# Rules that run, not a template. > OKRGoal ships the full OKR method as rules that run on your OKRs: a guided eight-phase cycle, twenty-six quality checks on your drafts and your cycle, publish gates, an alignment score and a closing diagnostic. Every rule is public and changeable. This is the Markdown version of https://okrgoal.com/product/method *The method* Most OKR tools give you a form and a help article. OKRGoal gives you the method itself, as rules that run on your real OKRs while you work: a guided cycle, twenty-six quality checks on your drafts and your cycle, six publish gates, an alignment score, and an honest diagnosis at the close. [Start free](https://app.okrgoal.com/sign-up) · [Read METHOD.md](https://github.com/open-okr/open-okr/blob/main/docs/specification/METHOD.md) *One document* ## The whole method, in one public document. Every rule, threshold, band, gate, session agenda and diagnostic OKRGoal enforces lives in one document, METHOD.md, published with the open source OpenOKR code. It compiles into a small library with no database, no network and no AI. The same rules run in your browser as you type, on the server before anything is saved, inside the agents, and in the importer. A test compares the document and the code on every change. If they disagree, the build fails. So what the documentation says is what the product does. *The eight phases* ## A cycle that knows where it is. The product knows which phase you are in, what is missing, who owes what, and how many weeks are left before the publication date. Completion is computed from the work, not ticked by hand. By default the phases guide. Anybody who can edit a space can write and change its OKRs at any time. Make them binding if you run a formal process, or hide them if you only want to track. 1. **Annual strategy** Once a year. Mission, vision, two to five annual strategies, up to five annual objectives, and the year's not-doing list. 2. **Prepare** A sponsor and a facilitator, and the seven-item input pack gathered and shared. 3. **Diagnose** Score last cycle, read the KPI baselines, rank at least three strategic issues by impact. 4. **Set direction** In the annual cycle, three to five priorities, each with what will be measurably different in twelve months, then the not-doing list. Each quarter checks that the annual frame still holds. 5. **Draft OKRs** Objectives and key results, with the quality checks running on every line. 6. **Align and commit** Map contribution, confirm dependencies, check capacity, clear the publish gates. 7. **Run the cadence** Weekly check-ins, monthly reviews and a decision log, with every session booked for the whole cycle. A target can change at any time, with a written reason when it gets easier. 8. **Review and learn** Score, retro, diagnose, and feed it all into the next cycle. *Twenty-six checks* ## Twenty-six checks, on every line and every cycle. Each check returns pass, warn or fail, with a coaching prompt and the reason it matters. The commonest mistakes also show a weak and a strong version side by side. By default only structural defects block publishing: a missing key result, target, date or owner. Everything else coaches, and you decide what blocks. 5 checks ### Objectives - `OBJ-1`Outcome, not output - `OBJ-2`Qualitative and memorable - `OBJ-3`Timebound - `OBJ-4`Owned by a champion - `OBJ-5`Not too many 7 checks ### Key results - `KR-1`Two to five per objective - `KR-2`Verifiable - `KR-3`Complete: target, date, owner, baseline - `KR-4`Leading and lagging - `KR-5`Impact, not effort - `KR-6`Ambitious but honest - `KR-7`Direction set 6 checks ### Alignment - `AL-1`Supports a bigger priority - `AL-2`One parent only - `AL-3`No level skipped - `AL-4`A company anchor - `AL-5`Dependencies declared - `AL-6`Not siloed 8 checks ### The cycle - `CY-1`Input pack gathered before session one - `CY-2`Prior cycle scored - `CY-3`Strategic issues ranked - `CY-4`Three to five priorities - `CY-5`The not-doing list is written - `CY-6`Capacity checked - `CY-7`Dependencies confirmed - `CY-8`Every session booked Some of what they catch "Your objective starts with a deliverable. If we do it and nothing changes, did we succeed?" · "This measures activity volume. Ask why: more calls, to what end?" · "A target but no baseline. Without the from, you cannot prove movement." · "Trim it. If your team cannot recite it from memory, it will not steer their daily decisions." · "Which priority does this move forward? Align it to one, or say why it stands alone." The checks are measured against real, published OKRs, and no check that matches words blocks by default. A person decides what blocks, not a word list. *Anatomy of a coaching card* ## Nothing is a mysterious red dot. 1. 