Dr. Gareth Smith explains how retailers can improve software quality to increase sales in this exclusive Q&A piece.

E-commerce spending is expected to keep growing by double-digit percentages through the next six years. And if retailers want to increase sales, it requires more than a functional e-commerce channel; instead, they need to deliver a seamless omnichannel experience.

In order to do this, retailers across the globe are rethinking how they test the quality of software.

To find out more about how the innovative technology is helping retailers improve software quality to increase sales, we asked Dr. Gareth Smith, Keysight’s general manager of Software Test Automation, to explain more.

What is intelligent software test automation and why should it replace manual testing and other forms of test automation?

Traditionally retailers have relied on testing that software and applications work. However, with the user experience (UX) determining success, retailers need to continuously test user journeys across all touchpoints, browsers, devices, and systems.

The only way to do this is by integrating intelligent technologies that automate testing from the user perspective and garnering insights in order to optimise the experience. Essentially, this is like unleashing an army of AI-powered bots onto an application, website, or process that behaves like real users and explores the workflows as real users would.

Until such AI-powered approaches were pioneered, manual testing – which is humans testing the website or application by hand, was the most common form of testing.

Manual testing approaches are slow, expensive, and aren’t comprehensive, making them obsolete in a digital-first world.

Similarly, traditional or legacy test automation doesn’t use super-smart AI – and is more rigid and predefined.

Typically, legacy test automation runs the same test scripts every time software is released. Maybe a bug will be found, if you’re lucky, but there is certainly no intelligence involved. These approaches fail to intelligently spot new bugs that have crept in, do not proactively find the weak points in a system, and lack the brains to identify what is critical to test from a user’s perspective. Rather than evaluating if software “sort of works”, retailers need a way to ensure it blows your socks off — from a quality point of view but, most importantly, from a user experience point of view!

Intelligent test automation can evaluate the functionality, performance, and usability of digital products rather than simply verifying code. It incorporates artificial intelligence (AI), machine learning (ML), and analytics to continuously test and monitor the digital user experience; it analyses apps and real data to auto-generate and execute user journeys. The result is a smarter way to test software and apps continuously.

How can intelligent test automation help retailers deliver a seamless experience that meets customer expectations?

Intelligent automated testing enables retailers to take complete control of the testing experience and understand devices in the same way a consumer does.

For example, intelligent test automation software can use the touchscreen on a self-checkout in the way a user would. It can understand the stream of images and text on the screen – using intelligent image and text understanding algorithms and it can send in clicks, swipes and key presses, just as a user would.

When it comes to checking out, it can control a robot arm (yes we don’t have to stop when things are not purely digital – we can bring robots to bear too!) to insert a credit card or tap a PIN pad.

The AI engine can verify any application’s functionality and user experience. To make this possible, a model is constructed that represents all possible flows through an application, website, or process. The model can be created interactively, by a designer, or automatically by watching a real app in use and charting every possible path. Behind each step of the model, automation actions emulate how a real user would interpret inputs and generate outputs.

The real smarts happen when AI-powered bots navigate through the model. They decide which paths to take, based on what they’ve learned and their imbued ‘behavior characteristics’.

For instance, they may focus on the user journeys that historically generate the most revenue – as these are important, or the user journeys that have changed the most since the last release. As bots navigate the model, they generate real actions on the real apps, which can be interpreted and responded to through intelligent automation.

Regardless of screen size or input devices, intelligent image and text understanding works like the human eye and brain to adapt to the form factor and interact with it. In this way, the intelligent automation can test any device, platform and is technology-agnostic.

Retailers can quickly roll out automated testing processes across systems, increasing visibility while saving time and money. Ultimately, this allows retailers to become agile, ready to adapt to unforeseen circumstances, and move quicker than competitors.

Why should retailers shift from multichannel to omnichannel testing?

In retail, UX is everything and brands need to provide an amazing and seamless omnichannel experience. This requires continuously testing user journeys across all touchpoints, browsers, devices, and systems. Multichannel testing looks at the usability of features across different devices and is an important baseline.

However, it’s not able to determine the user experience as the customer journey traverses different touchpoints – including mobile device, web, call center and in-store (including point of sale and instore terminals). With AI-driven test automation, the algorithm connects the data stack across channels, enabling omnichannel testing. Retailers can then evaluate the user experience and benefit from speeding up the delivery of software and applications, improving the quality, and saving money.

Tell us about AI and its role in improving the quality of software and applications?

