Any data warehouse is developed and managed on a set of business rules that are built into ETL jobs to process data. The data integrity depends largely upon conformity to these rules. While there are exceptions to rules, thresholds are created to establish the degree of acceptable error. These thresholds should be monitored regularly by automated integrity checks as part of a complete quality assurance process.
Database marketing applications are largely at the mercy of these database side quality assurance processes. Many read database tables at face value and depend upon the integrity of the data, rendering these tools vulnerable to data issues. However, IBM’s Unica Campaign can be used as a kind of last line of defense. While the tool itself cannot repair data issues, it can proactively capture and mitigate these issues. Below is a list of business cases and steps to alleviate the impact of potential data issues.
Business Cases for Campaign QA Checks
Business Scenario #1: Consider a business that markets educational programs to students. The process to acquire student contacts includes the submittal of teacher nominations. Database ETL jobs that process these nominations files, create separate contacts for student and teacher, and loads a series of tables used to associate these contacts. Each contact is also associated with a school and home address as received on the nomination and other data sources. Marketers then use this information to contact these students and include the nominator’s information in the messaging. In this business scenario, it is vital that the ETL processes responsible for loading the tables within the campaign data mart is trustworthy and that the matching processes are reliable. A data discrepancy, such as a scholar-teacher misassociation, negatively impacts the success rate of marketing initiatives. The implications of this faulty relationship data issue are disastrous because of the business use of this data element in marketing materials. In this example, the development of quality assurance processes with the campaign application tool would have captured the data issue.
Every business has its own set of marketing rules. In the previous business case, the scholar and teacher must have a valid association to appropriately market to the target audience. For every business rule, it is strongly encouraged to incorporate high adverse impact ‘rule breakers’. If a scholar and teacher must have a valid association then any relationship with a location distance outside of an acceptable range should be deemed suspect when comparing their respective address information. This check can be done using a single Select process box within IBM’s Unica Campaign. Longitudinal and latitudinal data for scholar and teacher can be leveraged to calculate the distance between the locations and any cases with a distance outside an acceptable range can be suppressed and investigated further. Rule breakers such as this should be a standard suppression process in campaigns.
Business Scenario #2: Another business case for this type of campaign side integrity check relates to purchased lists. Consider the scenario of a retail company preparing the launch of a new store grand opening. A lack of existing customers within the surrounding area of the store location requires the retailer to purchase contacts from a list vendor. While companies that choose to purchase list normally conduct a series of validations it is often limited to matching existing contacts to list contacts, ensuring only those that are new to the database are purchased. However, in the case of this business scenario, it would benefit the retailer to guarantee the list’s contact addresses are all within a reasonable distance of the new store location. To execute a campaign for this type of promotion, qualifying records based on location is a crucial marketing requirement. Naturally, this check should be managed on the database side as the contacts are loaded. However, businesses sometimes operate on a limited timeframe where time to market is important. In this case, it is not uncommon for a list to be imported directly into the campaign system. Regardless, dedicated quality assurance campaign process boxes should be a standard in list generated campaigns to capture rule breakers. An appropriate process check for this scenario would include a Segment process box that would group contacts by State. A campaign developer would then get a read of contact count by state and could immediately identify issues such as a count greater than 0 of contacts located on the east coast for a campaign promoting the grand opening of a store on the west coast.
Campaign QA: 4 Step Process
Both of the business scenarios above could substantially benefit from the addition of quality assurance processes within campaign. Campaign developers can follow four simple steps to ensure the data used for marketing purposes is accurate and appropriately serves campaign specific objectives. These steps are:
1. Develop a list of ‘Rule Breakers’
2. Create campaign logic to capture each type of rule breaker
3. Suppress rule breakers
4. Escalate issue to IT department (include suppressed contacts)
The first step is to develop a list of ‘Rule Breakers’. In order to create this list, a campaign developer must fully understand the business and the campaign requirements. These rule breakers would represent data instances that do not adhere to marketing critical business logic. When a list of rule breakers is created, develop logic within a campaign process box to represent each of them. These will serve as the actual QA campaign processes. (Helpful Tip: Name each box after the represented broken rule. i.e. ‘Contact > 50mi away’ or ‘Contact Count by State’). Use these process boxes as exclusions and suppress these instances of nonconformity to critical business rules. Once these rule breakers are suppressed the remaining contacts can continue downstream. The final step in the process is to analyze each of the rule break instances. These excluded contacts may represent valid instances of non-conformity. For example, in business scenario #1, a teacher and student may have addresses that are a hefty distance apart, however research uncovers both are associated with a boarding school. In this case, the distance is acceptable. Unfortunately, when rule breakers are discovered, there are few that can be explained by valid exceptions. Most often, these quality assurance campaign checks uncover holes in ETL logic that only worsen with time. This is more common when new database processes and job are introduced to the system.
The Campaign QA Process is a simple way for any campaign developer to proactively guard against the adverse impacts of looming data issues and to take responsibility for data integrity.
Feel free to contact me directly with any questions. I always welcome the opportunity to expand my network of CRM enthusiasts!- Candice M. Narvaez
No comments:
Post a Comment