Did we ever hear of Sampling Testing? The word sampling refers sample data to be tested. Let’s see why we have to go for sampling testing. How it can be done? Our examples will be based on Insurance project.
Example of Insurance project considered in this blog – Agents are being paid with commission, bonus, points and allowance for the Life, Traditional, health, Medicare supplement policies they bring to Insurance organization.
Technology Stack: IBM Varicent ICM (Incentive compensation management) to calculate the commissions, bonus, points and allowance. Informatica to load the data into ICM system from Policy admin systems.
Scenario considered: Legacy system built ago 30 years (old system) migrated to new system comprising above technology stack (New system). Each functionality (Base commission, Agent production credit points, etc.) have been measured and compared between two systems. But certain functionalities (Field training allowance, Field support program,) have been enhanced and modified in new system and not comparable with old system. There are 15+ functionalities (called as components in IBM Varicent ICM) considered as not comparable to old system. Business wanted to test with large set of data to ensure the business scenarios coverage in new system. We have designated as sampling testing henceforth.
Importance of Sampling Testing:-
- Different variations in sampling data with multiple dimensions (events, payee types, payee hierarchy, policy and plan attributes) would give the evidence and confidence to business that current system is working fine
- More sampling data would increase the functionality and code coverage to max of 98%.
Sampling Testing Process:-
There are four steps in sampling testing and are listed as below. Table explains the activities involved in each step and responsible owner.
|S.no||Steps||Involved activities||Responsible Owner|
|1||Template Preparation (or)|
|Template Preparation : |
Business to prepare the template considering the inputs on policy, plan and premium and intermediate calculations
Template Enhancement :
Template enhancement could be required if exhaustive coverage isn’t done. Business can’t be expected to cover all functionalities or calculations in template. Post mismatch analysis between expected results from our calculations and actual results from system, template could need to be enhanced to fix the template issues.
|2||Query Creation / Modification||Query Creation :-|
Create query based on template needs. Data extraction query could be complex as testing team is trying to cover the coding by development team to fill the template. could run in range of 1K-2K LOC in SQL query.
Query need to refer the history, product configuration, data tables, val tables, result tables (result tables are final data tables in ICM Cognos product).
Query Modification :-
Query modification could be resulted due to template enhancement or calculation of expected results couldn’t be precise.
Tester SME with insurance domain knowledge.
|3||Data Extraction||Data extraction will be simple step subjected to template preparation and query is available with us. But this is repetitive step if extraction needs to happen on every code base to ensure all transactions and functionalities are working as expected||Business Analyst (or)|
|4||Mismatch Analysis||This is the complex step to drill down the variance between expected and actual values. This step requires to analyze the issues from ICM layer to ETL layer to old system or Policy admins||Tester SME with insurance domain knowledge|
In next blog , I will describe with examples for more clarity.
Shanmugam Lakshmanan(Shanny) is techno-functional candidate having diversified experience in Insurance and Retail sectors. Excellence in Program management, hands-on in open source testing tools – Selenium, Jmeter, etc. Most important quality for tester should have development knowledge in digital era.