AI enables testing to move beyond simple rule-based automation. It utilizes algorithms to efficiently train systems using large data sets. The algorithm mimics human behavior through the application of reasoning, problem-solving, and machine learning. It works by executing automated test routines that reflect the actions of human users. The AI hunts for user interface errors, bugs, and performance glitches and then auto-fixes issues before they can impact the customer experience. The automation increases coverage as it can explore every potential user journey and predict and identify any bottlenecks that may affect performance. AI-powered testing enables retailers to accelerate the release of high-quality software, providing a significant advantage.

It’s like having your own tame army of human testers – except they don’t require salaries or lunch breaks and they can operate thousands of times faster!

With retailers’ success reliant on the performance of their digital products, how can they ensure that once they release software and applications, the applications continue to meet user expectations?

Once software and apps are released, retailers need to continuously test the performance to understand the customer experience and identify and resolve issues.

No retailer can afford an outage. Therefore, it’s essential that digital properties are prepared for surges in traffic and perform as users expect. With the increasing reliance on digital channels, the quality of software and applications directly impacts customer loyalty and sales.

AI-driven testing aims to understand the customer journey and how to influence a behavior to drive a specific action—like a purchase. For example, if the analytics from the performance testing identifies that 23% of customers are abandoning their basket at step 3. Then the algorithm will identify the root of the problem and come up with software updates that will improve the conversion rate.

Should retailers hire more IT/technical resources to manage intelligent testing?

Another benefit of intelligent automation is that it is easy to use. This removes the need for retailers to add technical employees with a programming background to run and manage testing efforts. 

Can you outline some examples that of how the retail industry harnesses intelligent automation?

  • Point of sale: These systems, including self-service, cashier, or mobile payments, are constantly being updated. Due to the cost, retailers often build new features, such as mobile ordering, offers, and contactless payments, on top of established platforms. This creates a complicated technology stack that was difficult to test before intelligent automation. 
  • The retail ecosystem: The AI-driven automation can test the full range of connected systems in a retail enterprise, spanning e-commerce, logistics, warehouse systems, order fulfillment, supply chain systems, merchandising, to ERP platforms.
  • Innovations: With the retail industry undergoing a technology arms race, there is a constant array of new technologies such as robots, RFID tags, augmented reality, and crypto payments to integrate. Retailers can now quickly and easily test these innovations and understand how they will impact the business.
  • E-commerce: The intelligent automation can predict changes in conversions, bounce rate, and revenue. It identifies which technical behaviors are impacting conversions and propose optimisations to increase sales.

How can intelligent automation improve a retailer’s bottom line?

By adopting intelligent testing, retailers can optimise their digital properties by using the insights from smart testing to make bolder, faster decisions regarding additional features and functionality that will drive up conversion rates.

For example, if the goal for your e-commerce website is to sell US$2 million in merchandise per week, but you’re only selling US$1 million, you need to determine why revenue has dropped off. The intelligent automation might find that it’s a basket issue during checkout that drives customers to abandon their cart. These are the types of insights that the AI provides retailers with, helping them, in turn, optimise their digital business.

As the retail industry continues to transform, what trends do you anticipate and what role will Keysight Technologies with Eggplant intelligent automation play?

IoT and Contactless shopping will continue as we leave the pandemic in the rearview mirror. Consumers have adjusted to the rapid acceleration in digitisation and with health remaining a global concern, the contactless trend will continue to gather steam. This will also fuel more “Walk-Out” retail stores that require an even greater set of IoT technologies that need to be tested to ensure customers are accurately charged, requiring an increased overlap of cyber+physical testing.

Augmented reality (AR) will deliver immersive experiences changing how consumers interact with products for remote shopping. This virtual commerce will enable consumers to visualise products, furniture, clothing, etc. in their real environment. These richly interactive and graphical applications increase the ways a customer can interact with the retailer. Intelligent automation of real user experience is key to ensuring end-to-end quality for these customers.

5G will help deliver in-store location-based services via standard and AR applications to offer a more finely personalised customer experience. This adds an additional level of complexity for testing as both time and location, as well as bandwidth and other network effects, will all impact the customers’ experience and need to be considered when testing new initiatives.

Dr. Gareth Smith leads Keysight’s software test automation group. Previously, Smith was CTO at Eggplant – the pioneer in intelligent test automation, acquired by Keysight in June 2020. He has a rich history of innovation in software, serving in leadership roles at Apama, Software AG and Progress Software.