Spreadsheets
No rhythm, no owner, no quality bar. Nobody is ever told anything.
From 
OKR Institute · Built on open source
Stop storing OKRs.
OKR Institute · Built on open source
Stop storing OKRs.
Start running them.
OKRGoal is where your organisation's OKRs live and run, with the method built in. The OKR Coach checks every objective as you write it. The OKR Champion keeps the weekly rhythm in Slack, Teams, WhatsApp and email. It runs on OpenOKR, which is open source, so you can also host it yourself, free.
Every feature on every plan. No credit card.
$
Or self-host OpenOKR in about half an hour
git clone https://github.com/open-okr/open-okr && cd open-okr/deploy/docker && ./openokr upQ4 · Company map
12 on track3 at risk
New customers reach value in their first week
Cut time to first value
Median setup from 9 days to 30.6
Hold twelve customer interviews0.5
New enterprise pipeline
Qualified pipeline from 1.1m to 2.0m0.3
Ten referenceable customers0.5
Support cost per account
Health band, in its own units▲
Recovery OKR draftednew
OKR Coach
KR-5Impact, not effortThis measures activity volume. Ask why: more calls, to what end? If you can measure that impact, make it the key result and keep the activity as a leading indicator.
OKR Champion
checkin.duein SlackCheck-in due today: Cut time to first value
Check inOpen in OKRGoal
26quality checks on your drafts and your cycle
8guided phases, from annual plan to quarterly close
7places it reaches you, from Slack to your own AI agent
0features behind a paywall
Built by OKR InstituteCoaching organisations through OKRs since 20171,000+ organisations50+ countries
Built on published sources: Doerr · Grove · Wodtke · Castro · Lamorte · Cagan · Google's OKR playbook
The problem
OKRs rarely fail in January. They fail in week six.
Somebody writes the objectives in a spreadsheet. Nobody checks in by March. The quarterly review is a meeting where everyone agrees things went reasonably well.
The method that makes OKRs work has been understood for twenty years. It just never lived in the software.
OKR trackers
A database with a progress bar. They store your objectives faithfully, then wait to be visited.
Consultants
The method is real. Then it walks out of the door with the consultant, and nothing keeps it going on a wet Tuesday in week six.
OKRGoal fixes this two ways at once. The method is in the product. And the product is active.
The method
Rules that run, not a template.
The whole OKR method ships as rules that run on your real OKRs: a guided eight-phase cycle, twenty-six quality checks, eighteen of which judge every line as you type, six publish gates, an alignment score that names its gaps, and a diagnostic at the close. It is all written down in one public document, and a test fails our build when the code and the document disagree.
Objective
Launch the new mobile app
OBJ-1Your objective starts with a deliverable. If we do it and nothing changes, did we succeed?
KR 1
Hold 12 customer interviews
KR-5This measures activity volume. Ask why: more calls, to what end?
KR 2
Weekly active users to 5,000
KR-2A target but no baseline. Without the from, you cannot prove movement.
Strength
41
- A guided cycle.
Eight phases. It knows what is missing and how many weeks you have left.
- Quality as you write.
Twenty-six checks, each with a coaching prompt. The commonest mistakes also show a weak and a strong version side by side.
- Publish gates.
Owners named, key results passing, alignment stated, capacity checked.
- Alignment that means something.
Contribution, not copying. A score that names every gap.
- An honest close.
Was the miss a strategy problem or a rhythm problem? It tells you which is more likely.
- Your way of doing it.
Twenty-five settings and five profiles. Every quality check and publish gate can block, warn or be off.
The agents
Two teammates who never forget a check-in.
Every workspace comes with two AI members. They have names and they show up in the feed. They start the conversation. You do not have to remember to ask.
The OKR Coach guards quality
Reviews every draft against the rules. With an AI provider connected, runs a nightly sweep for duplicated metrics and hidden dependencies. Speaks up when it matters: when the not-doing list is empty, when nothing was cut at the capacity check, when a goal says "on track" but one of its key results has not moved.
OKR Coach
quality.orphan_goalGoal has no parent: Grow enterprise pipeline
The OKR Champion guards the rhythm
Reminds the goal's champion before a check-in is due. Escalates, visibly, when it is late. Holds every blocker to the next check-in. Announces the weekly session, assembles the digest for the coordinator to post, keeps the streak, and drafts a recovery OKR when a KPI stays unhealthy.
OKR Champion
blocker.overdueto the coordinatorBlocker overdue
A dependency blocker has been open for 9 days.
