Some of those who read “Crash – 10 easy ways to avoid a computer disaster” may remember a warning that buying an IT system on the basis that it works well in another country and can therefore be adapted to the UK’s needs, is flirting with disaster.
First published in 1999, Crash said,
“There are graveyards of computer projects that began life as a simple adaptation of a package used elsewhere in the world.”
One example at that time was the failure of the London Stock Exchange’s Taurus project.
Says the report,
“The belief that most of the i6 system could be based on an existing IT system proved incorrect.”
It became clear well into project that
“a virtually fully bespoke system was required”.
The plan was for i6 to replace 130 paper-based processes and IT systems but on 1 July 2016, after many well-publicised difficulties and delays, the Scottish Police Authority and Accenture agreed to terminate the i6 contract.
Police in Scotland had chosen Accenture’s bid in 2013 largely because it had successfully implemented a system for Spain’s Guardia Civil police service.
To its credit Accenture refunded all the money the police in Scotland had paid for the i6 system, £11.06m, plus a further £13.56m – but Audit Scotland says the failure of the project …
“means that some of the benefits that should have arisen from implementing it, have been, at best, delayed. There was a need to modernise police ICT systems six years ago when the procurement of i6 began. That need has not been met. Police officers and staff continue to struggle with out-of-date, inefficient and poorly integrated systems.
“This also hinders how Police Scotland interacts and shares information and intelligence with the other parts of the justice system. There is an urgent need to determine what the next steps should be…”
The lessons are clear from the report:
- Don’t buy an overseas system without realising that it’ll need to be built almost from scratch for the UK. The ideal is for the business processes to be greatly simplified and adapted to fit a tried and tested system, not the other way around. Audit Scotland says the police programme team and Accenture believed that the majority of the i6 system could be based on an existing IT system that Accenture had developed for Spain, with the remainder being bespoke development work. But there was an “over-reliance” on Accenture’s work for Guardia Civil”.
- The “waterfall” systems development contributed to the fact that Police Scotland “only discovered the true extent of problems with the system when it was delivered for testing”. Waterfall meant that Accenture produced the software in distinct phases, in a sequence resembling a waterfall. Once a phase was complete, the process moved to the next phase – and no turning back. “It meant that all of the design, coding and construction of i6 would be completed before Accenture released it to Police Scotland for testing. Police Scotland would pay for each phase when it was completed.” [Agile, on the other hand, is a “test and see” approach and is far more flexible. It can adapted according to what the end-user needs and wants, and changes in those needs and wants.]
- Don’t trust the demonstration of a waterfall system. The demo may look great but rolling it out successfully across various regions may be a different story. Accenture had demonstrated i6 but much later, after a period of testing, the i6 programme team reported to the programme board in August 2015 that there were: critical errors in the technical coding, flaws that Accenture was unable to resolve as quickly as expected, serious concerns about the criminal justice module, which did not comply with the Integrated Scottish Criminal Justice Information System data standards, errors in the search and audit modules and “problems around the limited functionality in the administration module”.
- External assurance reports may tell you that you have complied with good practice and they may give you detailed praise for your attention to detail but they probably haven’t looked at the big question: will the systems ever work? Audit Scotland said external assurance reports such as the Scottish Government’s “Gateway reviews” suggested improvements but “raised no major concerns”. Throughout the course of the i6 programme, most of the external reviews suggested that delivery confidence was either amber or green.
- If the plan is for a waterfall development, doing everything by the book before a contract is awarded will not guarantee success, or even make it more likely, if you haven’t asked the big question: Is this ever likely to work given the complexities we don’t yet understand? For officials in Scotland, everything went smoothly before the award of contract: there were even 18 months of pre-contract discussions. But within weeks of the contract’s start, Police Scotland and Accenture disagreed about whether the proposed system would deliver the requirements set out in the contract. Soon there was a “breakdown in relationships and a loss of trust between Police Scotland and Accenture that never fully recovered,” said Audit Scotland.
