We know that IT projects often go wrong, but it's a bit much for 10 of Victoria's major IT projects to do a face plant in the same decade. We tell you what made each of the projects bomb.
The Victorian Ombudsman has published a report on 10 major Victorian IT projects, providing case studies on those he looked at. We dissect his findings for you.
1. LEAP of faith
The first was Victoria Police's project to replace its LEAP database, which had been used since 1992 to record crime incidents, with a new system called LINK.
In 2005, spurred on by a previous ombudsman's report, the then-premier announced $50 million in funding for a three-year roll-out, and a business case was later completed that said it would cost $59 million to build. This business case, completed by an independent consultant, "grossly underestimated the cost and complexity of the project", according to the ombudsman, because the scope of the project had been written to fit the funding, instead of looking at how much it would cost to get what was actually needed.
A leadership vacuum (there were two people heading up the project with joint responsibility, one of which had inadequate experience, the ombudsman said) meant that it took almost four years for the Victoria Police to realise that the project would be at least $80 million over budget.
Meanwhile, the team took the off-the-shelf product that had been chosen, and over-customised it, so that it would replace LEAP's functionality, instead of addressing future policing requirements.
When a new business case was finally created early this year, it laid out that an additional $127.7 million in funding was required in addition to the original $59.48 million. However, after this, the project failed its first gateway review, because it hadn't taken future policing requirements into account, and the project was put on hold in August. The team now intends to ask the Cabinet budget committee for $12 million to develop a business case that would focus not only on the replacement of LEAP, but also on transforming Victoria Police into a modern policing organisation.
Takeaway: rushed business cases don't cut the mustard. Project experience is paramount.
2. Health fart
In 2003, the Victorian government embarked on what was expected to be a $323 million program called HealthSMART, to create a consistent IT foundation (including a finance, patient-management and clinical application, as well as an organisation to manage them) for the Victorian public health service.
Most of these objectives were doable (even if they ran overtime), but the clinical application really put a spanner in the works. With the money available, the health department has only managed to roll out the elements of the application to four out of 10 health services.
Although the project was scheduled for completion at the end of 2007, the finance and patient-management applications were only finished in 2010 and the Department of Health estimated that it could take 10 years to get the other six health services onto the clinical application. This meant that by the time it was implemented in some of the services, it would be 10 years old. Part of the delay was because the department hadn't realised that each health service would need their board to approve a business case for the roll-out. In addition, the level of technology within the health services was woefully overestimated, the ombudsman said, while the level of training that staff needed was underestimated.
Meanwhile, pharmacists and doctors complained that the medications data in the clinical solution was up to 18 months out of date because of vendor delays, although the department disagreed that this was so. The vendor also hadn't sent experienced enough staff until a line was made directly to the US CEO to remedy the situation. An awkward three-way relationship between the department, the vendor and the health services made everything more difficult than it should have been. Staff weren't enamoured with the system, with minor irritants leading them to believe that it would have a negative effect on patient safety.
Moneywise, the government allocated an extra $104 million in the 2008-09 budget to the project, but with extra funding, which wasn't recorded against the project, and an additional $95 million that the department thinks it needs to complete the final six health services. The project is expected to end up costing $566 million; one third over budget. Confusion over whether agencies would be paying their way led to uncertainty.
Takeaways: plan first, and then do. Know exactly where the money's coming from. Have a simple vendor relationship.
3. Strike me, Myki
Despite the fact that ticketing projects had taken twice as long to implement in other countries, Victoria thought that it could deliver an electronic ticketing system within two years. This was to be achieved using the agile software methodology, and by hiring skilled staff to the project. However, the timetable turned out to be overly optimistic.
The project was conceptualised in 2002, and the Transport Ticketing Authority was created in 2003, with the hopes of replacing the existing Metcard ticketing system in 2007, when the contract with Metcard expired. The business case, conducted in 2004, said that the project would cost $741.9 million up to 2017. Yet, by 2005, the budget had already burgeoned to $999 million. In 2008, it had risen to $1.35 billion. Some of this added cost was due to the Metcard contract having to be retained, and $37 million has been spent on consultants.
Given that the project involved 20,000 physical devices to be deployed over many different kinds of transport with 17 contractors, the project was complex, and made more so by the fact that the authority decided to build it with open architecture, so that when gear is replaced, components from any brand can be used — a concept that had not been proven elsewhere. In addition, the ticketing authority had added hundreds of new specifications throughout the life of the project, increasing complexity further. The contract had 13,000 pages.
The ombudsman questioned whether the board had enough members with skills in ICT, which meant that it couldn't question details of the project. The relationship with the vendor was also not a good one, as the authority became frustrated at goals not being met, while the vendor was angry that the goalposts kept being moved. And the project will likely go up in cost again, because the scope of the project is still varying, as it's not finished yet.
Takeaways: trying new things is risky, and complexity costs and causes confusion.
4. Regret and Loss
RandL (registration and licensing) was a project to replace VicRoads' ancient licensing and vehicle registration systems, which are expected to fail within five to eight years. The ombudsman had suggested the replacement.
