Scott Kupor’s Test at OPM: Turning Reform into Better Service

Give improvement a place in the incentive system

Kupor describes arriving in a culture where avoiding adverse attention could become an overriding objective. When a failed experiment brings scrutiny but a persistently poor service brings little consequence, maintaining the familiar process can become a rational career choice.

His proposed response is permissioned risk: leaders should authorize appropriately bounded attempts to improve a service and accept responsibility for reasonable decisions that do not work as intended. He distinguishes that approach from the much larger risks a venture investor can take in pursuit of exceptional returns.

That distinction matters. A delayed benefit payment can harm someone even when a system is not usually described as life-critical. A public manager needs to specify who could be affected, how a pilot will be limited, what evidence permits expansion, and how service will continue if the change fails. Oversight can help make those decisions defensible.

Permission to improve a service becomes useful when people know their authority, their boundaries, and the evidence by which the work will be judged.

This is a leadership obligation. Staff need a practical way to raise problems, obtain decisions, and test improvements without carrying all the institutional risk themselves. They also need leaders who will respond to evidence that an initiative should change course.

Make merit visible in hiring and advancement

Kupor sees a connected set of barriers: hiring processes that mistake credentials for ability, promotion rules that reward elapsed time, and ratings that make it difficult to distinguish exceptional contributions. His preferred direction is demonstrated skill, appropriate entry levels, and rewards that recognize performance.

Several parts of that agenda have moved into formal policy. OPM’s February 2026 assessment guidance directs agencies using Direct Hire Authority toward job-related evaluation, including work samples, technical tests, and structured interviews. The guidance builds on earlier reforms; assessing actual skills is an existing federal practice that OPM is seeking to strengthen and apply more consistently.

The September 21 performance memorandum sets a combined 40 percent ceiling on the two highest ratings for covered career employee populations, subject to its exclusions and waiver provisions. It also says at least 60 percent of each covered population’s bonus pool should go to those performers. These are rules about ratings and awards, not evidence by themselves that output has improved.

The final time-in-grade rule removes the governmentwide waiting-period restriction for covered promotions, with an effective date of January 19, 2027. Occupational qualifications and other applicable requirements remain. A policy change on the books therefore should not be confused with a promotion flexibility already available everywhere.

The managerial challenge is to make these changes credible. Define successful work before assessing it; use evaluators who understand the job; document decisions; and recognize team contributions alongside individual accomplishments. A numerical ratings limit cannot substitute for that judgment. Employees should be able to see how an assessment relates to the work they actually performed.

Offer a compelling way into public service

Kupor’s recruiting pitch is deliberately concrete: come work on difficult national problems, develop valuable skills, and leave with more options. A candidate need not make a lifetime commitment to contribute meaningfully.

He cites the small share of federal employees under 30—about 7 percent in his account—as a reason to rebuild the entry pipeline. The relevant management problem is succession and skills continuity. Age alone establishes neither capability nor a retirement date; attracting new entrants works best alongside retaining expertise and transferring knowledge.

Tech Force embodies the shorter-service model: approximately 1,000 technology specialists are envisaged for two-year appointments, with training and industry participation. Its recruiting materials describe an intended corps, not proof that all those positions have been filled. EarlyCareers.gov offers a broader entry point across functions such as finance, procurement, HR, and technology.

For agencies, the assignment matters as much as the recruitment campaign. A short appointment needs a defined problem, a manager who can clear obstacles, access to the necessary systems, and a team that can sustain the result. Pairing new technical talent with experienced career staff can shorten the learning curve in both directions.

Long public-service careers remain valuable, too. The opportunity is to create more credible ways to enter, contribute, develop, and stay—or leave responsibly. A two-year program succeeds institutionally when the capability it builds remains after participants move on.

Retirement modernization makes the stakes concrete

The interview’s most vivid example is the retirement operation in Boyers, Pennsylvania. Kupor describes paper applications moving among agency HR offices, payroll providers, and OPM, followed by manual handling and calculations. Historical files stored underground still complicate service when an annuitant needs help.

The problem reaches beyond paper. An employee’s service history can be fragmented across organizations and systems. Reconstructing that history at retirement makes the retiree absorb the consequences of information that was never properly connected during employment.

There is substantive progress to report. OPM’s “Last Day of Paper” announcement says it ended paper processing for more than 95 percent of retirement applications, with more than 155,000 applications having used the online system. That is a meaningful change in the delivery process. Digitizing hundreds of millions of historical records remains an ongoing effort.

Kupor reports processing-time reductions of more than half and describes a goal of full retirement payments within 15 days for complete, healthy electronic cases. Those claims need their scope kept intact. OPM’s published September 2026 averages were 66 days for immediate retirement processing and six days for interim pay. Those measures cover different stages and case populations; they do not establish a universal 15-day result.

Earlier reformers also tried to solve the problem. GAO documented repeated modernization efforts, including a program terminated in 2011, and weaknesses in testing, cost estimates, and implementation planning. That history makes the current progress consequential while underscoring the delivery discipline required to sustain it.

For the retiree, the meaningful outcome is a correct payment when expected, understandable status information, and help when a case is complicated. Much of the improvement described here comes from digital workflow, usable records, and better service design. Those foundations also matter when agencies introduce AI; adopting AI does not remove the need to get them right.

Turn shared HR infrastructure into an operating capability

Kupor links the retirement problem to a larger ambition: consolidating fragmented HR technology so a reliable employee record can follow a person through federal service. In the interview, he describes roughly 120 systems and projects a reduction in annual spending from about $2 billion to $40 million.

Those figures are his estimate of the opportunity. OPM’s June 10 announcement confirms a contract award for a governmentwide HR platform and describes a phased transition, with projected spending reductions exceeding 90 percent. Neither the award nor the forecast establishes that agencies have completed migration or realized the savings.

The case for common infrastructure is strong where agencies repeatedly buy and maintain similar capabilities. But shared service is an operating arrangement as well as a technology decision. It needs agreed data definitions, accountable service owners, reliable interfaces, and a way to resolve agency requirements without recreating the fragmentation.

Measure the economics across the transition. Migration, data cleanup, integration, training, and temporary operation of old and new systems all consume resources. Savings become durable when agencies actually retire redundant systems and costs while maintaining or improving service.

The same reasoning applies to states and localities considering shared HR or benefits infrastructure: use scale where needs are common, retain necessary mission expertise, and make the service provider answerable for the experience of participating organizations and their users.

The next phase must make reform last

Kupor credits DOGE with disrupting established patterns and describes his own task as institutionalizing the changes. He cites an approximately 300,000-person federal workforce reduction in the interview. That is his rounded account; headcount alone cannot show whether government has become more capable or whether particular services have improved.

His next question is more operationally useful: what work should agencies stop doing so they can concentrate on what matters? Reducing staff while leaving every obligation and process intact risks moving the burden to remaining employees and the public. Prioritization requires decisions about work, service commitments, and the resources needed to meet them.

For this agenda, a practical public scorecard would connect management changes to observable results:

Kupor’s most useful contribution is the connection he draws among people, incentives, and delivery. Each part depends on the others. A strong recruiting message must lead to a workplace where people can contribute; performance management must recognize that contribution; and technology must make better service possible.

The opportunity at OPM is to make those connections routine enough to endure. That would give future leaders something more valuable than a compelling reform narrative: an institution better equipped to serve.