Unknown is a state, not a failure
You can draft an Objective, a metric and a baseline while the target stays open. The audit reports “target decision missing” and the next step is to agree a basis, not to accept a made-up number.
OKR Generator & Key Result Builder
Build team or company OKRs without invented targets: define the Objective, turn each Key Result into a measurable contract, keep initiatives separate from outcomes, and run lightweight check-ins as results change. This is business and team goal-setting, not HR performance management.
- Blank first: nothing is generated until you create it.
- Measurement contract: metric, population, baseline, target and basis per KR.
- Honest unknowns: unknown baseline and target are allowed states.
- Moving ledger: check-ins, learning and target-change history.
No login. Everything stays in this browser. CraftUp never invents a target, baseline, owner, cadence or initiative: missing business facts stay visibly unknown until you decide them.
Share privacy: a share link encodes the whole OKR workspace in the URL. Compression is not encryption, anyone with the link can inspect strategy, targets, owners and constraints. Remove confidential content before external sharing. Oversized workspaces are not shared; export instead.
Workspace context
Context shapes the audit and the draft skeleton. Changing it never silently rewrites your Objectives or KRs.
Your workspace is blank
Nothing is generated until you create it. Add an Objective, draft a skeleton from your own context, or load a clearly labeled synthetic example to see how the measurement contracts fit together.
Check-In Learning Ledger
Record the current value, freshness and what the team learned or decided. The metric refresh cadence and the OKR check-in cadence are independent. Initiative status never advances a Key Result value.
Add an Objective and Key Result first.
OKR Quality & Alignment Audit
Grouped findings, no aggregate score. Unknown baselines and targets are allowed; they produce a decision to make, not a failure.
Workspace summary: 0 Objectives · 0 Key Results · 0 initiatives · 0 check-ins.
How the OKR workspace works
1. Decide the outcome
Set the level, timeframe and strategic intent, then write the Objective you actually intend to pursue. CraftUp can draft a skeleton from the context you supply, but it will not invent a goal.
2. Make every KR measurable
Each Key Result captures a metric contract: definition, population, unit, source, baseline, target, target basis, owner and refresh cadence. Lower-is-better, thresholds, ranges and milestones are all first-class types.
3. Run the loop
Link initiatives to the results they are hypotheses for, audit the set, then check in with the current value, freshness, learning and the decision that follows. Completing an initiative never advances a result by itself.
The core asset: a KR Measurement Contract
A Key Result is not the sentence “increase activation 28% → 35%”. It is the small set of facts that let two people calculate the same number and understand why the target exists. The contract is inspectable and editable, and it can stay partially unresolved without pretending otherwise.
Metric + definition
What exactly is measured, including the value event and window.
Population / denominator
Which eligible population the number is calculated over.
Unit
Percent, count, currency, days, hours, score, ratio, yes/no or custom.
Baseline + period
Where the number stands now, and over which cohort or window.
Target + basis
The decided target and where it came from, or an explicit unknown.
Source + owner + cadence
Which report supplies it, who owns it, and how often it can refresh.
Progress is never forced into one universal percentage: threshold, range, milestone and guardrail results use met / not-met semantics, and lower-is-better metrics move in the opposite direction.
A target without a basis is a guess
The tool never fabricates a target from your audience, your stage or a Key Result position. When a target is decided, you record where it came from: historical trend, prior period, a contractual SLA, a customer commitment, a forecast, a benchmark with a source, a budget/strategy assumption, or explicit team judgment. No single basis is universally correct, and a number you chose is not a benchmark.
Baseline blank ≠ baseline unknown
Leaving the baseline blank means nobody has assessed it. Marking it “unknown” means it was considered and is genuinely not available yet. The audit treats those as different decisions, and a target set from a contractual source can legitimately exist before a baseline does.
Initiatives are hypotheses, not results
Every initiative links to the Key Result it is meant to move and states its expected mechanism. Its status , not started, in progress, blocked, done, paused, stopped, is tracked separately. Marking an initiative “done” does not move the Key Result; only the source data does. Measurement or instrumentation work is welcome, and the audit labels it so it is not read as outcome progress.