Although VicRoads conducted a business case and wanted $156 million to complete the project, it was only allocated around $50 million as part of a $115 million budget, which was spent on procurement. When the project again looked at a business case, it found that $115 million would be inadequate. When asked in 2011 for the $123 million that VicRoads needed to complete the project, the government refused. The ombudsman didn't understand where the government had come up with the $115 million budget, finding no evidence that it would be enough to complete the project. It also praised VicRoads for what it considered to be best-practice procurement on the project.
The team has stopped working due to lack of funding, with the $50 million wasted. The ombudsman thought that if the government had approved the funding, it could have been completed, with only $17 million more than the original business case. He also felt that VicRoads could have raised its registration and licensing fees as alternate funding possibilities.
Takeaways: ensure funding for whole project before spending.
The Client Relationship Information System (CRIS) project was begun in October 2001 within the Department of Human Services, with the hopes of replacing client-information systems for programs such as Child Protection, Juvenile Justice and Disability Services by 2005.
Its budget was $22 million. In 2002, however, it was rolled into the Client Services Model Strategy Project, which wanted to simplify the way in which the Department of Human Services managed clients across programs. This project was delayed, not starting until 2005. Then the CRIS implementation was halted because of functionality flaws, and further flaws caused later delays. Part of the problem was that although Child Protection had signed off on the business case, it later wanted changes to the system, ostensibly because the new system didn't have the same or better functionality than the existing one.
The project came in three years behind schedule, and three times its original budget ($70 million, and completed in July 2008). The shortfall in budget was taken from program areas, such as Child Protection.
Child Protection workers are still not happy with the system, despite it being in place for three years, the ombudsman said. He found that the system is, in fact, worse than its predecessor for Child Protection, and that it impedes efficiency. Training and the addressing of system changes is required, according to a later review, yet the ombudsman said that there are inadequate resources, and only one trainer for 2000 staff.
Takeaways: identifying business requirements is paramount.
In 2006, the then-government set off on a project called Ultranet to create an online education network to connect students, teachers and parents, providing access to online learning materials and student information.
A tender went out in 2007, but was cancelled at a cost of $5 million after no tenders were able to respond within the budget.
Feedback from the market said that the timeline was too aggressive, the budget was insufficient and the scope of the project wasn't defined clearly enough. DEECD reduced the 1260 detailed business requirements down to 131 high-level requirements. Another tender was released in November 2008, with the same deadline as the tender released the year before.
Having the same deadline meant that the project was released to schools on time, but without system testing, and not fully functional, leading to poor initial performance, the ombudsman said.
At June 2011, $63 million had been spent on the $60.5 million project, but the ombudsman believed that $99 million would need to be spent by the end of the project due to additional costs for maintenance and training. He did say, however, that the costs and scope had so been revised that it was difficult to say whether the budget had been met. He noted that schools were positive about the project and its benefits.
Takeaways: plan tenders with feedback from the market, keep an eye on the risks.
7. ICMS's multiple warheads
In 2005, a project to integrate the Department of Justice's technology for all Victorian courts and tribunals began. This integration consisted of CourtView to manage cases, Smart Courts for videoconferencing, the Judicial Officer's Information Network as an online resource, Courts Data Warehouse System for data extraction of case-management information and eServices. The project was called the Integrated Courts Management System (ICMS).
It was meant to be completed in September 2009, but, although the Judicial Officers Information Network and Smart Courts were delivered, Courtview was, at that time, "at best 60 per cent complete". Vendor-delivery failure had meant that the model for producing orders wasn't provided. The module is now defective and unusable, the ombudsman said, with a full roll-out of CourtView to all courts being put on hold pending a review. Courts Data Warehouse and eServices is waiting on the Courtview implementation.
The estimated budget required was $55 million, but, like RandL, only part of that funding was provided, despite the fact that a similar project in the US cost considerably more. The timeline of the project was shortened because of the lower budget, but the project deliverables weren't reduced.
The original business case assumed that the existing County Court Case and List Management System could be used as the case-management system component of the project; however, the vendor didn't submit a bid for the project, leading to the department choosing an alternative system, CourtView, which was considered to be riskier, as it needed products from multiple sources to be integrated into a working solution. To make this decision, the department relied heavily on a systems integrator.
The vendor that was chosen provided inadequate performance, the ombudsman said, failing to meet timeframes. Due to a change in ownership of the vendor, the department was also forced to renegotiate the contract. The courts were then unhappy with CourtView's performance, suggesting alternative solutions. The court said that it didn't get to see the business case.
The ombudsman considered that a product should have been rolled out to one court before rolling it out to others to check the product's viability. Again, the suggestion was made that the Cabinet budget committee only took notice of big projects.
In 2010, an aggressive law-reform agenda meant that the ICMS business case had to be reviewed, leading to increased costs and delays. Funding for the project has now run out. The ombudsman thinks it will require an additional $21 million to complete, but whether it will be rolled out to any more courts is uncertain. The Supreme Court and the Department of Justice are talking about the situation, but the Supreme Court doesn't think it wants to continue with the project.
