Performance Management Module

The Performance module helps you manage goals (OKRs), reviews, feedback, and team calibration to drive employee growth and organizational success.

Overview

The Performance module provides:

Goals & OKRs - a compact, expandable list of goal alignment
Performance Reviews - 360° feedback and formal reviews
Continuous Feedback - Real-time recognition and feedback
Talent Calibration - 9-box grid for succession planning
Check-ins - Regular 1-on-1 meetings

Contractors and freelancers are left out of review cycles, reviews and probation.

Goals & OKRs (Objectives and Key Results)

Goal View

Goals are shown as a compact, expandable row list: click a row to expand its key results. Each row shows the objective, owner, KR count, and a progress bar.

Visual Indicators:

Progress Bars - Real-time completion percentage
Status Badges - On Track (green), At Risk (yellow), Behind (red)

Alignment Benefits:

See how your work contributes to company goals
Identify orphaned goals (not aligned to anything)
Understand department priorities
Track cascading progress

An individual OKR linked to skills shows Learning for this goal on its detail page, under the progress bar, with the same course rows as My Development: the owner can enrol, ask their manager or ask for a course; a manager sees the courses.

Creating Goals

Creating and editing a goal are one page. New goal starts by asking how you want to begin:

Describe it in plain language: say what you are trying to achieve and Sparko drafts the objective and key results for you to edit. Sparko never makes up a starting value: where your workspace already tracks the measure (time to hire, time to fill, attrition, eNPS or onboarding completion) and you can see it, the key result starts from your own figure. Otherwise the Start field is empty and marked Set your baseline, and the goal saves once you fill it in.
Start from a roadmap item: pick a planned item and Sparko turns it into a goal. Available to managers and HR. A goal from a roadmap item is always a goal for the department that owns the item, so the owner is set for you (see Approving goals from the roadmap).
Write it myself: skip AI and go straight to the form.

Whichever you pick, you land on the same draft form: objective, description, key results, parent goal, timeframe, owner and collaborators. Nothing is saved until you save it, and anything AI drafted is fully editable first.

Editing an existing goal opens the same form, pre-filled. Once the quarter is running, the objective wording and each key result's measurement type become read-only, so progress recorded so far still means what it meant. Values and outcomes stay editable, and the form says which fields are locked and why.

Key Results

Each goal should have 3-5 measurable Key Results (KRs) that define success:

Good Key Results:

✅ "Close 100 enterprise deals"
✅ "Reduce churn to <5%"
✅ "Launch mobile app to 10,000 users"

Bad Key Results:

❌ "Improve sales" (not measurable)
❌ "Work on product" (not specific)
❌ "Be better" (not actionable)

Progress Tracking

Manual Updates:

Update progress percentage directly
Mark key results as complete
Add progress notes

How values show: each key result reads in its own measurement. A percentage shows as 11%, a currency amount in your company's currency with short units ($38M), a count with thousands separators, and a unit such as days or ms after the number.

How progress adds up: a key result's progress is how far its current value has moved from its starting value toward its target, so a lower-is-better key result counts down from its baseline. A goal's progress is the average of its key results. Status follows progress: 100% is Completed, 70% or more On Track, 40% to 69% At Risk, under 40% Behind, and 0% Not Started. The one status you set yourself is Completed: a goal you close with Mark complete stays Completed.

Trajectory and completion forecast

Open any OKR and you'll see a trajectory chart of how progress has moved over time, plus an AI completion forecast: Sparko's read on whether the objective lands by its end date, and what is putting that at risk. A factor that works against the goal shows a warning sign; one working in its favour, such as a key result ahead of plan, shows a tick.

The forecast is written when progress changes rather than when you open the page, so the OKR loads immediately instead of waiting on the model. If an objective has never been updated, the forecast fills in the first time someone views it. If the forecast cannot be written, the card says so and Sparko tries again an hour later.

What it uses: the objective's progress history, its key results, and how much of the period is left. It does not read anything outside the OKR.

Treat it as a prompt, not a verdict. A forecast is a projection from progress so far; a key result that was always going to land in the final fortnight will look at risk until it lands.

