Página de inicio de OpenText.
Customer stories

Pick n Pay

Major retailer accelerates software testing and increases test automation with OpenText™ Core Software Delivery Platform and OpenText™ DevOps Aviator™

About Pick n Pay

With more than 2,000 stores across South Africa and seven other countries, Pick n Pay Group is one of the leading retailers of consumer goods in Africa. Trading under two main brands–Pick n Pay and Boxer–the company traces its history back to 1967.
Two grocery employees smiling and looking at tablet
  • Stores:
    2,227
  • Headquarters:
    South Africa
  • Main retail lines:
    Grocery, home, clothing

Summary

Challenges

  • Managing a large portfolio of digital omnichannel applications.
  • Maintaining ERP applications, APIs, and integrations.
  • Constant time-to-market pressure.

Solution

  • Activated AI functionality.
  • Integrated AI into daily processes.
  • Investigated future potential.

Results

  • Reduced onboarding time by 50%
  • Cut cycle times 43%; increased release frequency 60x
  • Increased test coverage and delivered faster insight

Challenges

  • Large set of digital omnichannel applications
  • Requirement to maintain ERP applications, APIs and integrations
  • Meeting constant time-to-market demands

Pick n Pay is one of the leading retailers in southern Africa, operating more than 2,200 physical outlets across South Africa and seven neighboring countries. The company also has a strong online shopping presence for its grocery, home, and clothing businesses, including Pick n Pay asap!, a mobile app for ultra-fast local delivery of groceries.

Keeping physical stores replenished and digital applications running smoothly is vital in a highly competitive market. After all, if consumers cannot get the products they want in good time, they may turn to Pick n Pay’s competitors. This ramps up the pressure on the team responsible for testing and managing the Pick n Pay digital omnichannel applications and back-end systems.

Leon van Niekerk, SQA Manager at Pick n Pay, said, “We have a complex omnichannel application landscape supporting more than 25,000 API calls, three different mobile platforms, in-store scanners for stock picking, and integration with an external delivery partner. We’re also responsible for testing SAP Retail and all its integrations with the omnichannel apps. The big thing for us is always: how quickly can we get into production?”

As a long-term user of OpenText DevOps Cloud solutions to manage and automate software testing throughout the lifecycle, Pick n Pay currently manages more than 20,000 automated test scripts weekly using OpenText Core Software Delivery Platform (SDP). The solution uses OpenText MF Connect Core to sync releases, stories, and defects with Pick n Pay’s Jira development tools.

When OpenText announced its OpenText DevOps Aviator™ AI solution, Pick n Pay was quick to sign up as a beta tester. The retailer’s goal was to see whether AI technology could help it get new testers up to speed sooner and accelerate the creation of test cases.

“Our internal program to train junior testers used to take anything between six and twelve weeks,” said van Niekerk. “Equally, writing test cases for features took around seven days from notification to handover to the automation team. Given OpenText’s successful track record of helping us evolve our testing lifecycle, we were excited to see what OpenText DevOps Aviator could do in terms of making things easier and faster.”

Person sitting down after shopping and looking at phone

It was immediately clear that OpenText DevOps Aviator had added things that we had missed, giving us better test coverage from a feature or user story point of view.

Leon van Niekerk
SQA Manager, Pick n Pay

Solution

Following a successful proof of concept conducted during beta testing, Pick n Pay has activated OpenText DevOps Aviator in its production OpenText Core SDP landscape. The retailer is using the AI tool to accelerate the creation of test cases for features.

Products deployed

Activated AI functionality

Pick n Pay uses OpenText Core SDP, the OpenText cloud-based software delivery management platform, so deploying OpenText DevOps Aviator was simply a case of switching on the new functionality. In trialing the technology, Pick n Pay had two key objectives. First, to confirm that it could trust the results and second, to compare the performance of the OpenText solution against that of a generic AI tool.

As the first step, Pick n Pay selected a handful of stories from a recent release, pulled them into a sandbox in OpenText Core SDP, converted them to features, and asked OpenText DevOps Aviator to create test scenarios from them. The team then compared the output against the test cases it had created manually.

“It was immediately clear that OpenText DevOps Aviator had added things that we had missed, giving us better test coverage from a Feature or User Story point of view,” said van Niekerk. “We then put the same features into ChatGPT. With the historical data that we have in Core SDP, we could see that OpenText DevOps Aviator gave us less generic and more detailed ideas around test cases.”