Check-In Learning Ledger
A check-in records the current value, the data window and freshness, the change since the prior check, any confidence or data issue, a blocker, the initiative update, what the team learned, and the decision that follows. History is preserved, so a blocker without a decision, a target revised without a reason, or a no-longer-mature metric becomes visible.
Metric cadence ≠ check-in cadence
Revenue may refresh monthly, operational errors daily, and a survey quarterly, while the team meets on its own OKR rhythm. The tool keeps both explicit.
Decisions, not scores
Decisions include continue, adjust or pause an initiative, clarify measurement, revise a target with a rationale, escalate a dependency, or close the Objective. Nothing is auto-recommended from a threshold percentage.
OKR Quality & Alignment Audit
The audit reads the whole workspace and returns grouped findings. It never produces an aggregate score, grade or fake confidence number, because a “strong OKR” is a decision people make, not a number the tool can own.
Measurement blocker
The number cannot be trusted yet, missing metric, conflicting units, a benchmark with no source.
Needs decision
A real decision is open, baseline not assessed, baseline unknown, target not decided.
Review
Worth a look, task-like wording, generic Objective, unclear denominator, missing cadence.
Informational
Context and reminders, shared metrics, constraints, pre-PMF framing, migration notes.
A likely task-style Key Result, a numeric Objective, or an activity-only Objective produce a review finding, not a hard rejection. A verifiable milestone Key Result with a date is allowed.
What this tool owns, and what siblings do better
This route owns drafting and operating a team or company Objective set with measurable Key Results. Strategy, the North Star, driver decomposition, sequencing and experiment design are separate jobs.
| Adjacent job | Owner | Boundary |
|---|---|---|
| Why and where we compete | Product Strategy One-Pager | If the strategy is unresolved, the Objective may legitimately stay unresolved too. This tool does not choose markets or positioning. |
| A durable Product value metric | North Star Metric Finder | A Key Result may reference North Star movement, but this tool does not pick or audit the North Star itself. |
| Driver metrics and decomposition | KPI Tree Builder | A KPI tree decomposes an outcome into drivers; the OKR tool owns the committed outcome and its measurement contract. |
| Sequencing investments over time | Now Next Later Roadmap | The roadmap sequences bets; the OKR tool records the results those bets are meant to move. |
| Testing one initiative | A/B Test Plan Generator | When an initiative is experimental, experiment design and statistics live with the experiment owners. |
| Individual performance or compensation | - | This is not HR performance-management software. It does not rate people, produce rankings, or connect results to pay. |
Synthetic examples you can inspect
Each example is hand-authored and labeled synthetic. Its metrics and targets are illustrative, not benchmarks, and each one includes at least one genuine unresolved measurement issue.
B2C product growth (activation + retention)
A synthetic B2C activation and retention set that keeps a planning-judgment target separate from a benchmark and shows a partially mature cohort.
Synthetic, not a benchmark.
B2B SaaS (procurement readiness)
A synthetic B2B procurement set with a CRM-sourced time metric, a committed target and a still-undecided target.
Synthetic, not a benchmark.
Internal reliability (MTTR + incidents)
A synthetic internal reliability set that proves targets are not all percentages: MTTR in hours, P1 count, and an SLA guardrail.
Synthetic, not a benchmark.
Pro tips
- Write the Objective as the outcome you intend to pursue. Put every number in a Key Result, not in the Objective.
- Treat each Key Result as a measurement contract: metric, definition, population, baseline, target, target basis, timeframe, owner, source and refresh cadence.
- Never let a target exist without a recorded basis. Trend, SLA, forecast, benchmark or team judgment are all valid, a silent number is not.
- Only call a target a benchmark when you can name its source and population. A number you chose is a planning judgment, not a benchmark.
- Keep an unknown baseline or target visible. 'Needs a decision' is a real state, and pretending otherwise creates fake precision.
- Separate the metric refresh cadence from the OKR check-in cadence. Revenue may update monthly while you meet every two weeks.
- Do not auto-classify leading and lagging. Ask whether you actually need an earlier signal to decide before the outcome matures.
- Ask 'if every Key Result were achieved, would we be confident the Objective progressed?' rather than hunting for a specific KR count.
- Treat initiatives as hypotheses for moving Key Results. Completing an initiative never advances a result by itself.