Goal Alignment

Parent-Child Relationships:

Company goals cascade to department goals
Department goals cascade to team goals
Team goals cascade to individual goals

Alignment Rules:

Maximum 2-3 children per parent (avoid overcommitting)
All department goals should align to company goals
Individual goals should align to team/department goals
Orphaned goals trigger warnings

Analytics on OKRs

Timeline & prediction: each OKR has a timeline view of its historical check-ins plus an AI-projected completion probability.

Strategic Roadmap

A dedicated space for department/company-level initiatives that are bigger than a single OKR cycle. Create and manage roadmap items, get AI field suggestions while filling one out, and convert a roadmap item straight into a goal for the department that owns it.

Switch between List and Timeline with the icons at the top right of the header.

Approving goals from the roadmap

A goal created from a roadmap item goes through a short approval step before everyone can see it. Who does what depends on who converts the item.

When a manager converts an item

You can convert items that your own department owns. The goal belongs to your department.
The goal is saved with a Pending HR badge. Until HR accepts it, only you and HR can see it, and progress updates and check-ins wait until it is accepted.
If HR accepts it, the goal is published and you are notified.
If HR returns it, the badge changes to Changes requested and HR's note appears at the top of the goal. Edit the goal, then click Resubmit to send it back to HR.

When HR reviews a manager's goal

Reviewing needs the Approve OKRs from the Roadmap permission (performance.okrs.approve), which HR Admin and Super Admin hold.

Open Performance > Goals and pick the Pending tab to see every goal waiting on HR, or use the goal row in My Hub > Approvals.
Click Accept to publish the goal, or Return to send it back to the manager. Returning always needs a note that tells the manager what to change. There is no reject: a returned goal stays with its roadmap item until the manager resubmits it.

When HR converts an item

The goal is published straight away for the department that owns the item.
The department head is asked to accept it. Until they do, the goal shows an Awaiting manager badge.
The department head sees the goal in My Hub > Approvals and on the goal page, and can click Accept or return it to HR with a note (Return to HR on the goal page). A returned goal shows a Returned to HR badge; when HR edits and saves it, it goes back to the department head.
If the department has no head, the goal is published without this step and Sparko tells you so.

Performance Reviews

360° Feedback

Comprehensive feedback from:

Manager (formal review)
Peers (lateral feedback)
Direct reports (upward feedback)
Self-assessment

Setting up a cycle

New cycle in Manage Cycles opens a quick form with the essentials; its use the full setup link goes to the setup page described here.

Start a review cycle opens a two-pane setup page. The main column holds six numbered sections you work down:

Audience: who is getting reviewed, with a live count of how many people that is. Narrow it with the same audience filters used elsewhere in Sparko.
Cycle basics: name, review type (Mid-Year, Annual, Quarterly or Custom), and the period it covers.
Schedule: the milestones from launch to final delivery. Self-review, manager review, acknowledgement and final delivery are always on; skip-level review, HR review and the calibration session are optional, and the count shown in the section, the summary line and the rail is the number the cycle saves. Dates are derived from the ones you set.
Inputs: which review template the cycle uses. It follows the review type by default, and you can override it for this cycle.
Calibration: cross-manager rating alignment, tied to the calibration milestone above. Employees are never present.
Notifications: which reminders go out and when: cycle opens, mid-window nudge, self-review deadline, acknowledgement reminder, results released.

A rail on the right stays with you as you scroll, showing a live summary of what you have configured and the Launch button. Launch creates the cycle with the whole configuration and activates it in one step.

Launching needs the Manage Review Cycles permission. If you do not have it, the page says so before you fill anything in rather than after.

Review Cycle:

Launch review cycle (HR/Admin)
Employees complete self-assessment
Peers/managers provide feedback
Manager completes final review
Calibration session (managers align ratings)
Share results with employees

Pausing and editing a running cycle

A running cycle can be paused from its detail page. While paused, reminders stop and nothing can be submitted: self-assessments and manager reviews are both held until you resume. Resuming picks up where the cycle left off. Pausing is reversible, which is what makes it different from cancelling.

