John O'Leary

is a research fellow at the Ash Institute of the Harvard Kennedy School and executive editor of the Ash Institute's Better, Faster, Cheaper Web site. He is also a co-author of If We Can Put a Man on the Moon: Getting Big Things Done in Government, published by the Harvard Business School Press.

About This Column:
Governing and the Harvard Kennedy School's Ash Institute have teamed up to create a marketplace for ideas that will spread the adoption of creative public-sector solutions. The Better, Faster, Cheaper site and efficiency newsletter aspire to be the first place public officials look for ways to improve services and lower costs. Sign up now to receive ideas for better, faster, cheaper government in your inbox.

When you buy a ticket from JetBlue, it costs you an extra $15 to book a flight over the phone compared to doing it yourself over the Internet. The reason is obvious. Online transactions are much less costly for JetBlue, and the fee not only generates revenue but also helps the airline drive consumers to the lower-cost transaction. If you want to talk to someone on the phone, that's fine — but it's going to cost you. If you don't like it, you can always call another airline.

The fee helps JetBlue keep its prices low, making it one of the most popular and profitable airlines around.

But what about government services, particularly those for which charging a fee doesn't make sense? How do you drive consumers to the low-cost transaction channel without being able to charge a fee?

Unemployment insurance is a classic government monopoly. There are no competitors offering the same service, and an unemployment insurance recipient isn't — and shouldn't be — charged a fee for applying for benefits. So what is a manager looking to provide better, faster, cheaper services supposed to do?

How about trying better, slower, cheaper?

When I took over as the commissioner for the Division of Unemployment Assistance in Massachusetts in 2004, the organization was focused on answering the phones quickly. Call the DUA, and your call would be answered in an average of just about three and a half minutes. The problem was, this obsession with answering the phones quickly made all the back-office operations inefficient, since it meant pulling higher-skilled workers off their tasks to take claims calls. This meant that our checks went out very slowly, which resulted in a lot of "where's my check?" calls, calls that add no value and, in a more efficient operation, wouldn't need to occur at all.

Our disputed claims backlog was also huge. The federal standard said that 80 percent of disputed claims should get a hearing in 45 days or less. For us, less than 30 percent were being heard in that time period. On my first week on the job, a very nice lawyer from Greater Boston Legal Services shook my hand, welcomed me to my new position and told me if the backlog didn't come down they'd take the agency to court.

I decided that service had to be looked at as an overall package, and that the best thing I could do for customers was to extend phone wait times. We weren't going to lose any customers over it, and unemployment insurance claimants were, by definition, people who had time on their hands.

The results came swiftly. Without the organizational chaos caused by the constant pursuit of low wait times, our backlogs came down. Within 18 months, 83 percent of our hearings were being held within 45 days, exceeding the federal standard. Checks were going out much faster too, meaning claimants got their first unemployment insurance check days earlier. This resulted in about a 20 percent drop in "where's my check?" calls — which translated into 18,000 fewer calls each month.

Yes, customers now had to wait eight minutes instead of three and half, but there can be no doubt that their overall experience was improved. It's counter-intuitive, but just as JetBlue customers have an overall better experience with a $15 phone booking fee, DUA filers were better off waiting on hold for a few minutes.

The next area we tackled was also channel-related. In the early 1990s, the DUA offered just one way to "certify" your claim each week, which basically meant telling the DUA that you were still out of work and still looking. That one way was to go in person to a DUA office. This was inconvenient and costly for everyone.

By 2004, however, the problem was too many options. You could certify in person, over the phone, over the Web or by sending in a little postcard by U.S. mail.

I hated those little postcards. Twenty five percent of our certifications came in via postcards, burying us in roughly 25,000 pieces of mail. These postcards had to be reviewed and the data entered into our system by hand. It was labor-intensive, errors were common and it was a far more costly transaction than the automated "telecert" phone system or the Web.

We first tried to change behavior through training. Phone representatives were taught to encourage people to sign up for phone certification when they applied. This brought postcard usage down from 25 percent to 15 percent, but that was still too high for my liking.

When we investigated further, we found that postcards were overwhelmingly used by our Spanish-language speakers who couldn't follow the telephone prompts. So we introduced a Spanish-language version of the telecert system — and the mail all but disappeared, with 96 percent of our certifications taking place through automated processes.

Even in the absence of price mechanisms and competitive comparisons, it still makes sense to examine the channels through which you deliver services. Sometimes adding a channel (such as the Spanish telecert) will allow you to do away with one entirely. Sometimes diminished service levels in one area (phone wait times) can allow you to deliver a better overall bundle of services.