Countdown is one of New Zealand’s leading supermarket brands and serves more than three million customers every week. With the help of 18,000+ team members, customers enjoy a choice of over 20,000 different products in each of their 180 supermarkets across New Zealand.  The ‘Countdown X’ division manages the digital and technological aspects of the business, including online shopping, the Onecard loyalty programme, digital marketing, customer personalisation and customer care. The Countdown brand is part of Woolworths New Zealand, a subsidiary of the publicly-listed Australian company, Woolworths Limited.

The Challenge

Countdown’s online business has grown rapidly over the last few years. The leading online food and drinks retailer in NZ, they have a passion for creating truly great customer experiences. CountdownX recognised the need to provide their customers with self-service options in order to scale their customer support team at the same rate as their growth.

Introducing a digital assistant to serve multiple channels became a key part of their digital strategy. They wanted to provide quick and efficient service via web chat, interactive voice response (IVR), and ensure integration with virtual assistant platforms, such as Alexa and Google Assistant. Countdown decided to leverage existing work done by the WooliesX team in Australia, but needed to create a customised version for Countdown and the NZ market.

The CountdownX team chose ClearPoint because of our expertise in digital innovation and DevOps practices.

“The Clearpoint team provided technical and thought leadership to ensure we created something helpful for our customers and relevant for our business. They provided a great team who blended well in our Agile Tribe and added to our culture of experimentation.”

— Adam Hatley-Owen, Head of Digital Experience, Woolworths NZ

The Solution

ClearPoint started the journey with Countdown by building a proof of concept virtual assistant in two days. This prototype was validated and we began building Countdown’s Olive with the goal to go live within 3 months. Our engineers built the digital assistant using Google DialogFlow NLU (natural language understanding) integrated with a custom fulfilment service (using NodeJS), which connected to other Countdown backend systems. We implemented fully cloud-based containerised services using Kubernetes and Google Cloud. We created an automated deployment pipeline using Google Cloud Build. The creation of such a pipeline allows Countdown teams to deploy to test environments in under 10 minutes and promote to production in under 5 minutes. Very high unit test coverage and automated end-to-end tests ensure fast, safe, and frequent deployments to production – a crucial requirement when running a digital assistant.

"The team at CountdownX really understand Product Management. This starts with setting the desired outcomes and giving teams the space to deliver. Another key aspect is their grasp of how to evolve a product through ongoing evidence-based investment.”

— Marcus Simmons, Product Manager, ClearPoint

The Outcome

Olive went live on a single page in September 2019 (within three months of kick-off) with the ability to answer general questions about store details, opening hours, locations and FAQs. The team then introduced the ability to handle refund requests and Olive was subsequently rolled out across the rest of the site (including main ‘Contact Us’ page in November).

The ClearPoint team implemented an automated deployment pipeline (including end-to-end tests for each feature) right from the beginning – enabling high-quality releases multiple times a day.

Olive has already answered thousands of questions from Countdown customers, reducing call volumes to the Customer Support team. Historical conversations are used to both prioritise new features and refine training data for the machine learning model with a goal of 80% conversational success within the first year of operation.

The key Objectives for the platform are:

  • Fast, safe and frequent enhancements
  • Enable customer self-service delight
  • Provide customer insight gold

These three Objectives are being tracked via 12 Key Results.

  • Cycle Time for New Features
  • Deployment Frequency
  • Time to Deploy
  • Test Coverage
  • Change Failure Rate
  • Monthly Active Users
  • Conversation Success
  • Avge Session Length
  • Fallback Rate
  • Team Adoption
  • Team Usage
  • Insight Value

Tech Points

  • Google DialogFlow
  • Google Cloud Build
  • GKE - Google Kubernetes Engine
  • CI/CD pipelines
  • NodeJS
  • Kubernetes
  • Docker
  • Helm
  • Stackdriver

“It’s been awesome working with CountdownX - their engineering practices really align with ours, which means we were able to deliver a digital assistant really quickly, along with a CI/CD pipeline which has been instrumental in our success.”

— Robert Redgwell, Principal Engineer, ClearPoint