It blocks: Qualified pipeline from 1.1m to 2.0m
Next action: Legal review of the contract template
ResolveTake it onOpen the board
Every message cites the rule behind it. Open the rule. Disagree with it. That is the point.
Where it reaches you
It comes to you.
People do not open an OKR tool on a Thursday afternoon. So the Coach and the Champion go where people already are. Slack, Teams, WhatsApp and Telegram work both ways: the nudge goes out, a real check-in comes back. Email brings a link straight to the check-in.
Browser
Email
Slack
Microsoft Teams
WhatsApp
Telegram
Claude
ChatGPT
Cursor
Slack
Microsoft Teams
WhatsApp
Telegram
Claude
ChatGPT
Cursor
Cursor
ChatGPT
Claude
Telegram
WhatsApp
Microsoft Teams
Slack
Email
Browser
A check-in typed into WhatsApp runs the same permission checks as a click in the browser. No chat app connected? Email and the browser cover everything.
How is it going? Answer one of: on_track, caution, off_track.
on_track
How confident are you it lands, from 0 to 10? A number on its own.
6
One line on why. Anything you would say out loud.
Two deals moved to proposal. Legal is slow on the contract template.
What is "Ten referenceable customers" at now? A number, or "skip".
5
Checked in. Your reviewer has been asked to acknowledge it.
The rhythm
The meeting runs itself. Your team does the thinking.
At most one nudge per subject per day, and five per person per week. Quiet hours in your own time zone. An escalation to someone else still gets through, because that is its job. A snooze never hides what you owe.
AI-native, never AI-dependent
Turn the AI off. It still coaches.
Every rule, escalation, gate, score and diagnostic in OKRGoal is code, not a prompt, and so is every nudge but one. Switch the AI provider off and all of that still runs. That is why it fits regulated, air-gapped and AI-sceptical organisations that would turn away a product that depends on a model.
Always on, because it is code
running- Twenty-six quality checks and six publish gates
- Scores, confidence and health
- Every escalation, and every nudge but one
- The alignment score and its gaps
- KPI bands and the recovery draft
- The closing diagnostic
Added when you connect an AI provider
running- Rewrites a failing objective to pass the rule it broke
- Drafts an overdue check-in from where each key result stands
- Finds duplicated metrics and hidden dependencies across teams
- Drafts the digest, the minutes and the retrospective
- Answers questions about your workspace, with citations
- Puts all of it in plain language
Proposes, you approve
Sees only what it is given
Hard cost cap
Quiet by design
Bring your own model
Local models, zero egress
Anthropic, OpenAI, Gemini, OpenRouter, Ollama, or any OpenAI-compatible endpoint. Your key, your bill, your choice.
Agent-native
Bring your own agent.
Almost everything a person can do in OKRGoal, an agent can do, through one permission-checked contract. Connect Claude, ChatGPT, Cursor or your own agent. It signs in as you, through a consent screen, and works only within your permissions. Every call is audited.
- "What do I owe this week?"
- "Draft my check-in for the onboarding objective from what changed since Monday."
- "Review this quarter's OKRs against the method and tell me the three weakest key results."
In the product, one button copies a setup prompt you paste into your agent. No token to copy, and nothing secret in your clipboard.
claude mcp add --transport http openokr https://app.okrgoal.com/api/mcp
# Then run /mcp, choose openokr and sign in.# Add OKRGoal to Hermes Agent as an MCP server:
https://app.okrgoal.com/api/mcp
# Then sign in through the consent screen.# Add OKRGoal to ChatGPT as a connector:
https://app.okrgoal.com/api/mcp
# Then sign in through the consent screen.
Deploy
Same product. Your choice of where it runs.
OKRGoal
We run it. Sign up and start. Backups, upgrades and monitoring are ours. Your OKRs are yours.
OpenOKR on one server
Free, on your own machine. Docker Compose and a setup wizard. About half an hour. PostgreSQL is the only thing it needs.
$
./openokr upOpenOKR on Kubernetes
For universities and large organisations. A Helm chart with your database, your single sign-on and your backups.
$
helm install openokr oci://ghcr.io/open-okr/charts/openokrThe same release everywhere. Export your whole workspace from one and import it into another, in either direction, whenever you like.
Who it is for
One product. Every seat at the table.
One place, or one chat message, that says what you owe this week and lets you do it in under a minute.
