Devil's Advocate: Joined up working

Integration is supposed to be the magic dust that makes mergers work, the glue that allows best of breed purchases - you know the hype. Martin Brampton has his doubts...

Problems can only be solved one at a time, if they can be solved at all. But then we often want to pull the solutions together somehow, and that creates another problem. In IT we usually call it integration and people seem now to be recognising what a difficult problem it can be.

Questions about integration arise in other spheres, as I find on my train journeys. Somehow, splitting the rail system into a number of operating companies was supposed to encourage competition. It is hard to understand how that is when, for instance, the only practical route from London to York is on GNER trains. You wouldn't expect to get there on ScotRail would you?

Ever since the split was made, passengers have seen a need for integration but the system seems to work against it. It is galling when your train is delayed by signal failure (presumably the responsibility of the more or less defunct Railtrack) and a connection is missed. If you ask the company operating the connection why they could not wait, the answer is that if they delayed their departure, Railtrack would fine them. Some integration!

You might think that IT could do better. After all, the problems are mainly in software, which is infinitely malleable. Yet time and time again companies have entered mergers and take-overs with bold claims of huge cost savings and no thought about how IT systems would be integrated. After the event, IT managers have bemoaned the huge difficulties faced in attempting an integration of disparate systems.

Only last week, giant retailer GUS announced that it would make enormous savings by combining the IT facilities of its Argos subsidiary with its new purchase, Homebase. Around the same time, it was revealed IT managers have wasted more money on trying to integrate IT systems than on any other area. A third of managers admitted to having lost money on integration efforts. CRM has often been cited as an easy route to throwing away money but it lagged well behind with only a fifth of managers owning up to waste.

Even without mergers and acquisitions, problems have occurred. Devolvement of IT to individual operational units was highly fashionable for some years. Now the trend seems to have reversed and the pressure is on pulling together systems that were designed in isolation. The move is driven by factors such as the fronting of all systems by call centres or, even more demanding, the exposure of systems to customers through the internet.

A major reason for difficulty in integration is that the problem is not well defined. What has to be done to integrate systems? An extreme view that has been popular in some areas is to attempt to create a model of the entire organisation. Individual systems then have to be overhauled in a way that allows them to be brought into the total model. This is inevitably a demanding and time-consuming process.

It is especially likely to fail in situations where change is rapid and unpredictable. Some large companies, especially those that deal in fast moving consumer goods, are buying and selling companies routinely. The chances of succeeding with a tightly integrated system in such an environment are small.

Aiming to avoid this pitfall, others have preferred to build a message-passing layer on top of existing systems, keeping the changes in those systems to a minimum. That can certainly help with quality of service, by reducing the dependency of one system on another. But the conceptual problems remain. Do the messages mean the same thing to the sending and receiving systems?

These are seriously intractable problems and the high profile successes are overshadowed by the many less publicised failures. Beware claims that integration will be an easy way to save money. Are you making sure that your management is not being foolishly optimistic about your ability to integrate IT?

** Martin Brampton is a director and founder of Black Sheep Research (http://www.black-sheep-research.co.uk ), an independent consultancy providing research, writing and speaking services on a wide range of business and technology subjects. Martin was previously a director at Bloor Research, and has worked with IT as a user and analyst for over 20 years. He is a frequent contributor to silicon.com's weekly Behind the Headlines TV programme and can be contacted at silicon@black-sheep-research.co.uk.