In May 1999, the WA State Revenue Department finally signed off on one of the most innovative and ambitious information technology projects ever undertaken by a Western Australian government agency. Seven years in the making and developed at a cost of some $13.4 million, the department's new Revenue Collection Information System (RCIS) is at the cutting edge of technology and provides significant productivity improvements.

Numbers of federal government agencies, various Australian state governments, a number of US authorities, and several corporations - including international management consultants Andersen Consulting - have all hailed the creation of the RCIS as a world first. But while RCIS is delivering some significant benefits, those benefits came at a significant cost and forced IS staff to learn some powerful lessons.

Not least of these were years of frustration which taught the system's architects about the limitations of high-level design, the need to mitigate the toll taken on production areas by major developments, and the dangers of listening too closely to consultants. They also learned that you can never overestimate the data-conversion effort likely to be associated with such a significant project, nor the amount of testing involved.

The RCIS is a cutting-edge system that integrates all of the department's revenue-collection operations into a single, easy-to-operate, customer-friendly system. So successful has RCIS proven that the State Revenue Department has already sold licences for the technology to New South Wales, Victoria and Queensland. It has recouped $2.6 million in just over 12 months and reached 85 per cent of the Australian population. The system won a Gold Award at this year's Government Technology Event.

State Revenue says RCIS has the potential to generate significant benefits and an ongoing income stream for Western Australia for many years while continuing to lift productivity at all levels of government.

From a financial perspective the department gains in two ways. It benefits from the revenue it is generating by licensing the intellectual property rights of RCIS interstate. It also benefits through charging a "fee for service" to undertake collection activities for other agencies. And it believes the RCIS project will allow other government agencies to cut their operating costs by divesting non-core collection functions, thereby benefiting the entire community.

RCIS also allows the department more resources and time to address the compliance areas of revenue collection. This has the potential to further increase revenue and provide a much more equitable system for the entire state. The system is also directly responsible for the transformation of the State Revenue Department. "State Revenue management has completely turned the organisation around from a classic ‘working in the business' to an innovative ‘working on the business' approach," says IS manager Carmelo Naso.

The development has helped the department move from a highly process-focused department to a more commercial, customer- and outcome-focused organisation, Naso says. Development, implementation and licensing of RCIS has highlighted significant leadership within the department, he says, and has provided a role model for other state government agencies looking to develop leading edge technology.

Overall the RCIS project has significantly increased the department's profile and credibility within government and will allow it to pursue ongoing marketing opportunities within Australia and overseas well into the new millennium. "No longer is the department limited by an inflexible system and traditional funding sources," Naso says. "It is now infinitely more flexible, is looking globally to market its technology. and is taking on the collection of new revenue lines within government."

Integration Guaranteed

Developed to eliminate stovepipes within the department's revenue streams, the RCIS has successfully replaced six totally disparate and ageing revenue collection systems, all developed in Natural Adabas on an IBM compatible mainframe computer.

Naso says that at the time development began there were six systems that were becoming obsolete, were not Y2K-compliant, and would clearly either need to be upgraded or replaced. These systems covered land tax, payroll tax, stamp duty and receipting, debt management, records management (corporate) and returns information. The high cost of ongoing maintenance of these separate systems was also of significant consideration.

There were two alternative solutions to the department's IT requirements: either upgrade each of the six existing systems separately, or create a hybrid mainframe system. Careful research showed both would have proven more costly and less effective than creating one new fully integrated system. In addition, a fully integrated system carried with it the promise of another significant advantage.

Naso says the existing infrastructure gave State Revenue little flexibility. There was no network in place, no comprehensive records management system and no electronic communication. Personnel were tied to one area of the business, with little scope for movement. "With six totally different systems, there was a fractured environment within the organisation, with a lack of communication and no standardisation," Naso says.

The RCIS was designed to facilitate development of an entirely new management approach and to allow the entire department to operate as a single entity. By pooling the department's resources and implementing a team structure, the RCIS allows all officers to deal with all revenues and eliminates the learning curve that was previously involved in moving from one revenue stream to another. It also allows for cross-pollination of people and ideas. Equally, being able to work as "one team" has given staff a new sense of satisfaction, and vastly improved internal communication.

The system involved extensive re-engineering to create a more effective and flexible working environment. This ensures all employees can access and use the fully integrated revenue collection information system to work on any revenue line. "Originally a particular revenue officer would have only been in a particular division. To go to another area and learn a new revenue, not only did they have to learn the specifics of the revenue, they had to learn the specifics of that system," Naso says.

"We're slowly going to a situation where a revenue officer will be a single point of contact for a particular client or a particular company, which can deal with that revenue officer for all their needs across all the revenues," he explains. "And having an integrated system with the same look and feel across all revenues means we can throw resources around the department a lot better to handle the peaks and troughs of the various revenue lines."

Having a fundamental core common to all revenue lines also gives the government the flexibility to introduce new revenue lines with a minimum of fuss. "It's just a matter of developing the assessing component of the new revenue line and then bolting it into the core," says project consultant Andrew Hamilton. "That gives us a lot of flexibility in terms of collecting any new revenues."

Hamilton says the installation improves customer service and allows State Revenue to offer access to a far more effective, personalised and user-friendly system. It also serves to provide the public with a greater understanding of the policies, procedures and legislation underpinning the revenue systems, while still giving them a single point of reference for all enquiries.

"RCIS provides a single client-view across all revenue lines, which will enable both departmental personnel and customers to monitor all of their taxation requirements."

A Workers Revolution

Work practices within the State Revenue Department have undergone a revolution over the last three years through the introduction of RCIS. Previously the department had a rigid management structure, with limited objectives and little interaction outside the agency. It is now a commercially active, team-based and highly flexible entity that has created a challenging and rewarding workplace.

