AI + HREvidence-based performance evaluation

Evaluate the work,
not the hours.

Evaluator reads the work your teams already do - tickets, pull requests, key results, documents and peer feedback - and drafts monthly evaluations where every score cites its evidence. The AI drafts. A manager always decides.

  • Shadow mode by default
  • Append-only audit log
  • Every employee can dispute
Evaluation · September 2026

Jamie Rivera

Backend Engineer · Platform

September composite

77.4

▲ 3.1 vs last month

  • Business Outcome & Goal Achievement82
  • Problem Solving & Decision Quality80
  • Ownership & Execution71
  • Quality, Complexity & Business Impact76
  • Collaboration & Reliability74

Evidence behind this score

MSQ-1482platform/api#412KR 2.1Q3 launch review
DraftAI draft - awaiting a manager's approval
Meets expectations

Illustrative sample - not real data

Reads the evidence where work already happens

  • Jira
  • GitHub
  • Slack
  • Hubstaff
  • Notion
  • Google Workspace

Why Evaluator

Most reviews measure visibility. Evaluator measures value.

Three rules hold everywhere in the product - enforced by the engine and tested in every build, not just written in the copy.

Value over activity

Business outcomes, delivery complexity, ownership, problem solving and collaboration are scored. Hours are context, and activity percentages are shown but never scored.

Evidence for every claim

Every number links to the tickets, pull requests, key results and documents behind it. A competency with too little evidence holds a neutral 50 instead of pretending to know.

People decide

The AI writes a draft. A manager reads the evidence, records a written attestation and approves - and every step lands in an append-only audit log.

Product

Everything a review cycle needs, in one place

From the first KPI of the month to the annual summary, every step happens in Evaluator and every decision is on record.

Monthly evaluations

Drafted from the month's evidence, with a narrative, a competency scorecard and a citation behind every claim.

OKRs & monthly KPIs

Company, department, team and individual goals that roll up - with KPI actuals, attached evidence and lead validation.

Quarterly & annual scores

Team OKRs, individual results, competencies and 360 feedback combine into a quarterly score that freezes when HR finalizes it.

360° peer review

Role-aware forms for peers, leaders and team members, with autosaved drafts and an honest 'no basis to judge'.

Fair by construction

Peer bands by level and track, neutral scores for missing data, and no percentile for a group too small to rank.

Disputes with a conversation

Employees acknowledge or dispute. A re-evaluation snapshots the original and never lowers a score because someone objected.

How it works

From signals to a decision in four steps

  1. 1

    Connect

    Link Jira, GitHub, Slack, Hubstaff, Notion and Google Drive. Accounts are matched to people explicitly - nothing is guessed.

  2. 2

    Draft

    Each month Evaluator assembles the evidence and drafts every evaluation - scores, narrative and citations, clearly labelled as a draft.

  3. 3

    Review

    Leads validate KPIs and evidence. Managers record their own read before seeing the AI's, then approve with a written attestation.

  4. 4

    Publish

    People see their approved evaluation, acknowledge it or raise a concern - and the whole conversation stays on record.

Scoring

One model, three horizons

The weights on this page are read from the live configuration, so it always says what the engine does.

Monthly

A composite of five competencies, scored from the month's evidence.

  • Business Outcome & Goal Achievement30%
  • Problem Solving & Decision Quality20%
  • Ownership & Execution20%
  • Quality, Complexity & Business Impact20%
  • Collaboration & Reliability10%

Quarterly

The quarter's final score - frozen once HR finalizes it.

  • 30%Team OKR
  • 20%Individual OKR
  • 30%Role & core competencies
  • 20%360 peer review

Annual

The average of the year's frozen quarters, with a written summary approved person by person.

average of finalized quarters

Governance

Governed like it matters

Evaluator decides nothing on its own. These rails are enforced by the engine - not by a screen that could be skipped.

  • Shadow mode keeps scores admin-only until calibration is signed off
  • Self-review is impossible - the engine refuses it, not just the screen
  • A manager override needs a written note, and the original stays visible
  • Missing data scores a neutral 50 - never zero
  • Every action lands in an append-only audit log
  • An approved evaluation never changes - a later change is an audited re-evaluation

For everyone in the review

Clear for the person being reviewed. Efficient for the people reviewing.

Employees

Know what is expected and see the evidence behind your evaluation.

  • Track monthly KPIs and attach evidence
  • Give and receive 360 feedback
  • Acknowledge or dispute, with a conversation

Team leads & heads

Spend review time on judgement, not on gathering data.

  • Validate KPIs and evidence in one queue
  • Read the evidence before the AI's view
  • Approve with a written attestation

HR & admins

Run the whole cycle, with every decision on record.

  • Windows, deadlines and calibration
  • Completion tracking and nudges
  • Exports, PIP queue and audit log

FAQ

Questions, answered

The short version of how Evaluator treats the people it evaluates.

Does Evaluator track my hours?

Attendance is context only, and activity percentages are shown but never scored. Evaluator measures the value of the work - outcomes, complexity, ownership and collaboration - not the time spent on it.

Who decides my score?

A manager. The AI only drafts. Nothing becomes final until a reviewer approves it with a written attestation - and nobody can review themselves.

What happens if I disagree?

You can acknowledge your evaluation or dispute it. A dispute opens a conversation with HR, and a re-evaluation keeps the original on record - it never lowers a score because you objected.

What if there is not enough data about my work?

A competency without enough evidence holds a neutral 50 and says so on the page. Missing data is never treated as a zero.

Can the AI change a score after approval?

No. An approved evaluation is immutable. Any change after approval is an audited re-evaluation with a written reason, and the original is kept.

Evaluations your people can trust.

Sign in with your company account to open your dashboard.

Sign in

Evaluator

The AI drafts - a human manager always decides. Every action is recorded in the audit log.

© 2026 Evaluator. Built for fair, evidence-based evaluation.