How Telstra cut its dev time in half

Telstra has set a goal to get 400 of its development teams on the agile methodology by 2019 after a partnership with Pivotal found time spent coding could be cut by more than 50 percent.

The telco signed a joint venture with the Silicon Valley start-up in 2015 that saw one Pivotal lab jointly owned and operated by the pair opened in Sydney, and another in Melbourne run by Telstra offering Pivotal’s methodologies.

The telco has been a customer of Pivotal since before the joint venture, using the firm for both its customer offerings and its own internal development practices.

Pivotal’s main proposition is agile software development enabled by its Cloud Foundry platform-as-a-service and analytics products.

Telstra chief Andy Penn today revealed the telco was aiming to have 400 of its teams - around 5000-6000 staff - using agile and Pivotal Cloud Foundry for software development by 2019.

“We’re on a big journey in terms of adopting the agile methodology within the organisation,” Penn said.

“I don’t think there’s a single Australian organisation that isn’t impacted by technology innovation and needing to build new skills in new areas to be successful.”

At the moment around 100 teams have transitioned to the new approach, and all new projects will commence on this model. Pivotal has three teams made up of 12 staffers embedded within the 100 teams.

Telstra has also integrated Pivotal’s technologies into its consultancy offerings to customers. Early agile customers include Macquarie Bank, NSW Government, and CoreLogic RP Data.

Slashing time to market

The partners claim they have been able to cut Telstra’s software development time from between 6-8 months to 10-12 weeks by using agile and Pivotal’s technologies.

This has been achieved through an intentional shortening of the scope of development alongside an increase in efficiency, Penn and Pivotal CEO Rob Mee said.

“In order to be able to change what you ship very quickly, safely, and efficiently, you have to put a lot of disciplined software engineering practices in place that allow you to also go faster,” Mee said.

“There’s a number of aspects to this: efficiency, as in doing things more quickly, and shipping things sooner and getting feedback.

“It’s very much an experimental way of working in that you have a hypothesis and you attempt to validate it or invalidate it by getting feedback.

"It’s actually more scientific in that sense, as opposed to saying ‘well we know everything we need to know about what we’re building and we’ll just ship it in six months from now’, and then you find out you didn’t really know.”

In practice, this presents itself in several stages. The initial short “discovery and framing” phase works out what the team is aiming to achieve from a product and design perspective, Mee said.

The following inception phase - which lasts only a day or two - plots out exactly what the minimum viable product will look like, which in turn provides a list of features to build on a day-to-day basis and an estimated release time.

Replanning occurs frequently following the initial product release, Mee said, to incorporate lessons learnt from the build as well as early customer feedback.

The first project Telstra applied this approach to was the Ram Select app it created with Pivotal for the Sheep Co-operative Research Centre. 

The app, launched in 2015, aims to make it easier for breeders to select rams based on certain traits.

“The world and technology innovation is moving so fast that we can’t afford to wait six to eight months to get a product to market, because frankly by the time you do …[the initial customer specification] has probably moved on,” Penn said.

“What [agile] allows you to do is interact better with customers’ needs, get out to market much more quickly with a minimum viable product, and then be able to iterate and innovate, and be much more up to speed with developments and functionality for customers."