Nothing here is about arguing your way to a better rating. The routes for that narrowed sharply this year. Everything below is about building a record during the cycle that makes the accurate rating the easy one for your supervisor to give.
I What changed this year
OPM's final rule on performance appraisal for General Schedule, prevailing rate and certain other employees took effect on 6 August 2026. Four changes matter to you directly.
| Change | What it means |
|---|---|
| Grievance route removed | You can no longer grieve a rating of record |
| Forced distribution allowed | The prohibition on a standardized distribution of rating levels was removed |
| Level 1 review ended | Mandatory higher-level review of an Unacceptable rating no longer required |
| Supervisory element required | A supervisory critical element for all covered supervisors |
The rule also eliminated unnecessary summary level patterns and requires OPM to conduct biennial certifications of agency appraisal systems. Compliance with the summary level pattern provisions at 5 CFR 430.208(e)(1) and (2) is required from 1 January 2027, so parts of this will reach your agency during the coming cycle rather than immediately.
Where an agency applies a standardized distribution, your rating stops being purely a function of your work and becomes partly a function of how your work ranks against your colleagues'. That does not change what you should do — a specific, evidenced record is precisely what survives a comparative process. But it widens the gap between a documented accomplishment and an undocumented one, because in a comparison the undocumented one simply is not there.
II Why the stakes moved
A month after the appraisal rule, the reduction in force rule took effect on 2 September 2026 and made performance the opening score rather than the final tie-breaker.
Your performance credit is the sum of your three most recent ratings of record received in the four-year period before RIF notices issue: 7 points for a Level 5, 5 for a Level 4, 3 for a Level 3, and zero for anything below. Maximum 21. Veterans' preference is then added as points, and tenure and length of service are tie-breakers reached only after that.
Three Level 4 ratings score 15. Three Level 3 ratings score 9. That six-point gap is larger than the five points a preference eligible with a 30 percent disability receives, and vastly larger than anything length of service now contributes. Two colleagues doing similar work, one consistently rated Exceeds and one consistently rated Fully Successful, are no longer close. Full mechanics in performance ratings and their consequences.
III Start with the performance plan
The highest-return habit in federal performance management costs an hour and almost nobody does it: read the performance plan when it is issued.
Take each critical element and establish what it actually requires. If an element is vague — and many are — ask for clarification in writing, at the start of the cycle, when asking costs nothing and reads as diligence. The same question in the final month reads as an excuse.
Three questions per element are usually enough:
- What specifically would Fully Successful look like here?
- What would exceeding it look like?
- How will it be measured, and from what source?
The answers give you a target and a written record of what the target was — which matters later precisely because you can no longer grieve the outcome.
IV The contemporaneous record
Keep accomplishments as they happen: the date, what you did, and what changed because of it. One place, updated monthly, written so someone else could read it.
Why timing decides the quality
Two employees write self-assessments in the same week of the same cycle.
The other reconstructs from memory and writes: “Consistently demonstrated strong commitment to improving office efficiency and reducing backlogs.”
Same work, possibly. But the first is evidence and the second is an adjective, and any rater can tell which is which in a second. Under forced distribution, so can anyone comparing the two.
The record is also what you attach if the rating comes back below what it supports — see documenting everything for the broader practice.
V Self-Assessment Readiness Check
How much of the cycle you have left, what your current ratings are already worth, and what one level would change.
Where Does This Cycle Stand?
A planning aid, not a prediction. Your agency's appraisal system governs how levels are assigned.
VI Writing the self-assessment
Map it to the elements, one at a time, in the order the plan lists them. Under each: the accomplishment, the scope, the outcome.
| Do | Do not |
|---|---|
| Give numbers where numbers exist | Reach for adjectives where they do not |
| Say what changed because of you | List what you were responsible for |
| Follow the plan's element order | Write a narrative of your year |
| Name the scope — people, money, reach | Assume the rater remembers it |
| Attribute results to your decisions | Claim credit for the team's baseline work |
The most common failure is describing responsibilities rather than results. Your position description already lists your duties; repeating them tells a rater nothing. What earns a higher level is evidence of outcomes attributable to you — the same standard that governs ECQ narratives, for the same reason.
Ask for mid-year feedback in writing. A surprise at rating time is a documentation failure, and written mid-year feedback is both a course correction and a record of what was said at the halfway point.
VII When the rating is wrong
The ordinary grievance route for a rating of record was removed on 6 August 2026. What remains is narrower and mostly earlier.
| Route | Status |
|---|---|
| Grieving the rating of record | Removed |
| Raising it with the rater before it is final | Available, and now the main move |
| Within-grade increase denial | Own reconsideration process |
| Discrimination or retaliation | EEO or OSC, own deadlines |
So the practical answer is: raise it immediately, in writing, with the record attached, before the rating is finalised. A discussion supported by contemporaneous evidence is a different conversation from a complaint, and supervisors do amend ratings that rest on something factually wrong.
