One of the keys to a successful B2B marketing database is an effective data hygiene strategy. Data hygiene, for our purposes, can be defined as the software and processes that are used to parse, standardize, validate, verify, match and load data into a database.
Get your data hygiene strategy right and your marketing database can be used to drive results for your acquisition, growth and retention campaigns. Get it wrong and your database will be loaded with duplicates and will contain contacts such as Mickey Mouse, Big Bird and Homer Simpson.
The SIGMA Marketing approach to data hygiene is implemented as an integrated component of marketing database solution and customer intelligence hub. As one of the leaders on the development team of our customer intelligence hub, I am able to share some of the observations and best practices from our data hygiene strategy.
Don’t re-invent the wheel
Build vs. buy is an old axiom when it comes to software decisions. Unfortunately, data hygiene solutions are not so black and white. The best data hygiene solutions blend third party software products with proprietary software development. Think of it as the blending of art and science.
Your data hygiene strategy will be most effective when you find the build vs. buy blend that is right for you. Don’t make the mistake of going too far in one direction or the other. Address standardization and verification are absolute requirements for a B2B marketing database. Select a vendor that makes use of the USPS® and the Canada Post databases for the processing of North American data files.
Get to know your data
One thing that I’ve learned over the years is that no two data sources are the same. Data quality and fill rates can vary from one source to the next. Don’t make assumptions about your data that will come back to haunt you. Whether the data is from a finance system or a sales force automation tool it should be inspected and analyzed before it is introduced to your database. Your system should be flexible enough to handle the intricacies of each data source.
One of the many value propositions of our customer intelligence hub is the ability to bring disparate data sets together and present them in a common view. This would not be possible without the appropriate data hygiene solution.
Embrace best practices
Technology is ever changing and improving. Review your own custom code and the methods used by your partners to ensure that they are using up-to-date technology. Is your vendor able to offer you roof top level geocoding? Is your phonetic algorithm up to date?
At SIGMA, we periodically review our technology and that of our partners against industry best practices and adjust to them as needed.
Use flexible match keys
Some database implementations develop one or two standard match keys and use them for all of the major decisions regarding the de-duping and matching within the database. The flaw with this approach can be found in the makeup of the match keys themselves. Match keys rely on the presence of data to work properly. For instance, if your standard match keys require an address and your source data does not contain address information, your match keys are essentially useless. This will almost certainly result in duplicates in your database.
The marketing database technology inside our CI Hub allows for an unlimited number of match keys that can be configured to work based on the presence (or lack of) different data elements. Since these match keys are driven by metadata they are easily configurable for each implementation and can adapt to the specific data needs of a client.
Iterate, iterate, iterate
When testing your data hygiene solution, iterations and the resulting adjustments are extremely important. As I mentioned earlier, a data hygiene solution can be seen as a blending of art and science. The more often you are able to review your results and adjust your processes the better. Remember when testing to use a data set that will test for the exceptions. If you use “clean” data for your testing, your processes will not get any better no matter how many iterations you perform.