1 **The verdict.** Pass, warn or fail. 2. 2 **The rule.** Its key and name, `KR-5 Impact, not effort`. Click it to read the rule in full. 3. 3 **The prompt.** The question that exposes the problem. 4. 4 **The reason.** Why this matters, in one sentence. 5. 5 **The example.** A weak version and a strong version, side by side. With an AI provider connected, a rewrite of your own line you can apply or dismiss. *Alignment that means something* ## Contribution, not copying. A team objective that restates the company objective aligns nothing. In OKRGoal, vertical alignment means this goal moves that one. Horizontal alignment is a dependency both teams know about, which can be escalated when it stalls. The product scores alignment health as the share of goals that are aligned or stand alone with a stated reason, and names every gap: no company objective anchoring the tree, an orphan goal, an objective with no key results and, if you turn those checks on, a skipped level or a department with no cross-team dependency anywhere in its branch. With an AI provider connected, the Coach's nightly sweep adds what structure cannot see: two goals counting the same metric, a goal that would sit better under a different parent, a dependency between two teams who have not spoken. *Committed, aspirational, and four kinds of key result* ## Scored by what they are. Two kinds of objective ### Committed 1.0 Expected to land at 1.0. A miss asks for an explanation. ### Aspirational 0.7 A real stretch. Landing at 0.7 is a good result. Four kinds of key result ### Metric Move a number from a baseline to a target. ### Maintain Hold a number inside a band. ### Milestone Done by a date, or not. ### Baseline Measure something nobody has measured yet. Every key result has an owner, a weight, a due date, a confidence and its full value history. Metrics get a trend forecast. Easing a target mid-cycle asks for a reason and keeps the original on record. "It got hard" is not a reason. *Defaults, not dogma* ## Your flavour of OKRs, without starting from a blank page. Every rule is a recommended default, and every one can change. Twenty-five practice settings cover who may write and when, how binding the phases are, the kinds of objective and key result in use, which checks and gates block, warn or are off, how confidence is shown, score colours, how the quarterly review runs, and more. Five profiles set them all in one move. ### Recommended Most organisations. The defaults in the method. ### Google-style Organisations following [Google's OKR playbook](https://rework.withgoogle.com/intl/en/guides/set-goals-with-okrs) closely. ### Radical Focus Small companies and teams starting out, in the style of [Christina Wodtke's book](https://eleganthack.com/the-art-of-the-okr/). ### Lightweight Teams that want to track OKRs without the planning workflow. ### Governed Organisations that run a formal planning process. Every change is audited. A closed cycle keeps the rules it was graded under, so changing a threshold next year never rewrites last year's verdicts. Profile names describe published OKR methods. No endorsement is implied. *Built on published sources* ## We did the reading. And the coaching. OKRGoal and OpenOKR are built by OKR Institute, on what it has learned since 2017 coaching OKR rollouts and running its OKR platform, okri.io. Every rule also names the published source it rests on. Where the sources disagree, the document says so and picks a default you can change. The method is written in our own words. No third party's copy, branding or course material appears in the product. 1. John Doerr, Measure What Matters, and whatmatters.com 2. Andy Grove, High Output Management 3. Google's OKR playbook and Google re:Work, Set goals with OKRs 4. Christina Wodtke, Radical Focus 5. Felipe Castro, on OKR levels and alignment 6. Ben Lamorte, The OKRs Field Book 7. Marty Cagan, on product teams and OKRs 8. Rick Klau, How Google sets goals *Argue with it* ## Think a rule is wrong? Tell us. Every coaching message links to its rule. Every rule is public. If you run OKRs for a living and think a threshold, a check or a session agenda is wrong, open an issue with your reasoning and your source. If you are right, the rule changes for everyone, with your name on the change. [Read METHOD.md](https://github.com/open-okr/open-okr/blob/main/docs/specification/METHOD.md) · [Challenge a rule on GitHub](https://github.com/open-okr/open-okr/issues) ## Draft your first OKR with the Coach watching. [Start free](https://app.okrgoal.com/sign-up) · [Try the live demo](https://demo.okrgoal.com/) --- OKRGoal, from OKR Institute. A map of this site for AI agents: https://okrgoal.com/llms.txt