Take measures

If you focus solely on cost when outsourcing, according to statistics, your deal may be for the high jump. By measuring...

If you focus solely on cost when outsourcing, according to statistics, your deal may be for the high jump. By measuring performance, the deal stands a better chance of lasting its term. Mark Vernon reports

People outsource to buy improved performance - right? If a business seeks the services of an Internet service provider it does so to ensure that connectivity is fast and reliable. If a corporation outsources its datacentre it does so in part to cut costs but also to gain new flexibility with data processing.

But do these organisations ever check on the quality of the services they are receiving, beyond the cash savings made? Do they revisit past outsourcing deals to see whether they have worked and how they might be improved? In short, do they monitor the outsourcer's performance?

The answer for the most part is no, few firms measure return on investment. Julie Giera, vice-president and research leader at Giga Information Group, reports that only 10% of organisations monitor the performance of their outsourcing partners. Further, while many companies choose to outsource to improve performance, they then negotiate a contract that will drive down costs and ignores performance as a deliverable.

The monitoring of performance should be a good enough case, but if that is doubted, Deloitte Consulting has found that 25% of outsourcing deals fail in the first two years and a further 25% fail within the next five years, in terms of delivering on expectations. Or to put it another way, indiscriminate outsourcing, in the sense of simply passing whole areas of IT support to a third party, has only a 38% success rate.

On the other hand, and excuse just one more statistic, 77% of selective outsourcing deals work. And herein lies the key to an improvement. If you know what you want from an outsourcer and get them not only to deliver but also be seen to deliver, then sure enough the relationship will be a success. The problem is that it is not as easy as it sounds.

"One of the first issues is that such control is not really in the client's hands," says Giera. "Suppliers choose the metrics they will be measured by, since they know they can meet them. But there could well be a gap between that and what the business requires, and so the performance actually being delivered is compromised."

Typically what happens is that performance itself is barely considered, what stands in its place is cost. "A company believes itself to be happy if the costs go down. However, we believe that companies should be outsourcing to ensure that service levels go up."

So what kinds of stones can companies expect to turn if they monitor what is going on? Mercedes Benz Credit, the leasing side of the car maker's operation, had purchased the services of a T1 network cable at a cost of $1,100 (£775) per month when its imaging system seemed to have slowed to a virtual standstill. Surely more bandwidth was all that was required?

"Instead of improving the situation, things only seemed to get worse," says network engineer Chris Beyer. "Clearly the imaging application along with Lotus Notes and database replication could cause bandwidth issues. But we didn't want to jump to conclusions and throw more bandwidth at the problem. We had to look a little deeper."

First, Beyer installed a wide area network (Wan) probe from NetScout and a sniffer from Network Associates. "We got a lot of data from the NetScout probe, but not the kind of information we could easily present to non-network engineers," says Beyer.

But the Wisewan product from NetReality provided the kind of reports Beyer was looking for. "Contrary to popular belief, it showed us that no application was taking over the bandwidth. At no time were we reaching critical utilisation on any connections."

Armed with that information, Beyer used a Wan sniffer to record how long it took for the lease-tracking application to check on its database and respond to a client's enquiries. By seeing where the delays were occurring, Beyer tracked down the real culprit: a database server. "Without Wisewan it is likely we would have thrown bandwidth at the problem once again, and once again it would have solved nothing," he says.

Alternatively take the scenario of Web hosting. At first, a company may consider outsourcing its Web infrastructure because it does not want to carry the cost of the hardware and is inexperienced at maintaining servers. However, since the company is a retailer, most of the Web site's business is done in the run-up to Christmas. So a more significant reason to outsource is to gain the flexibility in infrastructure that reflects the economic cycle.

And then, it turns out that the number of hits reaches a massive peak at this time. It is, therefore, even more important that the Web hosting service performs optimally under high volumes of traffic. In other words, from the beginning performance should have been the key issue in choosing an outsourcer since performance - not cost - is in fact going to have the biggest impact on profitability.

But if the case for performance monitoring is so strong, why is it so neglected? Part of the problem is that IT is commonly treated as an unallocated cost: decisions about the investment are ultimately made by the finance officer who may know little or nothing about IT.

This attitude is compounded in the outsourcing scenario when organisations are already, in part, outsourcing to reduce their direct responsibility for the service that is pushed out.

However, this hand-washing approach may be changing. Increasingly, users are forcing service providers into contractual obligations to demonstrate that the technology provided is delivering quantifiable benefits. This move passes responsibility to service providers who many commentators would say have colluded with poor performance discipline by concentrating on winning new contracts rather than maintaining those in existence already.

Jacques Hale, director of methods research at Butler Group, believes that the next 10 years will see a dramatic shift in power from the service suppliers to service users, since buyers will increasingly insist that suppliers back up promises with tangible results. This shift will be triggered by a growing board-level scrutiny of IT budgets and a requirement to demonstrate returns on investment.

"Sweeping promises were bandied around by IT suppliers throughout the 1990s about the massive savings that could be made through implementing new technology," Hale says. "But the recent downturn in the US has made IT users suspicious and now promises won't suffice. If tangible proof of return of investment is not actually demonstrated after purchase, the supplier could well find that initial promise coming under the scrutiny of a legal court."

In a less adversarial approach, Gilbert Toppin, chief operations officer Europe at Deloitte Consulting, explains that real breakthroughs will be made when the "economics of collaborative" are reflected in mutually owned performance measures. "The real measure of trust and confidence would be when each party gets paid on the performance of each other," he says.

Such delivery heaven seems some way off. "But the winning companies have already begun the process of assessing outsourcing arrangements and whilst it is not yet endemic we believe everyone will be doing it in the next 18 months or so." By then the business environment could have made it an imperative.

Read more on IT outsourcing