When dealing with large volumes of data it is often difficult to identify trends and outliers within a dataset. MDM’s data profiling functionality allows data stewards to observe data in its current state and identify patterns and trends. A few examples of when a data steward would utilize data profiling include:
- Profiling and analyzing data
- Investigate a decline in quality of data as reported by data quality policies.
- A potential upstream problem may have been identified.
- Ensuring address quality of master data
- For a marketing campaign targeting a group of customers or specific geographical region.
- Increase in return mail or bounced emails.
- Running a search and reviewing the result of the search
- Determine number of resulting records from the search.
- Refer to examples of a search that returns a large subset of entity records (ex. more than 1 million).
- Refer to details of the various source systems that serve as input for organization records.
- List all data that came in from Sweden on a Saturday 2 weeks ago.
- Investigating what an existing collection contains
- Verify if an old collection can be deleted.
- Identify who created the collection. When? Why?
- Exporting a list of customers
- 1000 records with phone numbers for marketing research.
- 42 million records with address and anonymized data points for a BI platform.
Use Case 1: Ad Hoc Work on Lists of Records
Bruce is a data steward whose primary role is to support and ensure the quality of customer data for the ACME’s marketing team. The Marketing team is promoting the upcoming season and is running multiple campaigns which target existing customers residing in specific geographical areas within the United States. All campaign collateral us addressed to each individual customer and sent via both email and direct mail.
Existing data policy monitoring metrics are in place and show that a sizable percentage of marketing collateral is either returned by the United States Postal Service (USPS), emails have bounced back, or contact information is not present. In the last two campaign seasons the statistical breakdown is as follows:
- 9% of emails have bounced back
- 12% of direct mails were undeliverable and returned
Because the upcoming campaign is targeted towards individuals in a particular region of the US, the marketing team would like to understand the quality of contact data for those individuals before rolling out the campaign. As such, the team has compiled a mailing list comprised of customers and prospects they would like to mail collateral to.
Bruce’s task is to run this campaign list against the current MDM database to determine the quality of contact data currently on file. This determination can be made by determining if the following requirements are met:
- Customer has sufficient contact information
- Main Address is validated at least once within the last 365 days
- If address has a low quality, no flyer is sent
- Email is validated within the last 120 days
- If email is no longer valid, a process is started that notifies the customer and requests that they update their email address.
To accomplish this, Bruce can create a collection of customer records based on the mailing list and generate a data profile on the Collection. They then evaluate the data profile against the requirements.
Note: It is assumed that the profiling configuration Bruce uses includes the necessary data elements to be analyzed, i.e., Address and Email attributes.
With the data profiles in place, the MDM system can highlight the data quality problems for Bruce, allowing them to report the following back to the marketing team:
- 4% of the customers did not have mailing addresses.
- 18% of the addresses that are present are incomplete (i.e., missing state or zip code).
- 8% of the emails are marked by Experian as 'disposable', indicating the address is associated with a disposable email provider. Usually they are unreliable and could potentially be spam accounts.
- 4% of the emails are 'illegitimate', indicating it is highly likely they are spam accounts or inactive domains.
Based on this information, marketing may then determine the best course of action, i.e., to either proceed with the mailing (if the data quality is acceptable) or place a larger emphasis on an e-campaign if a large percentage of the target customers only have an email address.
Use Case 2: Investigate a Decrease in Completeness of Contact Information
Chris is a data steward whose primary task is to continuously monitor the quality of data within MDM. A focal data point of Chris’ responsibilities is to monitor overall quality of contact information. Data quality policies monitor contact information by utilizing a completeness metric which evaluates completeness of attributes such as Address, Email, Phone, Last Contact Date, etc.
In the last two weeks, Chris has noticed there has been an increase in frequency of policy breaches for quality of customer contact information. Since ACME’s marketing strategy includes heavy reliance upon informing existing customers of the latest products and promotions, having dependable contact data is a top priority.
Chris would like to use MDM to profile the records that contain incomplete contact information to determine what aspects within contact information are incomplete / missing. Once these customers are identified, Chris will then:
- Provide the results to the marketing team, so they may best determine how to proceed with their marketing activities.
- Collaborate with marketing in identifying and changing existing business processes that caused the capture of incomplete contact data so they may address the issue at the source.
By using the profiling capabilities of MDM, Chris was able to support the marketing team and their campaign efforts. In doing so, it was identified that ACME’s CRM system, which manages all customer loyalty programs and sign-ups, was lacking validations to ensure complete and accurate contact information is captured at the point of entry. Specifically, the CRM system experienced an issue where city and zip code information were no longer mandatory as part of the new customer sign-up process.
Furthermore, the marketing team was able to strategically accommodate for this in their marketing activities as they awaited resolution of the issue within the CRM system. Once the issue was addressed, marketing then proceeded to carry out their campaign with confidence in their data