How do you handle uncertainty? If you’re a modern leader sitting in a highly digital organization, especially if you own the QA or software development processes at that org, your answer should include both more testing and more automation. The formula that you need is simple – more tests plus more automation in the process equals more complete testing – but never a certainty.
One of the most difficult puzzles posed by the modern software QA process comes up when we consider completeness. Essentially, if your software tests are complete, they are considered representative of all possible future scenarios your application may encounter. Software that has passed a battery of complete QA tests can be considered virtually 100% error and bug-free – and that is an accolade you’ll want to shout from the mountaintops when you earn it.
The problem arises when you try to balance the number of possible future scenarios out there with the number of employees available to configure, conduct, conclude, and analyze the results of their corresponding tests. And this is all without the non-negotiable next step in the testing process, which is to have developers go back and correct everything that was discovered. Man hours start to add up fast as more scenarios and test cases arise by the day, making this particular issue a real thorn in leaders’ sides.
AL/ML Software Testing
Artificial intelligence (AI) and machine learning (ML) applications help conduct software testing without pulling devs away from other, higher-priority projects to run and interpret tests. Since AI/ML models excel at managing extremely large amounts of data, they’re a natural fit for the QA testing process – when introduced, AI proves both faster and more accurate than human counterparts at managing testing and analyzing results.
AI systems can also take some initiative at high levels. What that means is less grunt work for software QA testers – in the future, if an AI in the QA testing workflow detects a need for it, the system can create new test cases, either on a fully or partially automated basis. The same goes for test maintenance, which AI-powered insights will help similarly speed up and automate.
When global B2B payment provider TreviPay, for example, initialized a very large cloud migration effort, leaders ended up realizing that many of their org’s internal development processes – including software testing – were far too manual, and should be optimized, or better, automated. As leadership quickly realized, “In an agile environment, we cannot continue to do regressions manually. It’s not sustainable.” TreviPay, having simplified as much as possible across its 25 complex financial applications, was highly conscious of time savings, requiring that every QA tester join the automation effort. To empower QA throughout the process, TreviPay reached out to AI-based UI testing firm Testim, who helped teams rapidly reskill and author the many tests needed to ‘go automatic’.
The results were extraordinary, reducing scripting time for new test cases by 50% — all while simultaneously reducing maintenance needs by 50%. That is thanks to Testim’s “smart locator” feature, which helps keep tests up to date after UI updates and revisions. “The amount of manual testing hours we saved by using Testim is incredible. And it’s not a one-time saving; we continue to leverage these efficiencies with every sprint and upgrade we are doing now.”
[RELATED: AI Regulations Are Coming. Should IT Leaders Be Worried?]
QAOps
QAOps is best explained as the practice of integrating QA directly into software delivery pipelines and the overall software development life cycle, primarily via automation and testing enhancements. Without automation, it’s arguable that QAOPs as we know it wouldn’t even exist!
What is QAOps? If you’re familiar with the methodologies that make up DevOps, you could think of QAOPs as DevOps + QA. In fact, per recent Capgemini research, 40% of DevOps teams now report that 30% of their development time is dedicated exclusively to software testing – and over half of the surveyed organizations state that testing should be conducted as early in the process as possible.
Why would an organization pursue QAOps? For one thing, since it’s estimated that around 85% of code defects are introduced in a project’s active coding phase, running QA processes earlier could help catch minor bugs as they’re written, long before subsequent updates help build them up into big, lurking software errors. QAOps doesn’t just yield less buggy code, though; when compared side by side with legacy software testing procedures, it ensures higher quality in the finished product while speeding up intervals between subsequent releases.
“I believe this is something that grows organically,” explains Dmitry Gorbunov, QA Automation Team Lead at Devexperts. “Shifting left and migrating to faster delivery models constantly sets newer, higher standards for test stability and performance. To avoid turning into a bottleneck and continue providing meaningful results, testing shall transform and start solving these new challenges.”
IoT Testing
The benefits of more automation are also apparent in QA testing. The Internet of Things, or IoT for short, refers to Internet-connected physical devices with built-in sensors and other technologies that operate together or with more advanced systems. When we discuss testing in this context, we’re talking about tests conducted on IoT devices to ensure quality before release.
Since IoT wearables accompany us everywhere in daily life, there are thousands if not millions more test cases that QA teams must consider – a very high benchmark for “completeness” indeed. These tests extend into the physical world, where they soon stretch the boundaries of digital security in dealing with some of our most sensitive personal data. In addition to the now-familiar QA obstacle of balancing man hours spent testing vs coding, QA engineers now have another wrinkle to contend with – and unlike in application development, not every variable here is one they can control.
IoT devices come with us everywhere, and so on-device testing is required to fully simulate the conditions they will encounter in daily use. These devices allow us a greater degree of connection, but as any cybersecurity professional can tell you, that connection is only as secure as the end user is willing to keep it. While there’s not much to be done if a future end user invites risk via improper personal security practices, developers can and should ensure that QA testing of IoT devices addresses cybersecurity concerns and remains a priority. The consequences of failure are too great.
By the year 2025, there could be as many as 75 billion interconnected IoT devices in use globally. At the same time, cyberattacks have already identified these devices as an opportunity, making them targets of an attack approximately once every week. In January and February of this year alone, there was a 41% in the average number of these weekly assaults. The far-reaching effects of an IoT hack leading to a major breach of personally identifiable information have yet to be seen – but if developers have their way, they never will be.
[RELATED: Be a Cybersecurity Superhero.]
Conclusion
Hacking, errors, uncertainty, and glitches aren’t the cheeriest software development discussion topics around, but they can’t be ignored. Whether via new methodologies like QAOps or technological innovations like AI/ML models and IoT on-device testing, today’s leaders are already delivering on the promise of better QA.
If anything, automation won’t put professionals out of work — it will help them discover what work truly needs their efforts.