Edit settings changes a cycle after launch, within limits. Anything that would strand work people have already done is locked, and the modal says which and why:

Whether the cycle feeds compensation, performance or equity is fixed once the cycle launches, because it decides which worksheets and budgets exist.
Self-review can't be switched once reviews exist.
Peer surveys can't be switched once surveys have gone out.

A draft cycle locks nothing.

AI synthesis of a 360 round

Once at least three peers have answered, HR can generate an AI summary of the round from the reviewer's detail view. The summary groups what reviewers said into themes. It never names a reviewer, never quotes an answer word for word, and folds any relationship group with a single reviewer into "colleague" so nobody can be identified by elimination.

Choosing who gives you feedback

Performance > Request Peer Feedback opens with a suggested roster. Sparko ranks colleagues by how closely you have actually worked together lately: goals you share, recurring 1-on-1s, feedback either of you gave the other, and reviews of each other's work. Recent collaboration counts for more than older collaboration, and anyone already asked for this cycle is left off the list.

Each row says why the person was suggested. Deselect anyone who shouldn't be included, use Add peer manually for someone the suggestions missed, add an optional note, then send. Nothing goes out until you send. Your note appears at the top of each peer's survey under "What they'd like to know".

Answering a feedback request

When a colleague asks you for feedback, through the assistant or a peer survey, it waits on the To complete tab of Performance > Feedback, and My Hub links straight to it. A request from the assistant shows who asked and their note; Give feedback opens the feedback form with them already chosen and what they'd like to know, their note and any questions, above the form. Sending it answers the request and removes it from the tab. When the request asks for anonymous answers, Send anonymously is already ticked and cannot be cleared, so the colleague who asked never sees your name.

If the ranked suggestions can't be loaded, the page says so and falls back to a simpler list built from your manager, 1-on-1 partners and past feedback exchanges, so you can still send requests.

Peer survey questions

Every peer survey stores the questions it was created with. Changing the standard set later doesn't rewrite what an in-flight survey asked, so a reviewer's answers always match the questions they actually saw.

Before launching, HR/Admin can run an audience preview: a dry-run that resolves exactly which employees the cycle's audience scope/config will include, without committing anything, so you can catch an over- or under-broad audience before the cycle goes live.

During a calibration session, a budget-impact view overlays the recommended salary adjustments coming out of calibration decisions against the active merit-cycle budget envelope, so calibrators can see the cost of their decisions in real time.

Once a review is delivered, the employee reads and acknowledges it on the final review page. An AI-drafted manager comment there speaks to the employee directly ("You led the migration", "your planning"), where the manager's own view uses their first name. An Export PDF button on that page opens the browser's print dialog with a clean, chrome-free copy of the finalized review for saving or printing.

Review Templates & Question Library

Admins build the questions employees and managers answer during a cycle. Open the builder at Performance > Reviews > Templates (/performance/reviews/templates).

Question library (your reusable bank). Every question you write is saved to a tenant-wide bank so you can reuse it across templates instead of retyping it. Each question is one of two kinds:

Reflection. A self-assessment prompt the employee answers about their own work.
Open. An open-ended question a reviewer answers about the employee.

Composed templates. A template is a named set of bank questions tied to a review type: Annual, Semi-annual, Quarterly, or Custom. Compose it by picking questions already in the bank or typing new ones inline (new text is added to the bank automatically). Each review type has a default template the cycle picks up when none is chosen.

Draft a template with AI. Click AI generate, choose the review type, and describe what you want the review to cover in a sentence or two. Sparko drafts a full set of questions you can edit, reorder, or swap before saving. As with all Sparko AI, no cost is shown to you.

Viewing templates and the bank needs performance.review_templates.view (everyone who runs reviews has it). Building and editing them needs performance.review_templates.manage (HR Admin / Super Admin).

Growth Trajectory

An employee's profile Performance tab plots their growth trajectory: the overall review rating as a dot-and-line across every completed review cycle, plus the four fixed component ratings (goal achievement, technical skills, collaboration, leadership). It appears once there are two or more completed reviews, so there is a trend to draw. Before that the profile says there is not enough review history yet. It is a read view of ratings the employee has already been given, not a new rating surface.

