Imagine sitting in an empty office, with only your PDA, considering the conundrum of being tasked to design, build, test and implement Australia's first mobile virtual network operator (MVNO) GSM carrier service. Daunted? Well, imagine if you also had to design, build, test and implement an entire set of corporate infrastructure systems as well as hire the staff to do it - all within a six-month timeframe?
Barry Evans, director of IT for Virgin Mobile Australia (VMA) has lived to tell the tale of successfully doing all this. "I didn't think what was being planned was impossible. I could somewhat visualise the size of the undertaking, but the devil was in the detail."
He said the problem was not being a MVNO, as the company was 'piggybacking' on Optus' network, it was the issue of the time it would take to put the technology in place.
"We were ignoring the accepted timeframe to build a carrier network."
VMA was the first local company to adopt the MVNO model. The company uses the backbone network infrastructure of the Optus network, but has its own prepaid, switching, voicemail, short message service (SMS) and value added platforms.
Evans said since the business had committed to a go-live date of November 1 2000, it was imperative that all systems were built "from scratch" in parallel and be operational, with cross functionality, by that date.
Other business goals for the IT systems at the outset he said, were that it had to be flexible enough to accommodate changes over time, based on market pressures (for example new data services), scalable to accommodate a minimum of one million customers, and the architecture had to be capable of accommodating the future requirements of intelligent networking and mobile number portability.
To ensure the launch date was met, the company had to adopt an approach to problem solving that was "different" to most companies.
"We had to be prepared to make decisions. [In the case of IT] people had to recognise the urgency of the project."
With an IT budget of $45 million - which the project came in under -- Evans said the difficulty with the implementation plan was that it had to built backwards to ensure the go-live deadline would be met.
"Due to the constraint of the launch date we had to look at what could be realistically delivered in that time. We had to look at how much time we could spend on design, development and testing."
For example, PeopleSoft was implemented as the ERP system. "We had four months to get it in. So we knew we could only implement the minimum functions in that window.
"The key was to focus on the delivery of customer-facing functionality for the launch. It was understood that many of the processes and back-end functionality would be delivered post-launch."
This included putting the problem of resolving number quarantining - when a mobile number is disconnected or cancelled it is placed in quarantine for six months and then used again - on the backburner until six months down the track, when it would become a concern.
The round-off procedure for the end of year financial process was also attended to post-launch for the same reason that it would not be required until 12 months down the track.
Evans said during the early post-launch days, IT also focused on transferring the business ownership of system functionality back into areas such as finance, so it could turn its eyes to the operational maintenance of the systems.
"As the systems were deployed so quickly and other business units were also being recruited in parallel, the system knowledge resided solely within IT and not in the business units."
Evans said the mentality of concentrating on key launch functionality -- which still required working 100-hour weeks for six months -- was vital to the success of the projectVigilantly keeping to milestones and eradicating scope creep was also crucial to ensure systems were online at launch.
"As with most projects, people have ideas after the design has been finalised and then try to squeeze them in. The key to making sure we kept to our plan was to lock down the design; except of course for some exceptions."
To build the company's IT infrastructure and carrier service, VMA deployed 25 major systems and 20 minor systems with 20-plus interfaces from 50-plus vendors.
Therefore, a major concern for Evans was handling personnel involved in the project.
"I needed a lean and mean headcount total so [internal] people had to multi-task [they also needed to be able to] work with very little structure initially and hence be highly autonomous.
"During the height of the implementation, we had 310 people on board, including 21 internal Virgin Mobile IT people and vendor people."
While everyone involved in the project was considered "equally part of the team", Evans did not allow vendors to become single-minded about their project.
"Each vendor had to adhere to our approach [for the project] and report progress in the same way. They had to ignore their own [implementation] methodologies."
As well as adhering to the 'VMA' way of doing things, potential vendors had to comply with certain specifications to stand a chance of being engaged.
Evans said if vendors were not able to deliver within their time frame they were "discarded". Support was also critical; they had to show a willingness to partner. "We wanted long-term partners." Whether or not the vendor was an Australian company was also a consideration.
Since the launch, Evans said the carrier has been working to reduce the number of supported operating systems.
"We have eight to 12 major vendors that play together and the rest sit around them. Due to the large operating costs we would like to rationalise our vendors to about four or five."
Core vendors include ADC (customer care and billing system), PeopleSoft (ERP system), Argent (service node and prepaid platform), BEA Systems (WebLogic, used as an application server giving functionality to the Web, IVR and back end) and Oracle (data warehouse).
He said the company was hoping that, as vendors increase their functionality, it will be able to reduce the number of vendors it relies on.
Reflecting on all the hard work and what was accomplished so quickly in the first six months, Evans said he was "extremely proud of what was achieved considering the timeframes dictated by the business, the starting point (nothing) and the complexity of the solution".
Despite the difficulties of defining the IT requirements of a business that was completely new and the danger of creating a "Utopian" system, Evans said he relished some of the benefits of starting from scratch.
"We had no legacy systems, no existing processes to sustain and the ability to deploy the latest revisions of software and hardware."
And perhaps best of all, "There were no customers to consider sustaining during development and migration," he said.