The BBC DMI project - what went wrong?

The BBC and its former CTO have engaged in tit-for-tat allegations over its failed £100m digital media project. But who was right?

It is a situation familiar to many IT leaders. When business users see the result of a software development, they say it’s not what they wanted and it does not work. The IT team, in return, says the users kept changing their minds about the requirements. It’s an argument as old as corporate IT itself.

But in the case of the BBC, it turned into a very public row and cost the corporation – and licence-fee payers – nearly £100m.

The past week has seen tit-for-tat accusations between former BBC CTO John Linwood – sacked as the scapegoat for the failed Digital Media Initiative (DMI) to digitise video production processes – and the broadcaster itself.

In advance of a hearing in front of MPs on the Public Accounts Committee (PAC), first Linwood responded to BBC statements about its reasons for scrapping the project and his subsequent dismissal as the executive in charge. Then the BBC responded to Linwood’s response. And finally Linwood responded to the BBC’s response.

Linwood then defended his response in front of MPs, before BBC executives subsequently responded to his response. Confused? This argument will run and run yet, as Linwood is taking legal action against his former employer, and it may take a court case to determine where the blame lies.

Poor engagement

Computer Weekly has seen all of the submissions made to the PAC, and a constant theme is the poor engagement between IT and the business at almost every stage of the project.

“Throughout the project, the team informed me that the biggest single challenge facing the project was the changes to requirements requested by the business,” said Linwood.

He cited a report produced with the help of external consultants, which concluded: “DMI has also suffered from a lack of clear and consistent direction with respect to the business requirements and priorities. Business representatives were not empowered to make decisions on behalf of the whole production community, and regular reprioritisation of requirements has impacted the efficiency of programme delivery.”

The BBC, however, continues to insist that the technology developed by Linwood’s team simply did not work, and that there was no point “throwing good money after bad” in an attempt to save the project.

“Any change in business direction and conclusion that the original DMI vision was no longer valid can be attributed to the lateness of delivery,” said the corporation’s PAC statement.

Delays in the project, according to the BBC, meant that off-the-shelf software available on the market overtook the capabilities expected from DMI.

So where did it all go wrong?

Agile, or not agile?

Problems had already been identified by the time the project was brought back in-house from its original contractor, Siemens, in September 2009.

DMI timeline

October 2006: Approval of £2.8m for initial mobilisation of Digital Media Initiative (DMI)

March 2007: Approval of £6.6m for design of DMI

January 2008: BBC Trust approves £82m total budget

February 2008: Siemens wins £79m fixed-price contract to design and deliver DMI by May 2009

July 2009: BBC and Siemens terminate contract

September 2009: BBC brings project in-house with target date of completion for February 2011

June 2010: BBC Trust approves wider roll-out with revised budget of £133.6m

August 2010: Procurement delays push back DMI timetable by five months, with final delivery date set at July 2011

January 2011: NAO reports BBC has made good progress on straightforward parts of system, but faces challenge in the following stages

February 2011: BBC tells Public Accounts Committee it is on track to deliver DMI technology by summer 2011

February 2012: The project management office grades the status of DMI as ‘red’ and suggests termination or re-evaluation of the project

May 2012: Executive board requests review of costs, benefits and timetable. Whistleblower contacts the BBC Trust saying NAO, PAC and the trust may have been misled about the DMI’s progress

November 2012: Most work on DMI is stopped pending review

May 2013: Programme permanently stopped and CTO John Linwood suspended

Source: National Audit Office, based on various published and unpublished sources provided

The revised project was to be delivered using agile development methods – an approach that requires close and regular interaction between IT and business users for success. As it turned out, the business could not deliver on its side of the agile pact.

“Agile development was agreed upfront, not developing the whole system end-to-end from day one,” Linwood told MPs.

“Several months later [the business] decided it didn’t want to do that, but would wait for the full functionality; it didn’t want to continue down that path,” he said.

“The business objected to the [agile] approach. Small incremental releases would allow the business to get hands-on with the technology so it would not need to wait until the end of the programme. The business then said it didn’t want to spend time testing, but wait until large incremental pieces [were completed].”

As a result of scrapping agile, the IT team went ahead and developed large chunks of functionality ready for review by users.

According to Linwood, those users were still not happy. In one of his submissions to the PAC hearing, he cited two examples of business requirements undergoing significant change that caused additional work for the development team.

In one case, he claimed that users wanted a function to produce a "rough cut" of video output, which was subsequently developed, only to be told that those users now wanted to use an off-the-shelf product from Adobe instead. Once the IT team had integrated that Adobe product into the DMI software, users then said they wanted to use a different Adobe product altogether, said Linwood.

In another example, he quoted several changes to requirements for search functionality, at each stage requiring "significant and challenging" extra work, according to Linwood.

"These changes caused delays in delivery, which were appropriately reported at every stage," he said.

Under questioning at the committee hearing, Linwood added: “They wrote off more than they should have done. They wrote off software that was working and infrastructure that was working. They were written off because the business decided not to use them.”

A major business change

He described the project as “a major business change” that was dependent on users changing the way they did their jobs. It seems apparent from evidence on both sides that the business never fully understood the scale of changes required.

The BBC continues to point the finger at IT for failing to deliver, although former BBC Trust member Anthony Fry acknowledged that trustees did not have sufficient knowledge around the technology and that external consultants should have been appointed to keep a closer eye on the IT aspects.

Somewhat ironically, BBC director of operations Dominic Coles told MPs that the broadcaster plans to change the way it delivers technology in future by “chunking it up in definable separate projects” – in other words, agile by a different name.

The National Audit Office (NAO) report into DMI highlighted failures of governance and said that BBC executives “did not have sufficient grip of the programme”. The NAO also pointed to the lack of a senior responsible owner (SRO) from the business, and stated that the BBC’s management of the DMI was "focused more on the technological aspects of the programme rather than enabling BBC-wide change".

Coles told MPs that the BBC will, in future, appoint an SRO for every major project.

Throughout the project the team informed me that the biggest single challenge facing the project was the changes to requirements requested by the business

John Linwood

Former BBC CTO John Linwood

Even now, Linwood and the BBC disagree about what was finally delivered. The CTO said that one tool, called the metadata archive, is in use by 5,000 BBC users. Coles responded that the metadata archive has only 168 active users because the system is “incredibly clunky” and “difficult to operate”.

“It can take up to 10 times longer to use the archive database than the legacy system,” he said.

The BBC said it will now use off-the-shelf technology to support its production processes. Linwood countered to MPs that: “If the BBC didn’t do DMI, each department would go and buy each system and it would be more money.”

In his written statement to the PAC, Linwood said: “Commodity production tools do not offer the functionality that the DMI project, with all its integrated parts, offered. The use of commodity tools would require a significant compromise by the business in their requirements, as they do not integrate with BBC systems, cannot be shared across all production areas, and don’t acquire production metadata through the production process.

"DMI did all these things while still allowing the production teams to use the third-party commercial editing tools they wanted to.”

But Cole said the BBC no longer needed “one, integrated, all-works-or-nothing-works plan".

He said, they said...

DMI was not the first major IT project to founder on the rocks of disagreements between business and IT leaders – and it will not be the last. But few IT projects have to bear the public scrutiny that the BBC does, and the embarrassment of seeing its dirty technology linen washed in public.

John Linwood may yet have his day in court. But many of his IT leadership peers will recognise and sympathise with the problems that led to his departure.

Read more on IT outsourcing

Data Center
Data Management