More than two-thirds of State Revenue's 200 employees played a role in the development of RCIS, with a core of 50 providing significant input on the project. Naso says that since then all have seen improvements in their work practices and job satisfaction, and now have the ability to move freely within the organisation.

Introducing RCIS has also boosted the professional development of the department's employees. The system has become the cornerstone to creating a challenging, enjoyable and rewarding workplace. Naso says the pioneering workplace relations practices introduced into State Revenue ultimately proved crucial to the success of the RCIS project.

All the technical specialists used on the project had high levels of knowledge of State Revenue's business and technology. However, with their IT skills in demand because of Y2K, all were tempted to leave the project. In a bid to stem the movement of technical staff out of the organisation, State Revenue became the first government department to introduce a "Golden Handcuff" scheme for IT staff. Under the scheme team members received a bonus of around $1000 a month - all but unheard of for public servants at the time.

Naso says the scheme was so successful that up until the end of release three, four of the original six staff members remained on board, with two of them having left before the introduction of the bonus scheme. "Once we introduced the bonus scheme we kept them pretty much up until the death knock, which was pretty good given that their skills were in great demand."

No BPR Thank-You

The extensive re-engineering and introduction of electronic workflow has given all employees a new skills set and allows them to work on any revenue line. Yet all the innovations were achieved without the department undertaking a formal business process re-engineering (BPR) project. Naso says his examination of some organisations that had implemented a formal BPR process suggested BPR was difficult to do before a new system is in place. "It's sort of a chicken-and-egg situation."

Instead State Revenue engaged in a series of mini BPR exercises on individual core functions, looking at how it could extend these across all revenue lines. "Now [that approach] probably cost us in terms of timeframe in the overall project. However, it certainly made for a very focused outcome, because people had to really come up with suggestions and solutions to particular issues quickly without drawing the thing out. And it focused our attention well and truly, when we needed to come to one of these situations," Naso says.

Hamilton says State Revenue also learned a lesson about the need to provide "backfill" in production areas. "We used some of our own people on the project and we certainly didn't allow enough backfill in our production areas, so the project itself was a huge drain on the organisation, to the extent that it was affecting our production areas.

"If we were to do it again we would make sure that we had the necessary backfill into our organisation to make sure we didn't impact the production area," he says.

Significant Milestones

In partnership with a number of local and international companies and built on a client/server architecture based around Oracle Forms and an Oracle database, the RCIS runs on a Unix system and Windows NT network. The project took seven years from start to finish and involved a number of significant milestones.

Before development began RCIS project team members assessed and reviewed a similar system being introduced in Victoria. They also gained valuable input on overseas and interstate experiences from two of the consulting firms involved in the project. One was Andersen Consulting, which had previously worked on the development of an Inland Revenue System in New Zealand and taxation systems in the Philippines and Singapore. The other was Interim Technology, which had been involved in the development of a similar revenue collection system in South Australia.

After a planning phase running from 1992 to mid-1994, State Revenue spent six months in 1994 doing high-level design before moving into a detailed design and construction phase from September 1995 to April 1999. But while the results were ultimately satisfactory, that's not the way Naso would approach such a development in the future.

"We did the six-month high-level design on the advice of our consultants and then tendered again for the low-level design and the construction," he says. "The trouble is we found that the initial high-level design was perhaps a touch too high-level . . . When you actually took it down to the lower level, the level of detail blew out exponentially; it was actually a much larger piece of work than was perhaps originally anticipated."

To avoid such problems in future, Naso says, he would probably opt for a much higher level of specification initially, followed by a more open low-level design and construction phase.

"I don't think I'd do a fixed price. I would certainly not go into it with a high-level design, which is what's been quoted on, because then you get into a whole cycle of needing to deal with scope variations more often than you need to. It also leads to users perhaps not getting their expectations met, because what was envisaged in the high-level design has translated into something completely different," he says.

The detailed design and construction phase of the RCIS project was completed in three releases.

Release One included building the Stamp Duty Document Based Assessment Module and the core modules to support the department's stamp duty document-based assessment business. The development of this release commenced in November 1995 and went into production in March 1997.

Release Two included building the Return Based Modules and modifying the core modules to support the department's returns-based business. The development of this release commenced in November 1996 and it went into production in August 1998.

Release Three included building the land tax and other minor tax modules and modifying the core modules to support these areas of the department's business. The development of this release commenced in October 1997 and went into production in April 1999.

Naso says the complexity of the system led to some serious integration issues. "Because it was a large system different teams were developing different aspects. Occasionally the line of demarcation would become a little bit blurred as to where one team picked up and another team lost a particular piece of functionality.

"I wouldn't say they were turf wars, but when it came to testing it was clear that there were holes in the middle or occasionally they overlapped. So it just took a bit longer then we anticipated to iron out the bugs."

Testing also took a long time to get right and, equally, took much longer than anticipated.

"We seemed to have a lot of problems in going through a testing cycle," Hamilton says. "We underestimated that quite considerably, starting off by estimating that a testing cycle for a particular release would take us in the order of six weeks, when it ended up taking us somewhere in the order of three to six months."

And Naso says the quality of the information the department converted "hurt us quite badly". He says knowing data quality would be a problem, the department started data conversion the very day it started system development.

"One very important lesson with a project this size, where you're moving fairly large systems, is not to underestimate conversion," he says. "We certainly didn't do that; but for some of the systems that have had a reasonable amount of change over the years, we've found the quality of the data - certainly going back past four or five years - was fairly questionable. In the end the decision was taken to convert it. However, if I had my time over, I think I would have left a lot more of it behind, and gone with the stuff that was better quality."