AI-Powered Performance Review Generation

Sparko's AI can now generate high-quality performance reviews automatically by synthesizing data from across the platform.

Two Ways to Get AI-Generated Reviews

1. Via Ask Sparko (every plan; the /ai-chat page is Growth and above, Core asks from the command palette)

Ask the AI assistant to write a performance review:

"Write a performance review for Connie Hennelly for Q4 2025"

The AI will:

✅ Pull data from OKRs, 1-on-1 notes, and 360 feedback
✅ Structure feedback in STAR format (Situation, Task, Action, Result)
✅ Check for plagiarism and template overuse
✅ Generate uniqueness score (minimum 70% unique)
✅ Provide quality warnings if issues detected
✅ Display rich preview in chat with sections and badges
✅ Save the review as the type its period reads as: a quarter is a quarterly review, a half-year a mid-year review, a full year an annual review. Name another type ("a promotion review") and Sparko uses that instead.

Drafting takes a minute or so. After you press Confirm, the chat shows "Working on it" and, when the draft is saved, a card naming the person with an Open the draft button. If your connection drops while you wait, Sparko checks whether the draft is still being written before it reports anything, and pressing Confirm again never writes a second review.

✍️ AI-Generated Performance Review

✓ Uniqueness: 87%

✨ Accomplishments

Situation: Connie was assigned to lead the platform launch with a team of 4 developers...

Task: Her goal was to deliver 3 major features by Q4 end while maintaining code quality...

Action: She implemented daily standups, code review automation, and mentored 2 junior devs...

Result: All 3 features shipped on time, customer satisfaction increased 23%, and team velocity improved 40%.

📈 Areas for Improvement

Situation: During the Q3 incident with the payment system...

Task: Connie needed to communicate the root cause to stakeholders...

Action: She focused primarily on the technical fix but delayed stakeholder updates...

Result: While the fix was excellent, executive team expressed concern about communication gaps.

Review Period: Q4 2025 | Rating: 4 - Exceeds Expectations | Sources: 3 OKRs, 8 1:1 notes, 5 feedback items

2. Manual Creation with AI Assist (Coming Soon)

When creating a review manually, click the "AI Assist" button to:

Get suggestions based on employee data
Fill in sections automatically
Improve language and structure
Ensure STAR format compliance

AI Review Quality Features

STAR Format Enforcement:

All accomplishments and improvement areas structured as:
- Situation - Context and background
- Task - What needed to be done
- Action - What the employee did
- Result - Measurable outcome

Multi-Source Data Synthesis: The AI pulls information from:

✅ OKRs - Goal progress and key results achieved
✅ 1-on-1 Notes - Manager feedback and discussion points
✅ 360 Feedback - Peer and upward feedback
✅ Growth Data - Development-plan milestones completed, certifications earned

Plagiarism & Quality Detection:

✅ N-gram Fingerprinting - Detects template overuse
✅ Uniqueness Scoring - Minimum 70% unique content required
✅ Template Detection - Flags generic phrases
✅ Quality Warnings - Suggests improvements if score is low

⚠️ Quality Warnings

Review contains 3 common template phrases - consider making more specific
Accomplishments section has low uniqueness (62%) - add more concrete examples
Missing measurable outcomes in 2 STAR examples

Approval Workflow

For AI-Generated Reviews:

Review Created - AI generates draft and saves as "in_progress"
Manager Review - Manager receives notification
Approve & Send - from the Reviews list, an AI-drafted review assigned to you opens "Review the AI-drafted review", with the draft formatted as written and What each claim is based on under the accomplishments and areas for improvement. Approve and send it, add optional notes for the employee, or Reject with a required reason. Where the draft says "the employee", you read the person's first name instead, in the draft and in its claims. On every plan.
Employee Access - Once approved, employee can view

Consolidated Approvals Dashboard:

All pending performance reviews appear in the Approvals page
View employee name, review period, and uniqueness score
Approve multiple reviews efficiently
Track approval history

