When Indiana Governor Mitch Daniels took office, he inherited a social services department that had distinguished itself only in bad ways. The state's Family and Social Services Administration had high error rates and low job-placement rates, and more than two dozen employees had been convicted of fraud. It was facing federal sanctions due to poor performance.

At the heart of this dysfunction was antiquated technology. The benefit eligibility system that operated as the front end of programs such as Temporary Assistance for Needy Families, Medicaid and food stamps was entirely paper-based. FSSA assisted clients only through face-to-face interactions, and this approach caused struggling moms and others looking for help to make 2.5 million unnecessary trips a year to FSSA offices.

Dissatisfied with the status quo, Daniels took a chance on a big change. In 2006, FSSA signed a 10-year, $1.3 billion contract with IBM in which nearly 1,500 state employees were shifted to the private sector.

From the beginning, the outsourcing faced problems. This October, Daniels cancelled the contract, describing it as an endeavor that "just did not work."

What can we learn from Indiana's experience?

First, we need to recognize that innovation and change are risky, and that outsourcing doesn't guarantee either success or failure. Indiana's FSSA, like many social service providers, used contractors to deliver over 90 percent of its services -- mental health organizations, doctors, welfare-to-work nonprofits and so forth -- even before this modernization.

In fact, the moral of this story has little to do with outsourcing. The outsourced employees, even when using the same antiquated system, actually performed better under private management.

Clearly then, the biggest risk factor in this case lay in the technology foundation. When upgrading technologically, the bigger the potential gain, the greater the potential pain. Risk cannot be eliminated, but it can be managed. Indiana had relatively clear service-level agreements when the contract was fully implemented statewide. However, during the transition the contract allowed for IBM to gradually improve without facing tough contract penalties for missing performance goals. As such, the IBM team added considerably to its labor pool to improve their performance, but those input changes did not produce the performance Daniels expected and citizens needed.

Second, the challenge of migrating from the old system to the new was underestimated, and there were two major design decisions that contributed to the poor results. One was the multiplication of channels. The other was the attempt to integrate, through a technology platform, all aspects of the application process.

A profusion of channels was supposed to be one of the benefits of the outsourcing. Under the outsourced system, instead of having to appear in person, clients would have a number of ways to apply: in person, over the phone, by fax or via the Internet. Unfortunately, the multiplicity of channels never meshed. Moreover, the paper-intensive legacy system meant that many applicants had to begin the process anew. "With the very best of intentions, namely convenience for the application, we produced a system that was at least as untimely as the other one," says Daniels.

Starting over was hard. The new online application ran to 16 pages, a daunting task. Moreover, for the many state employees absorbed by the contractors, there was a steep learning curve for the new processes, procedures and technologies being introduced. The contractor constantly had to educate the public and train its own workforce. In an effort to make lots of improvements, the outsourcing may have jumped the shark.

Which brings us to the second major design flaw: the integration platform. The old system was a paper-based nightmare, but it had one redeeming virtue: simplicity. A single, dedicated caseworker had a file containing all of an applicant's information. This approach was prone to fraud, had little quality control and was wildly inefficient -- but it did produce determinations. The new approach was supposed to be a seamless network of digitized, virtual files. Unfortunately, it never lived up to expectations.

Moving forward, the state will go with a hybrid model, attempting to capture some of the good things that came out of the outsourcing. For example, the new approach separated functions in way that reduced fraud. Under the old system, a single person took in your information, decided if you were eligible and then administered your benefits. Says Daniels, "It was an invitation to fraud, and people accepted the invitation."

The Indiana experience demonstrates one other common mistake. It is easy to underestimate the cost and effort of meeting federal reporting guidelines. What looks like a straightforward application of commercial best practices (call centers and digital applications) is often completely mired down in federal red tape. A company that hasn't operated in that specific arena in the past rarely understands the complexity and cost of federal compliance.

Should public officials in Indiana be criticized for trying to improve a system that was, by all accounts, dreadful? Hardly. As I wrote in an article when the contract was awarded: "Success will be far from automatic with such an enormous undertaking. The modernization of eligibility in Indiana places the state ahead of its peers. But the project, regardless of its ultimate success, can't help but lay down guidelines that will improve the chances of success in future public-private partnerships of such size and complexity."

The Indiana experience highlights some pitfalls that should be avoided. Nonetheless, it is quite possible that in the coming years the Hoosier State will find itself better positioned than it was before the outsourcing effort. Bold efforts to improve citizen service won't always work, but they should always be encouraged.