0

I made a post in relation to this question on the SQA here but thought I would post on this board as it's more active.

I'm testing an underwriting engine with a front-end interface that asks a series of questions that either evaluate the input eligibility for a product, or prompt for further drill down questions which have the same purpose. These questions are in the form of boolean, select lists, checkboxes and free text with range validation applied.

I'm looking for some advice on the best way to test the system to ensure that it evaluates in the correct way each time. Obviously it's not feasible to exhaustively test all of the options, but I need to be able to verify that each question will evaluate correctly and return the expected result.

Unfortunately, there's no scope for automation so it is all going to be done manually.

  • To clarify, the output of the underwriting engine is just a boolean (isEligible), or does it produce more - e.g., prices? – Dan Pichelman Jun 8 '18 at 18:19
  • Hi Dan, sorry for not being clear. The output is boolean: IsEligible. – ahu77 Jun 8 '18 at 21:44
  • Why is there no scope for automation? Ideally it should have been written in a way that decouples the logic from the UI, so the correctness of the logic can be tested automatically... – Sean Burton Jul 6 '18 at 14:53
2

Last time I did this the insurance product had a pricing sheet that could be used manually to calculate the price.

It looked like a 2d array of percentage rates with the main criteria on the axes.

There were then various discount percentages to take into account for the other questions.

The test was simply to enumerate every possible combination and check the result manually.

There then ensued some 'discussions' about how the correct way combine multiple percentage discounts, followed by an audit of all the policies sold at a discount of over 100%

  • Thanks for the help Ewan. How many different combinations were there? In my case, it's more complicated as there are questions that are either boolean, range, multiple choice or single option from a list, which means there are 100,000+ potential combinations making it unfeasible to test every combination manually. – ahu77 Aug 2 '18 at 9:16
  • Yes mine was very similar, they tested manually with what they considered key values. which I then automated after a 'correct' answer was determined – Ewan Aug 2 '18 at 9:36
-2

The approach is very simple. Test it with all possible combinations and compare the results at the end if it is returning the correct answer or not.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.