Check-in: Cut time to first valueDue last MondayOverdue
Check-in: New enterprise pipelineDue tomorrowThis week
Blocker: legal review on the new contract templateNext action due by your next check-inBlocker
A change proposed by the OKR CoachThe Coach proposes. You approve.Review
A composer that tells you why your key result is weak while you write it, and an overdue check-in drafted for you from where each key result stands.
Key result
Hold twelve customer interviews
KR-5This measures activity volume. Ask why: more calls, to what end? If you can measure that impact, make it the key result and keep the activity as a leading indicator.
Status
At risk
Confidence
6 / 10
Value now
1.4m
Overdue, so drafted. You add the numbersPublish check-in
A weekly session the product runs for you: confidence, blockers with owners, commitments, a digest, and a streak.
Confidence round, revealed together
2 blockers, both with owners5 commitments3 winsStreak: 7 weeks
The whole company on one map, stale goals named in your weekly digest, escalations too if you join the ladder, and a close that names what to fix.
Cut time to first value0.7
New enterprise pipelineEscalated in week five, recovered0.6
Support cost per accountRecovery OKR in progress0.3
Closing diagnostic. Was the miss a strategy problem or a rhythm problem? It tells you which is more likely, and names what to fix next quarter.
Cycles across the organisation, KPI driver trees with recovery OKRs, alignment health with named gaps, and everything exportable.
72
Alignment health, with its gaps named
AL-1Grow enterprise pipeline has no parentDuplicateThe same metric in two spaces
ExportEvery cycle, every space, one archive
Single sign-on, directory sync, backups with tested restores, a tamper-evident audit log and hard AI cost caps.
Single sign-on OIDC and SAML 2.0
Directory sync SCIM 2.0
Backups restore provedVerified
Audit log hash-chainedChain intact
AI cost cap halts a run mid-flight
Open source
Built in the open. Argue with the rules.
OKRGoal runs on OpenOKR, which is open source under AGPL-3.0. The method is a public document, METHOD.md, with every rule, threshold and session agenda the product enforces and the published source each one rests on. If you think a rule is wrong, open an issue. If you are right, the product changes.
Questions
Still deciding? Talk to us, or open the live demo and look around.
What is OKRGoal?
The OKR platform your organisation runs its OKRs on, from OKR Institute. It holds your objectives, key results and KPIs, guides the planning cycle, checks every objective and key result as you write it, runs the weekly check-in and the quarterly review, and comes with two AI teammates, the OKR Coach and the OKR Champion, that guard quality and keep the rhythm.
Is it a course, or a way to learn OKRs?
No. OKRGoal is where your teams write, check in on and review their real OKRs. People do pick up the method along the way, because the Coach explains every rule it applies. If you want to learn OKRs before you start, OKR Institute runs courses and certification, separately from the product.
Who is behind it, and how do OKRGoal and OpenOKR relate?
OKR Institute, which has coached OKR rollouts since 2017. It built OKRGoal and OpenOKR on what it learned running its OKR platform, okri.io, and coaching organisations through OKRs. OpenOKR is the open source product, under AGPL-3.0. OKRGoal is OpenOKR in the cloud, run for you. It is the same release either way.
Is it really free?
Self-hosted OpenOKR is free, with every feature and no limit on people. OKRGoal has a free plan, and paid plans that differ in AI credits, seats and support. Never in features.
Do I need to use AI?
No. Every rule, escalation, score and diagnostic works with AI switched off, and every nudge except the one that needs a model to judge meaning. AI adds drafting, rewriting and plain-language summaries. It never makes a decision.
Will the agents change our OKRs without asking?
Not by default. An agent proposes a change and a person approves it. You can run an agent in sandbox mode, where it commits nothing, or give one narrow agent direct writes if you trust it.
We do OKRs our own way. Will it fight us?
It has twenty-five practice settings and five starting profiles, including Google-style and Radical Focus. Every quality check and publish gate can block, warn or be off. By default, only structural problems block publishing. Everything else coaches.
Can we move from a spreadsheet or another tool?
Yes. Upload a CSV or Excel file of goals, key results, KPIs, initiatives or tasks. You confirm the column mapping, see a dry run, then import.
Where is our data?
On your own servers if you self-host, offline if you need it. In the cloud, in the region we state, and nobody at OKR Institute can open your workspace through the product unless a workspace administrator grants a time-limited session.
Run your next quarter on it.
Start free in the cloud, or self-host in half an hour. Either way, every feature is there from the first minute, and the Coach is already checking your first draft.
Are you an AI agent, or is one helping you?