If discrimination or retaliation is genuinely involved, those claims run on separate tracks with their own deadlines — establish those deadlines before you raise anything, not after.
VIII The playbook
| Situation | What to do |
|---|---|
| Plan just issued | Read every element. Ask in writing what Fully Successful and Exceeds look like. |
| Element is vague | Clarify now. The same question in month eleven reads as an excuse. |
| Mid-cycle | Request feedback in writing and keep it. |
| Nothing written down yet | Start today. Reconstruction produces adjectives, not evidence. |
| Writing the self-assessment | Map to elements in order. Results, not responsibilities. |
| Agency uses forced distribution | Documentation matters more, not less — you are being compared. |
| Rating lower than the record supports | Raise it in writing before it is final. The grievance route is gone. |
| Rating rests on a factual error | Say precisely what, and attach the proof. |
| Believe it is discrimination or retaliation | Separate track, separate deadlines. Establish them first. |
| Three Fully Successful years | That is 9 of 21 in a RIF. Know where you stand. |
This cycle
- Pull this cycle's performance plan and read every critical element.
- Ask in writing what Fully Successful and Exceeds look like for each one.
- Start a dated accomplishment record today, wherever you are in the cycle.
- Add to it monthly rather than reconstructing later.
- Record scope and outcome, not just activity.
- Request mid-year feedback in writing.
- Find out whether your agency applies a standardized distribution.
- Score your last three ratings against the RIF credit table.
- Write the self-assessment mapped to the elements, in their order.
- If the rating is wrong, raise it in writing before it is finalised.
Frequently asked questions
Not through the route that used to exist. The OPM final rule on performance appraisal for General Schedule, prevailing rate and certain other employees, effective 6 August 2026, removed the option to grieve a rating of record. That does not extinguish every avenue — a discrimination or retaliation claim runs on its own track through EEO or the Office of Special Counsel, and a within-grade increase denial carries its own reconsideration process — but the ordinary path of grieving the rating itself is gone. The practical consequence is that everything now depends on the record you build before the rating is issued.
The same rule removed the prohibition on a forced, or standardized, distribution of performance rating levels. Where an agency chooses to apply one, your rating is no longer a function of your work alone but also of how your work is ranked against colleagues. That does not change what you should do — a specific, evidenced record is exactly what survives a comparative process — but it does change your expectations, and it makes the difference between a documented accomplishment and an undocumented one much larger.
Because it is now the first thing counted in a reduction in force. Under the RIF rule effective 2 September 2026, performance credit is computed from your three most recent ratings of record before tenure or veterans' preference is considered, at 7 points for a Level 5, 5 for a Level 4, 3 for a Level 3 and zero below that. Length of service became a tie-breaker reached only after that score. A run of average ratings now costs standing that decades of service used to protect.
The day the performance plan is issued, not the month the rating is due. Read each critical element and establish what it actually requires, in writing, while there is still a full cycle in which to deliver it. Vague elements are worth raising at the start when clarification costs nothing; raising them at the end looks like an excuse. This is the single highest-return habit in federal performance management and almost nobody does it.
Accomplishments as they happen, with dates, what you actually did, and what changed because of it. Keep it in one place, add to it monthly, and write it in a form you could hand to someone else. The reason is simple: a self-assessment written from a contemporaneous record contains specifics, and a self-assessment reconstructed in the final week contains adjectives. Supervisors can tell the difference immediately, and so can anyone reviewing the rating afterwards.
Map it to the elements, one at a time, in the order they appear in the plan. Under each element, give the accomplishment, the scope, and the outcome. Use numbers where numbers exist and say plainly what changed where they do not. Avoid describing responsibilities — a list of duties is what your position description already says, and repeating it tells the rater nothing they did not know. What earns a higher level is evidence of results attributable to you.
Raise it immediately and in writing, with the record attached. Even without a grievance route, a rating discussion supported by contemporaneous evidence is a different conversation from a complaint, and supervisors can and do amend ratings before they are finalised. If the rating rests on something factually wrong, say precisely what and attach the proof. If discrimination or retaliation is genuinely involved, those claims run separately and have their own deadlines, which you should establish before rather than after you raise anything.
Several things worth knowing. It eliminated unnecessary summary level patterns, ended the mandatory higher-level review of Level 1 ratings, and requires a supervisory critical element for all supervisors covered by the subpart. It also requires OPM to conduct biennial certifications of agency appraisal systems. The rule took effect on 6 August 2026, with compliance for the summary level pattern provisions in 5 CFR 430.208(e)(1) and (2) required from 1 January 2027, so some of it will land in your agency during the coming cycle rather than immediately.