An annual performance review template needs four things a quarterly one doesn’t demand: full 12-month scope, evidence pulled from more than one part of the year (not just the last quarter), ratings that hold up in calibration, and a whole-year recency guardrail so a strong, quiet spring isn’t erased by a loud autumn. The copyable annual template is below, structured half-by-half so each rating has a place to cite two dated moments from different points in the year. If you want the generic four-part template and the full bank of ~30 vague-vs-specific competency examples, those live on our performance review template + examples page. This page is the annual, calibration-ready sibling and links to it rather than repeating it.
The reason the annual cycle is harder than a quarterly one is simple: memory is a terrible filing system, and over twelve months it fails badly. The quiet save your strongest engineer made on a Thursday in March is gone by December, while the one rough sprint in November is still vivid because it just happened. An annual template that doesn’t force dated evidence across the year quietly becomes a review of the last quarter with a year’s worth of stakes attached.
The annual review template
Copy this and fill it in per person. It’s laid out so every competency has one evidence slot for the first half of the year and one for the second. That split is the structure that keeps an annual rating from collapsing into a recency rating. Keep one of these per direct report.
ANNUAL REVIEW · @[Name] · [Cycle, e.g., Jan–Dec 2026] Scale: 1 Below · 2 Developing · 3 Meets · 4 Exceeds · 5 Outstanding COMPETENCY RATING EVIDENCE: 1 dated moment per HALF of the year Impact / Results [ ] H1 (Jan–Jun): ______ H2 (Jul–Dec): ______ Technical / Craft [ ] H1: ______ H2: ______ Communication [ ] H1: ______ H2: ______ Collaboration [ ] H1: ______ H2: ______ Ownership [ ] H1: ______ H2: ______ Growth (trajectory over the year) [ ] Early: ______ Late: ______ TOP ACCOMPLISHMENTS (3–5, dated, with impact) - DEVELOPMENT AREAS (behavior + path, not character) - GOALS · NEXT CYCLE (specific, measurable) - CALIBRATION NOTES (2 dated moments per rating, spanning ≥2 quarters) - OVERALL SUMMARY -
Each row earns its rating in the evidence columns, and the columns are split on purpose. If both your dated moments for a competency come from the same quarter, that is a signal you are rating the last few weeks, not the year. Need the competency examples to fill those slots? The generic template page has the full vague-vs-specific bank by competency. Use those patterns and swap in your person’s real projects and dates.
How to use it across the year
- Read all twelve months first. Walk the cycle from January forward, not from last month. Early work counts as much as last week.
- Rate each competency on the scale. Score impact, craft, communication, collaboration, ownership, and growth, one row at a time.
- Attach dated evidence from at least two quarters. One moment from the first half of the year, one from the second, for every rating.
- Write development areas as behavior plus a path. Describe what happened across the year and what changes next, not who the person “is.”
- Prep calibration notes and next-cycle goals. For each rating, note the two dated moments you would cite in the room, then set measurable goals for the next cycle.
See how managers walk into review season with the whole cycle already in hand.
Make your ratings hold up in calibration
Calibration is where an annual review gets tested. You sit with peer managers, put your ratings next to theirs, and defend why yours land where they do. An “Exceeds” you can’t source gets talked down to a “Meets” in ten minutes, not because the work wasn’t there, but because you can’t point to it. The managers who hold their ratings aren’t better arguers. They walk in with a dated line under every score.
Before calibration, run each report through one pass: for every competency, can you name two dated moments from different quarters of the year? If your evidence for “Impact” is all from October, that’s a recency rating dressed up as an annual one, and calibration will catch it. Pull one strong Q1–Q2 example against one from Q3–Q4. “Led the billing migration in March (zero downtime) and the search-relevance launch in September (+11% click-through)” survives the room. “Consistently high impact” does not. The evidence is what turns your rating from an opinion into a record, and a record is what calibration is built to reward.
The “will it survive calibration?” checklist
Run this on each report before you finalize the annual ratings. It is a whole-year audit: the same discipline the recency-bias deep dive applies to a single cycle, tightened for the calibration room.
