McDonald's is gearing up to pilot its new homegrown and cloud-driven ecommerce platform in Australia after switching the new system on in North America.
The fast food giant is breathing a sigh of relief after rolling out the AWS-backed transaction engine - which will eventually link up point-of-sale systems on 200,000 registers and 300,000 devices in 36,000 stores worldwide - in the USA and Canada earlier this year.
It will now commence pilot runs of the platform in Australia and the UK, and is preparing to go live in China in the first quarter of 2017.
It has been a project on a staggering scale, chief technology officer Tom Gergets told the annual AWS summit in Las Vegas this week.
“Orchestrating change and transformation of a company the size of McDonald's from a traditional technology organisation to a digital innovator is very challenging,” he said.
The end-game for the chain is to inject consistency of customer experience, agility, and speed into its global food operations, which serve 6000 menu items to 69 million customers a day.
“Each of our restaurants runs a varying availability configuration, and they need to come together at one moment in time," he said.
“But our customers are looking to us to be available 24/7, both inside of our restaurants and out. They are looking for a personalised experience, they’re looking for recognition of their loyalty and even options for delivery.
“To meet those expectations we are investing significantly in this technology."
After a worldwide market survey to find an ecommerce platform that could meet the company's scale and demands, Gergets and his team came up empty handed, so proceeded to build their own system internally on AWS.
It wasn’t a hiccup-free process, he admitted.
“We started with IaaS. Looking back it would have been a better choice for us to start sooner with platform services.”
The ecommerce platform stitches together 35 different Amazon services: it is built on EC2 and leverages Elastic MapReduce and Redshift, Elastic Load Balancing, ElastiCache, CloudWatch, and Identity and Access Management.
Gergets boasted that the go-live in North America had exceeded the chain’s expectations.
“We were targeting an ordering throughput at 350,000 transactions per hour concurrent, and we achieved 500,000 - a 43 percent improvement," he said.
“Our offers target was 600,000 per hour concurrent and we achieved over a million - a 66 percent improvement. That equates to 31 million transactions per hour."
Paris Cowan travelled to AWS Re:Invent as a guest of Amazon Web Services