Cracks appear in Revenue's flagship outsourcing deal

A report for the NAO suggests the Inland Revenue's deal with EDS is less than perfect, Tony Collins reports

A report for the NAO suggests the Inland Revenue's deal with EDS is less than perfect, Tony Collins reports

For six years the public face of the Inland Revenue's 10-year partnership with Texas-based Electronic Data Systems has worn a smile. But now it is beginning to crumple at the edges.

In 1994 the £1bn contract - now a £2.4bn contract - was not only central government's biggest computer outsourcing deal, but the largest ever instance of contracting out by a department of state to the private sector. It was held up as a model for the rest of the world.

The theory was this: if Britain was willing to contract out the systems that collect the nation's tax revenue, without which a country could collapse, did this not show absolute trust and faith in the ability of the private sector to manage the administration of a key central government department?

Indeed, EDS invited governments around the world to see how well the partnership works. So important was the deal that it was written into the contract that the director general of the Revenue and the chief executive officer of EDS should meet every six months in Texas or London.

Various IT executives at the Revenue were promoted on the strength of the privatisation. More than 1,900 IT staff were transferred in two tranches to EDS.

But, although the deal is regarded generally as having successfully blended mutual trust and professionalism with hard-nosed negotiations that keep both parties in a constant state of respect and noble tension, the relationship is also beset with the sort of problems that one would expect in any partnership that appears on the surface to be as near perfect as it could be.

That the problems have now reached the public domain is thanks to a 77-page independent report commissioned by the National Audit Office from Lorien, a company that has specialist knowledge of public sector procurement.

Nobody who has read the report can disagree with its conclusion: "The Inland Revenue's strategic partnership with EDS has, by comparison with many other major projects in the IT field, been successful."

But it is in the detail that one can see how even a highly experienced IT user such as the Revenue can run into problems. Sometimes these are as basic as failing in projects because end-users were not consulted sufficiently.

The Lorien report is not an all-encompassing look at the contract. It concentrates on the processes for commissioning new work from EDS. "Our study did not examine EDS' performance in delivering the service after completion of the development phase," it states.

Among its findings is a disclosure that it is difficult to measure and prove that the contract has delivered value for money. Lorien also found that there is a real risk of lock-in, which may deter the Revenue from awarding the contract to another supplier (see Computer Weekly, 30 March).

Lorien also noted that, although the Revenue has an IT strategy which seeks to standardise all new work - all developments should be based on open, not proprietary, systems, for instance - the strategy's boundaries are sometimes bent to meet the demands of deadlines and resources.

For example, the pressures to complete the tax self-assessment systems in time has led to EDS and the Revenue departing from the strategy to adopt only open systems. It was expeditious to incorporate the existing ICL VME-based mainframes into the project.

"The target dates for introducing self-assessment would not have been met if a completely fresh start had been made," said the report.

"Departures from the department's strategy have occurred and could still occur," if the Revenue has to "meet legislative or other deadlines", Lorien warned.

This may be surprising, given that ministers never make policy or budget announcements without first checking with the Revenue to see if IT systems will be in place in time to support such changes.

This suggests that either ministers have knowingly compromised the IT strategy because they wanted to introduce policy changes quickly, or the Revenue has not plucked up the courage to urge ministers to delay a policy change so that new systems could be built rather than tacking a module onto the existing archaic architecture.

As well as departures from the IT strategy, there were compromises in the procedures for commissioning new work. These compromises were due to "insufficient levels of resource or expertise" or "constraints on development times".

Lorien also found that the Revenue (or rather the taxpayer) sometimes had to pay more than was necessary for systems because they were inadequately specified. This happened with the Integrated Repayments System, which in part serves financial intermediaries. Lorien said the work on the project was "93% more than had been estimated at the outset".

It was noticed too late that the specification did not identify the need for the software to calculate the tax repayment due. This was "fundamental" said Lorien which added, "Had the proposed specification been discussed with users the omission might have been identified."

Another project, an integrated debt management system, was increased in size and complexity partly because, in the words of the report, "The original specification had been developed without consultation with users in the Inland Revenue office network."

It is surprising that an experienced IT practitioner such as the Revenue would forget or ignore such a basic project management tenet as consult users.

On the other hand, a construction industry scheme that is currently underway meets the original estimates - so far. In this case there has been "very clear specification" and adequate time has been allowed to prepare it.

There are several areas in which the Revenue has been making some basic mistakes. Sometimes, after the completion of a project it did not check whether there was a difference between the original estimated and actual cost, or whether the features in the delivered software were those that had been specified originally.

A failure to set key performance criteria for new work delivered by EDS prevents the Revenue from obtaining a reliable view of the quality of EDS service delivery says the report.

Lorien also highlighted the risk that the close joint working could "blur roles and responsibilities" and reduce the ability of Revenue executives to evaluate EDS proposals objectively. However, this risk has so far not crystallised.

Indeed, the Revenue succeeded in negotiating a reduction of £17m in EDS' estimate for the Infrastructure 2000 project to supply and install more than 50,000 Windows NT-based workstations.

But Lorien also said that there are too few experienced people at the Revenue to provide an overview of any new work that could increase the potential for lock-in.

Also there are only "very few people" with critical and perhaps "irreplaceable knowledge." In key strategy areas there have been "no significant changes of staff since the start of the contract".

The National Audit Office, which commissioned the report, said, "These findings are of concern, particularly as the risks of managing EDS successfully have been increased by the Revenue's new responsibility for Andersen Consulting's £143m National Insurance Recording System."

The all-party Public Accounts Committee will meet next month to question civil servants at the Revenue on the report.

Read more on IT risk management