What an AI-drafted review gives you

A first draft, not a finished review. You read it, edit it, and approve or reject it. Nothing reaches the employee until you approve.
The same structure for everyone. Accomplishments follow the STAR format (situation, task, action, result), so every review in a cycle covers the same ground.
A draft built from the record. Sparko drafts from the period's objectives, feedback, recognition, shared 1:1 notes, self-reflections and 360 input, so the draft starts from what was written down during the period rather than from what you happen to remember.
Claims you can check. Each sentence that rests on evidence lists it under What each claim is based on, and a sentence with no evidence behind it is left uncited.
A check on reused wording. Template phrases and low-uniqueness sections are flagged, so a copy-pasted review does not go out unnoticed.
A rating only when the record supports one. The suggested rating comes from goal outcomes, the self-rating, 360 input, constructive feedback and your own notes. Praise and recognition alone are not enough to place a rating, so when that is all the period holds the draft leaves the rating blank and shows Set the rating rather than defaulting to "Meets expectations".
The person's name, not "the employee". Sparko never sends the name to the AI, so the draft says "the employee"; the review dialog and the chat card put the person's first name back in when you can see their profile.

Where each AI-drafted sentence came from

When Sparko drafts a review narrative, it cites the evidence behind each specific claim. Under the draft you'll see What each claim is based on: the sentence Sparko wrote, and the objective, feedback item, recognition post, 1:1 or reflection it rests on.

If a source check is unavailable, your draft stays available and a short message explains why some sources could not be checked. Confirmed evidence remains visible alongside that message; unchecked sources stay hidden. Reopen the review after the issue is resolved to retry the remaining checks.

This exists so you can check the draft rather than trust it. A generated paragraph reads the same whether it is grounded in someone's actual work or not, and the person being reviewed has no way to tell the difference.

Three behaviours are deliberate:

A sentence with no supporting evidence is left uncited. Sparko will not attach the nearest artifact to a claim that artifact does not actually support. Every citation is checked against the text of the artifact it names: a claim that quotes words the artifact does not contain, or that the artifact is only loosely related to, loses the citation and keeps its sentence. One artifact backs at most two claims in a draft. Drafts saved before this check get it when their reviewer opens them, with deferred checks retried on later opens.
Editing a sentence drops its citation. The citation described the sentence Sparko wrote; once you rewrite it, that link is no longer true, so it is removed rather than silently re-pointed at your words.
A citation has to name evidence Sparko was actually given. Every citation on a draft is checked against the artifacts that went into it, and anything that points somewhere else is dropped before you see the draft. Sources shown under the draft have passed that support check.

What a draft is grounded in:

Objectives and key results that overlap the review period, so a quarterly OKR still counts toward a half-year review
Feedback the person received in the period
Public recognition and shout-outs, including the ones where they were named alongside other people. Private recognition sent as a DM is never used
Shared notes from 1:1s, reflections, and 360 responses in the period

Manager assessments read as stories, not labels. When you draft your own assessment of a report against a competency, Sparko writes it in your voice and walks through what they faced, what they did, and what came of it, as ordinary prose. There are no "Situation / Task / Action / Result" headings to delete. Each draft closes with one development push for next period.

Team accomplishments in a manager's self-review

If you manage people, your own self-review includes a Team accomplishments section: achievements your direct reports described in their self-reviews, re-framed into your voice as a suggestion about your contribution: the coaching, unblocking and prioritising behind their result.

Your reports submit their self-reviews. You'll be notified as each lands.
Open your self-review and use Refresh from team to pull in anything new.
For each suggestion, expand View source to read your report's original wording, then Add to my review, Edit first, or Not this one.

It is a shortlist, not a list of everything. HR sets how many highlights a manager may carry into their own review: 3, 5 or 7, under Admin Settings > Performance. The section header says which cap applies to you and how many of it you have used.

Ranked by impact. Sparko orders candidates by how confident it is in the match, weighted by how widely the sources spread across your team, so one report's achievements cannot crowd out everyone else's. Drag a card to override the order. Candidates ranked below the cap are still listed, so you can promote one.

Nothing is added unless you add it. Suggestions arrive pending and stay that way until you decide.