Integrated AI into daily processes

“The beta testing program proved to us that we can trust OpenText DevOps Aviator and that it will add value in the daily lives of our testers,” said van Niekerk. “However, it’s not a replacement for human expertise. You can use AI to generate test cases, but you still need to check them and make sure that everything is correct.”

Pick n Pay is continuing to analyze data on how its teams use OpenText DevOps Aviator and how best to adapt its processes. Typically, the solution accelerates the creation of test cases, giving employees more time to work on refining and testing the test cases.

Investigated future potential

As Pick n Pay extends its usage of OpenText DevOps Aviator, it is seeing improvements in accuracy and detail over time. “We can definitely see that the AI is learning over time,” said van Niekerk. “It’s becoming more Pick n Pay, in the sense that it seems like it remembers our way of doing things and prompts us to use previously identified steps. So that's very good.”

The next step is to investigate the capabilities of OpenText DevOps Aviator in generating automated script files and the resulting impact on test coverage at Pick n Pay. The retailer uses OpenText™ Functional Testing and has developed its own keyword-based, data-driven framework for test automation. “We are already up to about 95% on in-sprint automation coverage, so it may be difficult to increase the percentage substantially,” said van Niekerk.

Grocery store employee holding tablet and smiling

With [OpenText] Core SDP, it takes literally two seconds to understand where we are per sprint, per release, per feature level, for every application we test in our omnichannel space.

Leon van Niekerk
SQA Manager, Pick n Pay

Results

Using OpenText DevOps Aviator has accelerated the onboarding, reduced cycle times, increased the frequency of releases, improved coverage for both manual and automated testing, enhanced quality, and delivered better support for Go/No-Go decisions.

Cutting onboarding by 50% and boosting efficiency

Onboarding new junior testers at Pick n Pay previously required at least six weeks. With OpenText DevOps Aviator on top of OpenText Core SDP, the company can get team members writing and understanding test cases within just three weeks.

OpenText DevOps Aviator is also helping Pick n Pay make more efficient use of existing team members. By freeing up time in the generation of test cases, the solution has enabled Leon van Niekerk to move one person from a functional testing role into an automation role.

“At no additional cost, we’ve been able to take a team member who understands the application and bring them into the automation team,” said van Niekerk. “That means we’re bringing the two skillsets and knowledge bases closer together, which should translate into higher quality.”

Reducing cycle times by 43% and increasing release frequency by 60x

It previously took Pick n Pay about seven days to write test cases and hand them over to the automation team. Today, with OpenText DevOps Aviator, it takes just four days.

“In the past eight weeks, across the 17 applications we test in omnichannel, we've completed four to five releases a week,” said van Niekerk. “The guys on the team are starting to say we’re almost like Google in that we could release two or three times a day if we needed to.”

Before Pick n Pay deployed OpenText Core SDP, the company would typically release once every six months. By contrast, the company recently completed more than 60 releases in a six-month period. “Our reduced cycle times stem from a combination of test automation and AI enablement using OpenText technology,” said van Niekerk.

Increasing test coverage and quality

The AI-enabled acceleration of test-case creation opened the way for Pick n Pay to adopt in-sprint and in-release automation. This has enabled the retailer to automate much earlier in its two-week sprints, driving a massive increase in automation coverage from about 65% to about 95%.

“With OpenText DevOps Aviator, we no longer really have to write manual test cases,” said van Niekerk. “And with the move to in-sprint automation, we’ve also added 25% to 30% coverage on our testing for platform specifics. That increased coverage definitely improves our test quality: I cannot remember when last we had an outage in production caused by something we missed in testing.”

Delivering faster insight and supporting decision-making

Deploying OpenText DevOps Aviator has also provided some unexpected benefits at Pick n Pay, where the ability to query status using natural language is helping the team make Go/No-Go decisions.

“We no longer have to send out a dashboard,” said van Niekerk. “We are online in those sessions with OpenText Core SDP, and we can say, ‘Show us the number of open defects for this sprint.’ The information is there immediately, so Go/No-Go decisions are a lot easier to make using OpenText DevOps Aviator.”

OpenText Core SDP also helps Pick n Pay to identify features that are under-served in terms of their test coverage and to see the current status of releases and sprints without having to chase test leads for statistics.

“With OpenText Core SDP, it takes literally two seconds to understand where we are per sprint, per release, per feature level, for every application we test in our omnichannel space,” said van Niekerk. “Software testing is not random; we have to base all our activity on facts. And those facts are all readily accessible in OpenText Core SDP.”

How can we help?