- The supplier may be just as optimistic as you. “As the design and development of i6 progressed, it became apparent that Accenture would need to develop significantly more than had been originally anticipated. Despite delays and serious problems throughout the lifetime of the programme, Accenture provided regular assurance, in the face of strong challenge, about their confidence in delivering the i6 system. This assurance proved misplaced.”
- When planning a waterfall system that has complexities and inter-dependencies that are not fully understood at the outset, expect ever-lengthening delays and projected costs to soar. At one point Police Scotland estimated that the level of effort Accenture would require to complete i6 was around eight times greater than the resources Accenture had estimated when signing the original contract. “The i6 programme team believed that the functionality of Accenture’s solution did not meet the requirements it had agreed in the contract. Accenture maintained that Police Scotland had not specified a detailed description of business requirements. This issue had not emerged during months of pre-award dialogue. Accenture also believed that it had set out clearly what its solution would do and maintained that Police Scotland, as part of procurement process, had accepted its qualified solution. A dispute followed about the interpretation of the contract requirements. Police Scotland argued that, after months of competitive dialogue, the requirements of the i6 system were well-defined, and that in line with the contract, these took precedence. Accenture argued its solution had precedence and that Police Scotland was trying to extend the scope of the programme. Accenture stated that, to meet Police Scotland’s interpretation of requirements, it would require more time and money.”
- As soon as things start going badly awry, stop and have a re-think. Cancel all existing work if necessary rather than plough on simply because failure isn’t an option. Above all, take politics out of the equation. The Scottish Police Authority was anxious about i6 being seen to be a success after the failure of a previous police ICT project in 2012 – the Common Performance Management Platform. At the same time the i6 programme was “extremely important to Accenture at a global level. “This may have led to misplaced optimism about the prospects of success and unwillingness to consider terminating the programme,” says Audit Scotland.
- When things start to go wrong, the truth is unlikely to emerge publicly. Even those accountable for the project may be kept in the dark. “Police Scotland were cautious of commercial sensitivities when providing assurances on i6 publicly. The Scottish Parliament’s Justice Sub-Committee on Policing held a number of evidence sessions with the Scottish Police Authority and Police Scotland to explore progress with the i6 programme. In March 2014, the Sub-Committee expressed frustration at the lack of information about the problems with the i6 programme that had been ongoing since August 2013. Police Scotland did not disclose the severity of the issues facing the programme, nor was it overly critical of Accenture. This may have reflected a desire to maintain relationships with Accenture to keep the programme on track or to maintain the commercial confidentiality of the contract.”
“As the report acknowledges, the scope and the complexity of the solution for i6 increased significantly during the project. This was driven by the client. There were challenges and issues on both sides, but we worked closely with Police Scotland to review the programme and recommend revised plans to successfully deliver i6.
Despite our best efforts, it was not possible to agree the necessary changes and we mutually agreed to end the project.”
In May 2017 Audit Scotland is due to publish a report that summarises the lessons from a number of public sector ICT projects it has investigated.
Some of what i6 was intended to cover …
Tis a pity officials in Scotland hadn’t read Crash before they embarked on the i6 project – or if they had, taken more notice of the dangers of assuming a system that works overseas can be tweaked to work in the UK.
We commend Audit Scotland for its expert investigation and a fine report.
Clearly the failure of i6 is not entirely Accenture’s fault. The project was commissioned on the basis of assumption and when things went wrong politics intervened to prevent a complete stop and a fundamental re-think.
Fatally, perhaps, there appears to have been no discussion about simplifying police administration to make the IT more straightforward. If police administration is so enshrined in law that it cannot be simplified, officials would have to accept before awarding the contract that they were buying an entirely new system.
The UK armed services simplified volumes of rules and practices before it introduced pay and personnel administration systems. It was hard, inglorious work. But simplifying ways of working first can make the difference between IT success and failure.
Waterfall approach damns £46m Scottish police system – Government Computing