What is our primary use case?
Creating models for requirements definition and alignment, starting with high-level requirements definition and expanding models into detailed requirements to test models (manual testing). Also, creating models and analyzing and selecting appropriate test scope/coverage.
How has it helped my organization?
Using the test modeling approach helped
- get early alignment on requirements
- prevent design and implementation defects in earlier development phases if models were used from the start to get everyone’s alignment
- improve test coverage (with manual and automated testing).
In terms of meeting business challenges, it helped to shorten the dev/testing cycle by identifying requirements gaps early in the process, by having models shared within the development team. It helped increase test coverage and reduce the number of issues experienced by clients/customers. It made test design & writing process more efficient.
What is most valuable?
The ability to create models/diagrams at multiple levels (nest/embed them) helps in taking models from high-level business requirements and building them into detailed requirements models and test models. Plus, it helps reuse lower level models. It also allows maintaining models at appropriate levels, even for very complex systems/solutions.
The above features also help to avoid redundancy and use appropriate level models for different levels/phases of testing and/or communicating to different audiences.
Integration with TDM, test data management tool, provides the ability to generate data or use identified (preset or parametrized) test data. It allows significant expansion of test coverage and flexibility, without creating new tests and needing to maintain them.
Integration with automation is one of the reasons we started to consider moving to this tool from our original tool for implementing test modeling. ARD appears to have better integration with Selenium. It also has the ability to record scripts/flows using Selenium Builder and import them into ARD, which will then create and optimize a model based on that.
Measuring test coverage helps in one of the most challenging tasks. It has logic that can help to select the right set of scenarios and know what coverage it will provide.
Handling loops in models was one of the challenges that we came across in our previous tool. ARD seems to have better control and logic over handling loops in models.
What needs improvement?
Aligning models so they look clear and readable without having to move boxes around.
Saving different subsets of test cases, use cases, etc., of a given diagram, for different purposes and providing an easy way to name those subsets.
For how long have I used the solution?
What do I think about the stability of the solution?
Did not use in largescale capacity so I can't provide a sufficient evaluation.
What do I think about the scalability of the solution?
Did not use in largescale capacity so I can't provide a sufficient evaluation. However, when working with very large models (producing 250,000 scenarios), the ARD performed well and was able to provide meaningful options for test selection in short response time
How are customer service and technical support?
Did not use actual technical support, but pre-sales engineering support has been outstanding.
Which solution did I use previously and why did I switch?
We used another test modeling solution. ARD has better integration with automated test scripting and with test data management. We also switched to it because we were looking for ways to make testing more effective and more efficient; reduce defect leakage.
How was the initial setup?
For the standalone option it was easy, since we are familiar with modeling approaches, in general. For the repository option and connection to other tools, it required more work (as any integration would).
What's my experience with pricing, setup cost, and licensing?
Recommendation is to go with concurrent licenses as oppose to seat license; this gives more flexibility.
Which other solutions did I evaluate?
ADPART from Cognizant and Eggplant AI, part of Eggplant Digital Automation Intelligence suite from TestPlant.
What other advice do I have?
It is important to ensure that people first understand modeling concepts and the organization's defined modeling best practices before jumping into tool implementation.
It is critical to get management support, as for some people test modeling is a very new way of thinking and they would find many reasons to push back.
Check CA's videos on modeling available on CA's Youtube channel, before starting to implement the tool itself.
Which version of this solution are you currently using?