A more complex example for MDM ROI (part eight)

Product Director, Master Data Governance

In my last post, I discussed a relatively simple business case scenario where an enterprise intends to implement MDM because their current process heavily relies on a third party supplier of customer data.

Typically, though, things aren’t quite that simple.

Today, let’s discuss a more complex MDM business case for marketing applications typically addressed by analytical MDM. A company performs a few direct mail campaigns with a total of 10,000,000 mailings annually. The company determined that an estimated 10 percent of mailings are sent with at least one of the following errors:

  • The mailing address is incorrect and, as a result, the mailing does not reach the recipient
  • More than one letter is sent to each recipient
  • More than one letter is sent to the same household because the enterprise lacks capabilities to identify households even though the business would like to be able to market to households rather than to individuals

The company decides that if it implements MDM, the 10 percent error rate will be practically eliminated. This yields estimated savings of $500k annually, assuming each mailing costs 50 cents. In this scenario, using the MDM costs from the previous example, the company will save only $100k annually: $500k savings less the $400k spent on the ongoing expenditures.

In order to offset the initial MDM investment ($3M) and break even, the company will need 30 years! This is too long, especially considering the drop in the real value of a dollar over this period of time.

This estimate cannot justify the MDM investment. It does not mean, however, that the enterprise has considered all benefits that can be realized from the MDM implementation.

For instance the marketing department considering an MDM investment may want to develop an advanced market segmentation model leveraging social media information, which requires MDM as a component of the solution to bring together and cross-reference the data from multiple sources. MDM integrates customer information from multiple systems and possibly third party sources and also creates and maintains one view of a customer or household.

MDM enables marketing segmentation software that cannot operate efficiently without MDM.

This scenario highlights some of the challenges typical for business case estimations:

  1. The company estimated 10 percent error rate in mail delivery. In real life it is a challenge to determine the error rate. A 10 percent rate is most likely a guess. The real rate can be significantly higher or lower.
  2. If market segmentation is one of the applications planned for implementation, the invested amount should be calculated for the whole market segmentation project, with MDM being its component. Hence the baseline project cost will be higher than $3M and the annual cost will be higher than $400k.
  3. The total of the benefits should be calculated as a sum of the mailing savings and the savings and revenue opportunities associated with the new market segmentation process.
  4. A number of books and numerous articles discuss metrics estimating efficiency of marketing campaigns. These metrics require accurate measurements of responses to marketing campaigns, which is not simple.

Many companies do not measure the efficiency of marketing campaigns, which complicates the quantification of the MDM business case, since the impact of MDM cannot be measured in business terms. If the business itself does not measure its efficiency, it is difficult to estimate the business impact of MDM.

However, in light of both of these scenarios, the business case is not always so simple. MDM often cannot be justified based solely on projected cost savings from a single project or team; rather, those building a business case must also be able to address the effect on resources and potential new revenue opportunities.

Next week, we’ll go beyond cost savings and look at other ways to justify MDM.

Catch up on the entire series so far:

Follow @IBMbigdata

Follow @IBManalytics