Data appending: Dive deeper beneath match rates
The need to append a prospect or customer data file with third-party data is a necessity for most intelligence-driven marketers. Missing data, stale or inactive files, data kept in silos within enterprises and inaccurate data entry are all sins that can creep into the database and wear down the success of prospect and customer engagement initiatives. Depending on the application, such data mishaps can even undermine a brand.
One of the questions asked most often in the client-data vendor dialogue is: “What is a typical match rate on a given data append effort?” However, answering this question is like answering the age-old question about “average” response rates. We all know there are too many variables in price, product, promotion and offer to give an answer that is anything more than educated guesswork.
When appending data to a customer file – whether e-mail, mobile, telephone, demographic, behavioral or, in business-to-business, firmographic - match rates run the gamut. No matter how established and sophisticated a data-driven marketer, these rates can vary significantly – both because of the marketer's existing data, and because of the vendor's various data sources.
Foremost, there must be a conversation between client and vendor about data confidence levels. Match rates are meaningless unless there is corresponding knowledge about data confidence. How confident are we that the appended data are accurate? Confidence is about granularity: Does the data record match to name only, to name and address, to name and ZIP Code or to multiple “touch points” across media channels?
Second, how timely are the data sources? Are data sources updated within the past day, month or year – or perhaps in real time?
Third, any data vendor worth its salt will enable testing regimens, not just to determine match rates, but to determine match rates at various confidence levels. Some marketing programs can get by with looser matching business rules; others require a strict matching regime. Either way, a data vendor should be flexible to enable such dedication to testing to prove their data sources work for the client in advance. Clients, too, shouldn't settle for less, and should always be prepared to test.
Testing applies to digital marketing applications, just as it does for traditional media, and to both consumer and business-to-business markets. When comparing one vendor's match rates to another, it is vital to know if “apples are being compared to apples.” One match rate may “beat” another – but if the data confidence levels behind matching rules are different, then the metric comparison is pointless. Testing enables the best outcome for the client.
Finally, there may be data variations on the client side. Some companies have customers who turnover quickly. Some update their data in real-time, while others do it on a pre-set schedule. Some companies have data that enter the enterprise via channel partners. Some have already pulled aside selected data into a distinct data mart. So only a segment of the database is being analyzed.
Data appending requires a deep-dive conversation between database manager and vendor, as well as dedication to a periodic testing regimen, to determine if those highly-sought-for “data matches” are performing up to par.