The requirement tightened exactly as the comparison came apart
Two things happened to federal training evaluation in 2026, and they pull in opposite directions.
The first tightened the requirement. On June 25, 2026, OPM published a final rule overhauling Senior Executive Service Candidate Development Programs, effective July 27, 2026. It rewrote 5 CFR 412.303 to require agencies to complete and maintain SESCDP program evaluations on OPM-developed templates â per cohort, annually for the overarching program, and expressly for program improvement. Re-approval now turns on an evaluation that reports initial SES placement rates for QRB-certified graduates and demonstrates a minimum placement rate set by OPM.
The second removed the reference point. In a notice published July 2, 2026, OPM stated that it âwill no longer centrally administer the survey on behalf of agencies. Instead, each agency will administer the survey to its employees,â and proposed cutting the prescribed annual employee survey questions from 16 to 10. Agencies still report prescribed-question data to OPM and OMB once a year âto enable governmentwide comparisonâ â but the instrument, the fielding window, and the local item set now vary agency by agency.
One point of status, because it changes what an agency is actually obligated to do. That reduction is still a proposal. The current text of 5 CFR 250.302 continues to prescribe all sixteen questions, and the section has not been amended since 2017. Agencies planning a 2026 survey are therefore designing against a proposed instrument while a sixteen-question regulation remains in force â a gap that has to be managed rather than assumed away.
That leaves a familiar problem in an unfamiliar form. You must still prove your training works, but you can no longer do it by pointing at where you sit relative to everyone else. You have to build the evidence yourself.
What the two sections actually require
Read them narrowly. Both are shorter than their reputations.
5 CFR 410.202 is a single sentence: âAgencies must evaluate their training programs annually to determine how well such plans and programs contribute to mission accomplishment and meet organizational performance goals.â
Three obligations sit inside it. The cadence is annual. The unit of analysis is plans and programs, not individual courses. And the standard is contribution to mission accomplishment and organizational performance goals â for satisfaction, not completion, not seat time.
It does not stand alone. Section 410.201(d)(4) requires agency heads to assess the overall talent management program âperiodically, but not less often than annually.â Section 412.201(c) requires management succession programs to â[i]nclude program evaluations pursuant to 5 CFR 410.202.â
5 CFR 412.303 is the specific case. Paragraph (a) requires an agency to âcomplete and maintain program evaluations pursuant to training evaluation requirements in 5 CFR 410.202â using OPM-developed templates, completed respectively by individual SESCDP participants and agency program managers: (1) upon completion of each individual cohort; (2) annually for the overarching SESCDP; and (3) to collect evaluation data for identifying and implementing program enhancements or alternative approaches to program administration.
Paragraph (b) carries the enforcement weight. Evaluations âmust include initial SES placement rates for graduates who receive a QRB certification and demonstrate that the agency maintains a minimum placement rate as specified by OPM policy and guidance.â Cohort templates are not required at re-approval, but OPM âreserves the right to request templates for each individual cohort during the current approval periodâ â a records-retention instruction in all but name.
Design against the rubric you will be graded on
The most efficient move available is also the most obvious: build your evaluation against the criteria a reviewer will apply to it. GAO published those criteria and has not withdrawn them. Human Capital: A Guide for Assessing Strategic Training and Development Efforts in the Federal Government (GAO-04-546G, March 2004, superseding the GAO-03-893G exposure draft) organizes the process into four components â planning/front-end analysis, design/development, implementation, and evaluation â and supplies key questions with âlook forâ elements under each. GAO states it has relied and will continue to rely on this framework to report on training within agencies and across government.
The evaluation component poses eight key questions. Use them as the section headings of your plan: (a) systematic planning for evaluation; (b) appropriate analytical approaches; (c) performance data, quantitative and qualitative; (d) feedback returned to planning, design, and implementation; (e) multiple stakeholder perspectives; (f) cost and delivery tracking; (g) benefits achieved; and (h) comparison with other organizations.
Question (h) is the one the survey change complicates â and GAOâs answer was never âcheck the governmentwide average.â GAO describes benchmarking against public and private organizations undertaking notably innovative and effective training efforts, done âwithin the context of that agencyâs unique environment and situation.â That option survives intact. It requires only that you name your comparison set rather than assume one.
On question (b), GAO describes a five-level approach: participant reaction, learning, on-the-job behavior change, organizational results, and return on investment. Not all programs require or suit the higher levels, and reviewers look for criteria determining which programs get evaluated at which level. A defensible design states that gradation in advance. An indefensible one runs reaction surveys everywhere and calls the result evaluation.
Designing for defensibility: what a reviewer actually opens
What survives review is not methodological sophistication. It is traceability. Five artifacts carry most of the weight.
A written evaluation plan. GAOâs âlook forâ list under question (a) calls for a data collection and analysis plan setting priorities, methods, timing, and responsibilities. Name the responsible official. Date the document. Keep it version-controlled.
Decision rules for evaluation level, set before the data arrives. Which programs get behavior- or results-level treatment, and on what basis â cost, audience size, management interest, visibility, anticipated life span.