Takeaways: start small to pilot a project before starting a massive roll-out.
8. Face PALM
PALM was born in 2008 when two Victoria Police projects were merged; the Forensic Information Management System (FIMS) to deliver a laboratory information-management system for the Forensic Services Department, and the Seized Property Management Project (SPMP) to improve the management of seized property. Both projects selected the same vendor and application, and there was to be concurrent implementation, which was why they were combined.
The ombudsman had issues with the way the tenders were conducted, however, since some of the same members were on both teams. This meant that requirements from FIMS were considered in the SPMP tender. A joint tender would have helped, and would have meant better value for money, because there would have only been one contract, according to the ombudsman.
Also, after they were combined, the ombudsman said there was evidence to suggest that the projects continued to operate separately. A program director was appointed in 2010, but was of a lower rank than the individual project managers, and didn't have control over the overall program that budget had, so this didn't help, with fights breaking out over requirements and the allocation of resources. It wasn't until 2011 that a real project manager was appointed. Meanwhile, the steering committee failed to hold the project-management team to account for delivery of the program, with frequent changes to go-live dates given the go-ahead.
Originally, the business cases had intended for the SPMP and FIMS projects to be completed in 2008 and 2007, respectively, but there were delays in the individual tender processes, and since tender responses indicated that the former project was underfunded, it was paused while extra funding was sought. Once PALM was created, it was to be completed by November 2010, but it is now expected to be completed in June 2012, with delays costing $5 million in additional costs. The delays could have been due to vendor issues, but also due to optimism with timeframes. The ombudsman believes that the police underestimated the complexity of merging the projects.
Takeaways: just because two projects are similar, doesn't mean they'd be better combined.
9. HR, desist!
The business case was created for HRAssist, which intended on delivering an integrated human resources/payroll system for Victoria Police in 2005. A budget of $18.4 million was set. The implementation was to be completed by 23 October 2009. However, there were delays in approving the business case, meaning procurement was late in starting. After the tender process, the budget was revised in June 2008 to $28.4 million. After the project-design phase, an additional $3.5 million was added to the budget. By October 2009, the budget increased again, this time to $42.1 million — over double the original budget. The project was implemented in April 2010, six months behind schedule.
The ombudsman said that the business case failed to identify how complex the project would be, not going into such issues as system-interface requirements, data-migration issues or training needs. In 2009 and 2010, reviews were held, which found that the project was struggling, and also identified that the contract didn't have abatement and time-of-essence clauses that would have given Victoria Police the ability to impose penalties for poor performance by the contractor. High turnover of both the project team and the steering committee was also giving the project grief. The project found it hard to hire experienced staff, and didn't have the funding to hire experienced contractors.
The ombudsman noted that funds had been made available from Victoria Police's accumulated surplus, and had been classified as operating expense instead of capital expense, which meant that it wasn't put under scrutiny by the Department of Treasury and Finance and the Cabinet budget committee. Although the program is operational, the ombudsman said that the Police Association wrote to the chief commissioner late last year, requesting fixes for a "raft" of problems with HRAssist. The ombudsman wants a post-implementation review, as he does for many of the other projects that have already been completed.
Takeaway: it's important to make sure no clauses of a contract are excluded.
10. My HIIP is broken
The Housing Integrated Information Program (HIIP), which started in 2002, was meant to replace a 20-year-old Department of Housing system that manages client property, financial and lending information. The budget was $93 million, and it was scheduled for completion in 2004. Housing engaged a vendor in 2002, but settled the contract in 2006, after extensive delays in development, receiving $26 million from the vendor in compensation. In 2007 and 2008, Housing entered into contracts with two new vendors, one for the financial module and one for the rest of the functionality. The state's auditor-general said that the issues leading to the cancellation had included project governance, project management, contract management and risk management.
Although Housing still thought the project could be finished by 2008, and within budget, vendors believed that this would be virtually impossible, according to the ombudsman, and now the project is not likely to be finished until next year; a tenancy management module is still outstanding. This concerned the ombudsman, because the legacy system was considered past its use-by date in 1999. The current budget is $123.5 million.
One problem behind the delay was that Housing wanted all data from the old system to be transferred to the new one, which required much more data cleansing than thought. Another option would have been to have the old system's data made available in read-only format, but this was rejected by Housing; something that the ombudsman disagreed with. The ombudsman also identified that not all of the requirements laid out before the project went to tender, with delays and costly scope changes occurring as a result. Training for the system occurred before it was operational, meaning that it had to be conducted again later.
Meanwhile, housing had issues with one of the vendors, including a significant number of defects that weren't managed until 2010 via a memorandum of understanding. The low price named by one vendor on the tender also suggested to the ombudsman that it did not understand the complexity of the work. Although it was a positive monetary outcome, the ombudsman was concerned that the vendor would minimise its output to maximise its return. He felt that the department should have recognised this.
The parts of the project that have been completed received positive feedback, especially since the old system being used was a green-screen system that required staff to recall numerical codes in order to use it.
Takeaway: cheap tender responses aren't always a win.