年终 review 季自查包：证据、影响、失分点，一次对齐

Year-end review season is here, and most self-reviews fail in the same quiet way: they list work, but they do not make the impact easy to believe. Calibration does not punish effort. It dilutes claims it cannot verify. So before you write a single polished paragraph, go collect proof.

## 1. Gather evidence first, then write

Do not start from memory. Memory favors recent work and big meetings, and it drops the unglamorous fixes that actually kept a team moving. Pull a simple list from five places:

- Calendar: launches, reviews, incidents, cross-team decisions
- Tickets and PRs: what shipped, what you owned, what you unblocked
- Docs and design notes: where you shaped the direction, not just commented
- Emails and chat threads: praise, decisions, and follow-ups you drove
- Metrics dashboards: only numbers you can explain and defend

This matches the University of Colorado HR guidance on writing a self-evaluation: gather your accomplishments and evidence before drafting, and be specific rather than relying on general claims. Source: https://www.colorado.edu/hr/2020/01/22/performance-reviews-tips-writing-self-evaluation

A useful rule: if you cannot point to an artifact, a date, or a witness, treat that item as a draft claim, not a finished achievement. Keep it in a separate list until you find the proof.

## 2. One sentence per achievement: evidence -> what you did -> what changed

For each key item, use this formula:

Evidence -> action -> change.

For example, as a template (fill in your own numbers, do not copy these placeholders as facts):

- "[Ticket / launch link]: I redesigned [process / service], which cut [metric] from [X] to [Y] / saved the team [N] hours per week."
- "[Doc / review thread]: I proposed [decision], which resolved [blocking disagreement] and let [project] ship on [date]."

The rewrite test is the point. A reviewer who has read 800+ self-assessments puts it bluntly: task lists describe activity, impact statements describe change. Source: https://www.linkedin.com/posts/simoschiassi_i-reviewed-800-self-assessments-in-15-years-activity-7519397906743791616-zOdm

Weak, task-only version:
"Worked on Project A and attended weekly syncs."

Stronger, impact version (template):
"[Evidence: launch notes / dashboard]: I owned [part of Project A], shipped [outcome] on [date], and [team / users] could then [concrete change]."

Notice the difference is not adjectives. It is a verifiable result, an owner, and a consequence. If your sentence still works after deleting the project name, it is probably too generic.

## 3. Use a four-part structure

. 1point3acresA clean self-review does not need to be long. It needs to be easy to skim in calibration:

1. Overview of results: 2-3 sentences on the overall outcome for the year, tied to team or business goals.
2. 2-3 key achievements: one paragraph each, each using the evidence -> action -> change formula above.
3. One growth area + a concrete improvement plan: name the gap plainly, then state what you changed, what you will practice next quarter, and how you will measure it. Vague growth areas ("improve communication") help no one. Specific ones do ("I will write a one-page decision summary before cross-team reviews, starting with [project], and ask my manager for feedback after two reviews").
4. Next-year goals: 2-3 goals, each with a measurable signal of success, not just a topic.

This four-part shape — overview, key accomplishments, growth plus plan, and forward goals — is the structure recommended in year-end review guides. Sources: https://www.sybill.ai/blogs/how-to-write-year-end-reviews and https://instaprompt.app/blogs/how-to-write-a-self-evaluation-for-a-performance-review.html

## 4. Common ways to lose points

- Listing every task instead of choosing the 2-3 items that moved the needle.
- Claiming team results as solo results, or hiding your role so reviewers cannot tell what you did.
- Using superlatives with no numbers, dates, links, or outcomes.
- Writing only about the last two months.
- Saving the growth section for a throwaway line, which reads as if you did not reflect at all.
- Copying last year's review and changing the dates.
- Burying the strongest result in the middle of a long paragraph.

Short version: if a busy reviewer cannot find the impact in 30 seconds, the paragraph is not finished yet.

## 5. Pre-submission checklist

Before you hit submit, check each item:

- [ ] Every key claim has evidence: a link, a date, a metric, or a named reviewer.
- [ ] Each achievement sentence names what changed, not just what you did.
- [ ] The growth area has a dated, testable next step.
- [ ] Next-year goals have a success signal, not just a theme.
- [ ] You removed claims you cannot defend in a calibration discussion.
- [ ] Someone who was not on your team could follow the story without asking for context.

One last stance, because it matters: do not inflate to survive calibration. A smaller set of provable wins travels better than a long list of impressive-sounding tasks. Reviewers remember the one result they could repeat back in the room. Make that result easy to find.

Sources:
- University of Colorado HR, performance review tips for writing a self-evaluation: https://www.colorado.edu/hr/2020/01/22/performance-reviews-tips-writing-self-evaluation
- Google reviewer perspective on 800+ self-assessments: https://www.linkedin.com/posts/simoschiassi_i-reviewed-800-self-assessments-in-15-years-activity-7519397906743791616-zOdm
- Year-end review structure: https://www.sybill.ai/blogs/how-to-write-year-end-reviews
- Self-evaluation structure: https://instaprompt.app/blogs/how-to-write-a-self-evaluation-for-a-performance-review.html

Note: all examples above are templates and public guidance, with blanks left for your own numbers. No personal experience is claimed here — plug in your own evidence before you submit your review.