Can AI provide the answer?
August 1st 2019
- by: The Sogeti Digital Assurance & Testing Practice
This is a question every IT organization is faced with and there are not always simple answers. However, one thing remains true: When the cost of the testing itself outweighs the cost of the risk of releasing with potential untested items, the time to stop testing might have been already surpassed. It might be easy to see the cost of testing in terms of effort and budget used for all testing activities, but how can you know when the risks are decreasing below the cost of testing effort?
There is a saying that “measuring is knowing”. But then one needs to be measuring the right things with the right measures. Defining Test Metrics that are useful and gathering these metrics in dashboards that are finely tuned to pick up on realtime testing results and product quality data can be a source of valuable information in such decisions. Of course, the results of Testing and the data this generates are not the only measure to consider when deciding to release but it can be a very important dimension. The data is the key.
With the advent of AI guided Testing where the correct algorithms are chosen to provide the right dashboard view on the test data, steps can be taken to harness that data to even make predictive dashboards which can help in defect analysis, test case design choices and in looking at the risks a system has in relation to the test effort spent to test it. Cognitive QA from Sogeti is looking into the ways to leverage this data to make decisions and even “predictions” for such testing questions like “When should we stop testing for a particular release” ? more possible based on choosing the right data with regards to priorities for the business and combined with the quality results for those relevant features combined with the effort needed to test them as well as many other factors that bear upon such a decision.
There is no such thing as perfection perhaps but again – the data is the key. When it comes to such decisions there is always the right key to fit but then you need to know for each client context this differs. So, for your context you might want to ask, “When do we justify stopping testing and release?” And part of the answer lies in choosing to get the right data along with solutions like Cognitive QA. So, What is your lock? Maybe Cognitive QA can help provide a Key to unlock it.