Their words are always shown. You are being asked to adopt a rewrite of what someone else said about their own work, and you cannot judge that against a paraphrase alone. The original sits behind View source so the card reads as your contribution, with theirs one click away.

It does not cascade upward on its own. If you accept a report's achievement, it does not appear in your manager's review. They have to accept it themselves, from yours, so each person in the chain makes their own call.

If a report edits the achievement away after you accepted it, Sparko marks the suggestion as withdrawn and tells you. It does not rewrite your draft.

If several suggestions come from one person, Sparko prompts you to add a paragraph about your own managerial contribution, so the review describes your work rather than theirs.

HR and calibration can see the full trail: who contributed each item, what Sparko suggested, and whether you accepted, edited or ignored it. Your reviewer and the employee see only the finished text.

Rating Scales

Standard 5-point Scale:

5 - Exceptional: Consistently exceeds expectations
4 - Exceeds: Frequently exceeds expectations
3 - Meets: Consistently meets expectations
2 - Needs Improvement: Sometimes meets expectations
1 - Unsatisfactory: Rarely meets expectations

Continuous Feedback

Real-Time Feedback

Give and receive feedback anytime, not just during reviews.

Types:

Praise - Recognize good work
Constructive - Improvement-focused feedback

Your feedback inbox

Feedback arrives as cards in Performance > Feedback. Unread ones are marked. Each card shows who sent it, unless they sent it anonymously, and a Sent tab shows what you gave.

Cards carry a small row of reactions (👏 ❤️ 🎉 🔥 ⭐), with a + for a wider picker. Click one to add yours, click it again to take it back. Counts show how many people reacted with each. It is a quick way to say you read something without writing a reply.

You can also reply to a piece of feedback, or acknowledge it to clear it from unread.

Public recognition, kudos, awards, and milestones are a separate module: see Engagement for the recognition wall, awards, and points.

Talent Calibration

9-Box Grid

Calibrate talent using a 2x2 grid:

X-Axis: Performance (ratings from reviews)
Y-Axis: Potential (manager assessment)

9 Boxes:

High Potential, High Performance - Stars (promote, retain)
High Potential, Medium Performance - Diamonds in rough (develop)
High Potential, Low Performance - Question marks (improve or exit)
Medium Potential, High Performance - Solid performers (reward)
Medium Potential, Medium Performance - Core contributors (maintain)
Medium Potential, Low Performance - Performance issues (needs a development or exit plan)
Low Potential, High Performance - Specialists (keep happy)
Low Potential, Medium Performance - Limited growth (role fit)
Low Potential, Low Performance - Action needed (exit)

Calibration Sessions

Plan: calibration sessions are on every plan until 31 December 2026. From 1 January 2027 they are on Growth and above. If your workspace ran a calibration session before 1 January 2027, you keep calibration sessions on your current plan.

Process:

Each manager places their team members on grid
Leadership team reviews placements together
Discuss and debate (avoid bias)
Align on ratings and potential
Create development plans

Known limitation: dragging an employee's chip to a different box on the grid does not currently save: the move is visual only and is lost on refresh. Use the calibration session canvas below (with its per-row decision status) to actually record calibration outcomes.

Finding calibration sessions

Performance → Calibration lists every calibration session you can see: HR and anyone who manages calibration see all of them, a manager sees the sessions their own team runs. Each row shows the session, its review cycle (or "No cycle"), the date, how many employees it holds and its status. Filter by status with the strip above the table, search by session or cycle name, and click a row to open the session. The tab appears only for people holding a calibration permission.

Employees is the number of people with a result in the session, the same number the session page shows.

Calibration Session Canvas

The session page at /performance/calibration/sessions/{id} is a facilitator-led analysis tool with four blocks:

Rating distribution: bars per rating bucket (1–5) with target bands. Bars at rating 5 turn amber when >10% of the cohort lands there; bars at rating 1 turn amber when >15% lands there. Click any bucket to drill into its detail panel.
Cohort breakdown: toggle Manager / Level to see per-cohort mini-distributions. Cohorts with 2+ employees at rating 5 OR any at rating 1 get flagged as outliers.
Rating detail: table of employees in the selected bucket with four scores: Self and Manager are the ratings from the person's review in the session's cycle ("Not rated" until that review has one), Data is the score calibration computed from goals, feedback and reviews, and Final is the calibrated score, which equals Data until someone adjusts it. Each row also has a decision status pill and a Discuss/Update action (manage-permission gated).
Session outcomes: three tiles tracking Ready for promotion, Needs developing, and Adjustments made.