CALIBRATION-READY CHECKLIST (per report, annual cycle) - Read the WHOLE 12-month cycle before writing a single rating - Every rating cites 2 dated moments from >=2 different quarters - No competency is scored entirely on the last quarter's evidence - One strong early example AND one strong late example per rating - Separate the person from one bad sprint (or one lucky launch) - Development areas are behavior + date, not character - Note trajectory over the year, not just a year-end snapshot - The overall summary matches the dated evidence, not the mood
Whole year, not last quarter
The single biggest failure mode of an annual review is that it silently becomes a review of the last quarter. Over twelve months the recent sprint feels like the whole year: a quiet-but-productive spring disappears, and one rough autumn drags down a strong cycle. Below is the same employee, the same year, rated two ways: once from memory in December, once from a year of dated evidence. Nothing about the person changed. Only the record did.
| From memory (December) | From a year of dated evidence |
|---|---|
| “Had a rough end to the year.” | “Led the billing migration in March (zero downtime) and the search-relevance launch in September (+11% click-through). One slip on the October export feature, flagged two weeks early with a revised plan.” |
| Overall rating: Meets. | Overall rating: Exceeds, with one dated growth area. |
The December version isn’t unfair on purpose. It’s unfair because the rough patch was recent and loud, and the two strong launches were spread across months that memory had already sanded smooth. The dated version isn’t kinder. It’s just more accurate, and accuracy is what a review owes the person being rated. For why this happens and how to design against it, the recency-bias guide is the deep dive.
The annual self-review (when you kept no notes)
If you’re the employee writing your own annual self-review and you didn’t log a thing all year, you’re not stuck. You left a trail everywhere. You don’t need perfect notes to write a real annual review; you need the record you already made. Reconstruct the year from the systems that recorded it for you: your git and pull-request history, your calendar, your closed and shipped tickets, the Slack threads where a decision landed (search your own name and “thanks”), your sent email, and any 1:1 or goals doc. Walk the twelve months from January forward and write down what you find, and you’ll have more specifics than you expected, all sourced rather than remembered.
Then fix the actual problem so next year isn’t the same scramble. Start a running log today: one dated line whenever something lands. By next annual cycle, the year is already written and your self-review is reading, not archaeology. For turning those reconstructed facts into specific review lines, the self-evaluation examples page has the vague-to-specific patterns for the employee side, and how to keep a work diary covers the capture habit itself.
Write next year’s review by reading it, not reconstructing it.
Write the annual review from a year you already wrote down
A whole-year review needs twelve months of dated evidence, and that is exactly what Notivo’s live loop produces. Type @person, write one line when something lands, and tag it #delivery or #growth. It files into that person’s dated timeline and into a topic timeline you can read at review time. Months later you search the name, the whole year comes back grouped and dated, and you pull two examples per competency (one from each half) straight into the annual template. Prep stops being recall and becomes reading.
The AI assistant, available on paid plans and gated on your consent, works over your own notes only: it can recap the year you already captured to speed the read. It does not draft, auto-write, or submit your review or self-review, and it never invents work that didn’t happen. You write the review; the organized, dated year in front of you is what makes it specific. What’s live today is the part this page depends on: web, iOS and WhatsApp capture, the ChatGPT and Claude connectors, @people and #topic timelines, search, and export.
- Type @person, write one line. The note files itself into that person’s timeline. No folders, no “annual-review-v2-FINAL.”
- Tag the competency. Add #delivery or #communication and the same note also lands in a topic timeline you can read at annual-review time.
- Search the name at year-end. The whole cycle comes back dated and grouped, so each rating cites two real moments from different quarters.
On storage, the honest answer: your notes are private by default and scoped to your own account, and they are stored on secured infrastructure. That is platform-level security, the protections that come with reputable cloud storage. It is not a claim that your note content is encrypted end-to-end, and we are not going to pretend otherwise. The one exception is the Black Box vault, which is PIN-locked and end-to-end encrypted for the truly sensitive lines. For everything else, use the same judgment you would use with any cloud notes app.
Give everyone credit for the whole year, not just the last quarter.