched duplicates in a file.
It is easy to be cynical and assume that data owners are more- than-happy with this situation. After all, they get paid twice if a record is duplicated, and often receive a rental income whether the records are mailed or not. The suspicion arises, therefore, that they may deliberately leave known duplicates on file to bulk up the gross size of files going into the pot.
Could this possibly be the case, or is deduplication something that data owners fret about at night? And do the latest generation of pre-merged datasets and prospect pools deliver genuinely unique records with a lower overall level of dupes?
"All of the data we take in we check for internal duplicates and against each other. I can't think of a single file that didn't have to be considered for the number of internal duplicates. The number has never been low enough to ignore," says Barry Leeson-Earle, director of Tri-Direct.
"With any dataset, some people get mailed three or four times, not just twice," he adds.
But Leeson-Earle doubts that this situation is wilful on the part of data owners looking to bump up their fees. "What you do tend to get are errors which are genuine, such as wrong codes. It...