Each row's decision is recorded as pending (default) / discussed / adjusted via the decision modal. Status drives the outcomes tile + the sticky-bottom tally + the exported CSV.

Finalize calibration marks the session complete (read-only afterward). Export decisions log downloads a CSV of every result + status + notes for record-keeping.

Check-ins & 1-on-1s

Regular Check-ins

Recommended Frequency:

Weekly: For new hires or struggling employees
Bi-weekly: Standard for most employees
Monthly: For senior/autonomous employees

Check-in Agenda:

Wins & celebrations
Challenges & blockers
Goal progress
Career development
Action items

Meeting Notes

Take notes during 1-on-1s
Tag action items
Review previous action items
Share notes with employee

Development

The detail pane lists development milestones your report proposed from their AI Career Coach. Approve, Adjust the title or date, or Decline with a short note. Approved milestones move onto their plan; a declined one is removed and the note reaches them. The AI agenda includes a Development topic, marked "AI Career Coach", whenever proposals are waiting.

Performance Analytics

Manager Dashboards

View:

Team performance distribution
Goal completion rates
Review cycle completion
Feedback frequency
Calibration data

Needs your attention panel. The performance dashboard lists things worth a look: overdue reviews and check-ins, at-risk goals, 1:1s that have slipped or were never scheduled, and stale goal check-ins. Reminder rows carry an x to dismiss permanently for the whole company; dismissed rows move to the Dismissed items list under the panel, where anyone with the same access can restore them, and every dismissal records who made the call. Rows that are real work waiting on you (pending approvals, feedback requests) have no dismiss control: acting on them is what clears them.

Company-Wide Insights

Performance trends over time
High/low performers by department
Goal alignment coverage
Review completion rates
Feedback culture metrics

Cohort Comparison

The Performance Analytics page at /analytics/performance includes a Cohort comparison tab that surfaces how performance signals move across teams over time. Select any combination of:

Metric: Avg rating (review overall_rating averaged per cell), Promotion rate (career-path level changes per cohort headcount), or Attrition (terminations per cohort headcount)
Cohort: Department, Tenure band (<1y, 1–2y, 2–4y, 4–7y, 7+y), or Hire cohort (e.g., "Hired Q3 2024")
Period: Quarter, Half, or Year (always 6 buckets, newest-right)

The resulting heatmap uses a diverging red↔neutral↔green color scale for Avg rating (centered on 3.0) and a sequential single-hue ramp for Promotion rate and Attrition (both non-evaluative percentages).

Click any cell to drill into the employees behind that (cohort, period) bucket. The drill table shows Name · Level · Rating · QoQ delta · Suggested next step (Promotion packet / Stretch assignment / Calibration discussion / PIP risk).

Scope behavior: Callers with performance.analytics.view_team see cells/drills restricted to their org subtree; view_all sees company-wide.

Best Practices

Goal Setting

Use the SMART framework:
- Specific, Measurable, Achievable, Relevant, Time-bound
Limit goals to 3-5 per person/team
Align all goals to company objectives
Review quarterly, update as needed

Feedback Culture

Give feedback frequently (not just during reviews)
Be specific with examples
Focus on behavior, not personality
Balance positive and constructive
Follow up on feedback given

Reviews

No surprises - Discuss performance regularly
Use data - Reference goals, projects, feedback
Be fair - Calibrate to avoid bias
Focus forward - Development plans for growth

Integration with Other Modules

People - Review employee profiles during calibration
Recruiting - Performance data informs promotion/hiring
Career - OKRs link to career-framework skills; skill gaps surface stretch assignments
Compensation - Performance ratings inform merit increases

Still need help? Reach us at [email protected].

← Back to Help Center