When marketers decide to send it out for data hygiene, they will get exactly what they ask for-but what they ask for may not be what they wanted. A recent study offered a heads-up regarding the differences between marketers’ expectations and the services various data firms return under the rubric “data hygiene.”
Ruth Stevens, president of eMarketing Strategy, and Bernice Grossman, president of DMRS Group Inc., amassed a sample of 10,000 business customer records, each with 12 data fields. This file was sent to four different data hygiene vendors. In addition to the naturally occurring glitches, Grossman added a wide swath of errors to the file by hand, an exercise she described with a grimace as “character building.” All the records, which contained live data, were destroyed at the end of the project.
Not that she needed to strain herself. “Name” and “phone number” fields are often a mess, with the former often containing titles instead of names, the company name, a first name only, a prefix, such as “Ms.,” in the first name field, or out-and-out gibberish.
Phone number fields didn’t fare much better. These can include zeros, or alphabetic characters, or missing area codes, or a single number repeated 10 times. Grossman admits to adding decimal points to some fields, which were occasionally “rounded up” to the next highest – and quite incorrect – number.
The four participating vendors (Acxiom, DataFlux, Donnelley Marketing and Harte-Hanks) were given wide latitude in their instructions. They were asked to provide data hygiene as they defined it. In terms of purely correcting existing records, the four companies provided similarly. But there were wider swings in terms of which ones automatically included phone or e-mail appends without being asked, although all were capable of providing this information if requested.
Each of the participants was also asked to provide definitions of what data hygiene meant to them.
Acxiom said it was “purpose-driven data management practices and/or processes that promote data accuracy. Typically is applied to name and address data content, correction and completion.”
DataFlux deemed data hygiene “a five-phase data management cycle, including profiling (inspection), quality (correction), integration (merging and linking), augmentation (enhancement), and monitoring (auditing and control). Understand the data problems: Improve the data.”
Donnelley Marketing said it was “a broad range of processes that collectively deliver the highest deliverability of an address: standardizing, correcting, updating and verifying.”
And Harte-Hanks called it “the processs of solving business problems resulting from inadequate data quality: Accuracy, completeness, timeliness, validity.”
To Grossman and Stevens, hygiene is an objective process, involving only standardizing data a client sends out contact purposes. Processes such as de-duplication and appending are more subjective – as well as being separate processes from data hygiene, with separate pricing structures.
“People will be pissed, and they have no right to be, because they don’t know what to ask for,” Grossman said.
For example, a typical marketer may see that several B-to-B records are missing phone, fax and e-mail information and call his file a mess. It’s not. It’s incomplete, but not messy, according to Stevens. But consider a sales person who is trying to follow up on a sales lead. If the salesperson is givens the name of someone who is no longer with the target company, or an incorrect phone number, or a series of address fields that are garbled, then the definition of what makes a file a mess is very different.
As it happened, the four companies came up with similar results. Grossman stressed that the purpose of this exercise was not to determine which of the four participating vendors was the “best.” Rather, the purpose was to gain a picture as to what data service providers consider hygiene, and to offer a set of questions marketers could ask of themselves and of their vendors regarding their files.
So what steps can marketers take on their own? The first is enter data correctly in the to begin with. This includes training data entry personnel and paying them well. Customers should always be included to update their own data files. Any campaign involving first-class mail can use nixies as part of the updating process.
Finally, the most valuable accounts can be segmented and confirmed through an outbound process. While this last step is expensive, it will pay for itself if it is applied to the highest-valued customers, Stevens said.
Stevens and Grossman presented their findings at the National Center for Database Marketing conference in San Francisco, which runs through Wednesday.
Network