A stated link from program to mission measure. Name the organizational measure and the logic connecting training to it. Where outcomes take years to surface, GAO points to logic models and intermediate measures as the accepted bridge.
Cost and delivery data that reconcile. Question (f) expects learning management system delivery data alongside accounting and financial systems producing consistent numbers. Reviewers test whether your training cost figure ties to the financial system. Many do not.
Evidence the findings changed something. Question (d) looks for fact-based determinations used to refine or implement programs, and for resources reallocated on the strength of evaluation data. An evaluation that recommends nothing reads as a formality, because it is one.
A minimum-viable evaluation design you can stand up this fiscal year
You do not need a research shop. You need four moves and one signature block.
Move 1 â Define the annual unit and write the plan (Q1). Decide what âprogramâ means at your agency, list them, and publish a one-page evaluation plan for each. This alone closes more of the 410.202 gap than most agencies have closed.
Move 2 â Set the level gradation (Q1). Assign every program an evaluation level with a written rationale. Reaction data across the board. Learning measures wherever competency targets exist. Behavior measures for your highest-cost and highest-visibility programs. Reserve ROI for a small, named subset; GAO anticipates that agencies evaluate only a minority of programs at that depth.
Move 3 â Build one behavior-level instrument and field it twice (Q2âQ3). A paired participant-and-supervisor instrument, fielded roughly 90 days after training on your highest-stakes program, answers questions (c) and (e) at once. Two well-run instruments beat twenty smile sheets.
Move 4 â Close the loop in writing (Q4). Produce a short annual memo per program: what was measured, what it showed, what changed, what gets measured next cycle. Route it for signature. This answers question (d), and it is the artifact agencies most often cannot produce on request.
For SESCDP, layer the OPM templates on that spine: participant and program-manager templates at each cohort close, the overarching template annually, and continuous tracking of initial SES placement rates for QRB-certified graduates â maintained as you go, not reconstructed before re-approval.
Where agencies typically get this wrong
Substituting completion data for evaluation. Completions, training hours, and seats filled are delivery metrics. Section 410.202 asks about contribution to mission accomplishment. Delivery data belongs in the file; it does not answer the question the regulation poses.
Evaluating courses instead of programs. The regulation addresses plans and programs. Four hundred course-level reaction surveys with no program-level assessment is volume without responsiveness.
Treating an annual requirement as an annual event. GAO is explicit that agencies should view training evaluation ânot as a static, after-the-fact requirement but as a continual, ongoing effort.â A fourth-quarter scramble produces documents that look exactly like what they are.
Waiting on the OPM template before designing anything. Where 412.303 requires an OPM-developed template, use it â but a template is a reporting format, not an evaluation design. The data collection must be running long before the template comes due.
Abandoning benchmarking because the governmentwide instrument moved. Cross-agency comparison is harder and more manual now, not unavailable. State your comparison basis instead of assuming your reader shares it.
The through-line
A centrally administered governmentwide survey let agencies answer âhow are we doing?â without ever defining âdoing.â That shortcut has closed. What replaces it is unglamorous and entirely achievable: a written plan, a stated standard, data that reconciles, and a signed record of what you changed because of what you found. That is also, not coincidentally, what survives the review.
Sources
5 CFR 440.201, Responsibilities of the head of an agency [74 FR 65387, Dec. 10, 2009]. https://www.ecfr.gov/current/title-5/chapter-I/subchapter-B/part-410/subpart-B
5 CFR 410.202, Responsibilities for evaluating training [74 FR 65387, Dec. 10, 2009]. https://www.ecfr.gov/current/title-5/chapter-I/subchapter-B/part-410/subpart-B/section-410.202
5 CFR 412.201, Management succession. https://www.ecfr.gov/current/title-5/chapter-I/subchapter-B/part-412/subpart-B/section-412.201
5 CFR 412.303, Senior Executive Service candidate development program (SESCDP) oversight and evaluation [91 FR 38246, June 25, 2026]. https://www.ecfr.gov/current/title-5/chapter-I/subchapter-B/part-412/subpart-C
Office of Personnel Management, final rule on Senior Executive Service Candidate Development Programs, 91 FR 38235 (June 25, 2026), Vol. 91, No. 121, FR Doc. 2026-12811; effective July 27, 2026. Amendments to 5 CFR part 412 appear at 91 FR 38244z
Ôâ38246.
U.S. Government Accountability Office, Human Capital: A Guide for Assessing Strategic Training and Development Efforts in the Federal Government, GAO-04-546G (Washington, D.C.: March 1, 2004); supersedes GAO-03-893G. https://www.gao.gov/products/gao-04-546g
Office of Personnel Management, âAgency Information Collection Request: Federal Employee Viewpoint Survey, OMB Control Number (3206-NEW),â 91 FR 40599 (July 2, 2026), FR Doc. 2026-13443. https://www.federalregister.gov/d/2026-13443
5 CFR part 250, subpart C, Employee Surveys (§§ 250.301â250.303) [81 FR 89367, Dec. 12, 2016]. https://www.ecfr.gov/current/title-5/chapter-I/subchapter-B/part-250/subpart-C