Citigroup’s doomed IT strategy

By Felix Salmon
May 22, 2009

Why do I get the feeling I’ve seen this movie before?

Citi had originally estimated it could save $3bn over three years by rationalising its operations and technology functions, which employs 140,000 people including 25,000 software developers – more than many IT companies.

But recent progress in reducing overlaps between systems, and linking IT infrastructure across businesses that had previously run individual systems has prompted Citi executives to sharply increase its cost-saving target.

I’m going to go out on a limb here and say this ain’t gonna happen. The first thing any new management team or management consultancy does, on looking at the nightmare that is Citi’s web of incompatible IT systems, is decide that they should be unified and rationalized. And the second thing they do is blame Sandy Weill for the current mess. (Yes, that’s in the article too.)

The problem, of course, is that these legacy systems — all of which support vital operations and databases — won’t just die and go away. Citi’s IT honchos have the same problem that Microsoft’s do: everything they do needs to be backwards-compatible with everything that went before, or else they have to build some brand-new system which nobody currently knows how to use and into which all the old data can be seamlessly imported. Both options are so complex and difficult as to be, practically speaking, impossible — especially in a dysfunctional company like Citigroup which is simultaneously putting significant resources towards splitting itself up as opposed to integrating itself.

It’s a rare sign of good news within Citigroup that executives reckon there has been “recent progress” on the IT front. But extrapolating from that progress to billions of dollars in new cost savings is bound to prove overly ambitious.


We welcome comments that advance the story through relevant opinion, anecdotes, links and data. If you see a comment that you believe is irrelevant or inappropriate, you can flag it to our editors by using the report abuse links. Views expressed in the comments do not represent those of Reuters. For more information on our comment policy, see

12 years ago, my bank decided not to teach new traders a specific operations system, because it was on its way out within 6 months at most.The system was decommissioned about 6 weeks ago, only 11 1/2 years behind schedule. Its not that our IT group was stupid, it was just so damn complicated…

Posted by nicolas | Report as abusive

yea, unless its a new operation, most older banks and financial institutions have legacy systems that they cannot part with easily- just to maintain those is quite expensive and even more expensive to replace. Mostly these things have front-end clients made by IBM or something … hey, it keeps the cobol developer around for a while.

Good point, and not at all limited to Citi. In fact, this is likely a generational issue – the current decisionmakers at most large financial institutions grow up with the old technology, but didn’t really understand it.The next generation is starting to have more power to make decisions, but would like to ditch the legacy systems but still have to work thru the old fogeys who don’t want to spend the money.Over the next two decades, that should begin to change somewhat…

Posted by Eric Dewey | Report as abusive

Felix, I have been following a company called ants software, Accordingly, they seem to have the answers to the questions you are doubting. Whether true or not, since I’m not a techie, I can’t confirm. However, they do have customers who are verifying the integration probably is no longer a problem, and was not a major issue.Whether or not they could help citi or others of such size is unknown to me. I found the concept interesting enough to research. I would hope you do the same.

Posted by Mc | Report as abusive

Legacy systems are always an awful mess. It’s pretty rare to find a project manager who can design a new project around those problems, though.What’s more plausible is incremental change. If Citi can just upgrade one group at a time to a newer, low-maintenance system, they can probably get their entire company system back on track.

Having seen IT struggle with this problem in a much smaller bank, I can appreciate why perhaps Citi IT are failing to make the expected impact on costs.When you get to a certain level of complexity it becomes exponentially more difficult to exert close control over the way every business process is supported by IT services. The instinct of managers (and IT is no different) is to impose a centralised command-and-control environment that subjects all proposed changes to ordeal by business case. This often relies on purely fictional benefits invented to pass the Project Management Office’s rules, benefits that are imperfectly tracked after implementation and reliant on market conditions anyway.All this was a problem in my bank, a fraction of Citi’s size, but it seems to be the industry standard way of managing IT. Far better in my opinion for a CIO to try to instigate a cultural change in the IT organisation where people are rewarded for doing the right thing in the right way. Implementing industry-standard interfaces, using common securities data and counterparty data and reuse of components should all be factors in setting IT bonuses.Another example: getting the business to buy into Agile project management is easier than getting IT’s accountants to accept the same thing (yes, corporate IT has accountants coming out of its ears) but how else can you delegate decision-making to the people with the information and experience to make the right choice of solution? If the project teams can’t make the right decisions then Citi IT is screwed anyway.The virtuous circle created by a clear technology vision at CxO level and properly delegated authority to individual project teams has probably never been tried on such a scale and it possibly never will, which is one of the reasons why organisations the size of Citi are fundamentally doomed.

Mega sized companies suffer the same project mistakes of governmental projects, namely:- Risk Analysis Was Limited In Depth, and…- Top-down Planning With Little Input From Those Working On The Project, and (last but not least)…- Timeline is not realisticI’m sure had whoever was involved in managing that project studied made his/her homework about supporting the legacy systems, the project would’ve been in a much better shape. But then again there is little accountability in such organizations.