- Record target and definition changes with a reason and a comparability note so reviews cannot quietly move the goalposts.
Common OKR mistakes
Targets appear from nowhere.
Why: The tool or the team fills in a number before anyone decided what success means.
Do instead: Set the target from a recorded basis (trend, prior period, SLA, forecast, benchmark with source, budget, or explicit judgment). Leave it unknown until then.
Every target is a percentage.
Why: Treating OKRs as a percentages-only format.
Do instead: Match the unit to the metric: hours for MTTR, days for cycle time, counts for incidents, percentages only for true ratios with a defined denominator.
The Objective is a task list.
Why: Writing work items ('launch', 'build', 'migrate') as the Objective.
Do instead: State the outcome and move the work into initiatives. A verifiable milestone Key Result is fine when it is explicitly typed and dated.
One number hides the denominator.
Why: 'Activation 35%' without saying which population or window it is measured over.
Do instead: Record the population/denominator and baseline period so two people cannot calculate different numbers.
Check-ins become a status ritual.
Why: Recording progress without a blocker, learning or decision.
Do instead: Each check-in captures the current value, data freshness, what the team learned and what changes next, including 'no change yet' as a decision.
Initiative completion is reported as result progress.
Why: Marking work done and treating that as the Key Result moving.
Do instead: Update the Key Result from its source data only. Initiative status is separate evidence that a hypothesis was tested.
Leading/lagging is assigned by habit.
Why: Forcing one leading indicator onto every Objective regardless of causal timing.
Do instead: Classify only when the label is meaningful, or leave it unknown. Add an earlier signal only when it helps you decide sooner.
A target changes mid-cycle with no trace.
Why: Editing the number when the result is behind.
Do instead: Add a change note with old value, new value, reason and whether history stays comparable.
OKR Generator FAQ
Is this for business and team OKRs or HR performance reviews?
This is an OKR drafting and operating tool for team and company goal-setting, not HR performance review or compensation software. It helps you define Objectives, write Key Results as measurement contracts, keep initiatives separate, and run lightweight check-ins. It does not rate individuals.
Does the tool invent targets or objectives for me?
No. A blank workspace stays blank. When you ask it to draft structure, it only uses the strategic intent, focus areas and metrics you supplied, and targets stay unknown until you decide them. CraftUp may help you structure a target, but it never invents the target.
What is a Key Result measurement contract?
It is the full description of what a Key Result measures: metric name and definition, population or denominator, unit, data source, baseline and its period, target, target basis, timeframe, owner and refresh cadence. The contract is what makes two people calculate the same number and understand why the target exists.
What if I do not have a baseline yet?
Leave the baseline blank to mark it as not assessed, or write 'unknown' to mark it explicitly unknown. These are different states. The audit turns an unknown baseline into a decision, establish it, instrument it, or set the target from a contractual source that does not need one, instead of blaming the draft.
Can a target stay unknown?
Yes. A draft with a clearly defined metric and a baseline but no target is valid. The audit reports 'Target decision missing' rather than 'invalid OKR', and the next step is to agree the target basis. That is usually better than a fabricated number.
Are tasks ever allowed as Key Results?
Outcome-oriented Key Results are the strong default, but practice varies. A clearly typed, verifiable milestone with a date can be legitimate in some operating contexts. The tool flags likely task wording for review instead of rejecting it automatically, and never claims all milestone Key Results are invalid.
Do I need a leading indicator?
Not always. Leading measures help only when they are plausibly predictive and the team can act on them before the outcome matures. If the Objective is best measured by mature outcome metrics, that is allowed; the tool asks whether an earlier signal would actually help you decide, rather than forcing one.
How many Key Results should an Objective have?
Aim for a small number of decision-useful Key Results. There is no universally correct count, some Objectives need one or two, others three or four. The tool warns only when quantity clearly undermines focus, and explains that as a heuristic rather than a correctness rule.
How do check-ins work, and how often should I run them?
You choose the OKR check-in cadence, and each metric keeps its own refresh cadence. A check-in records the current value, data window, change, blocker, learning and the decision that follows. The ledger keeps history, so a blocker without a decision or a silent target change becomes visible.
Is my OKR data private, and where does it go?
Everything runs in your browser and is saved to local storage only. Nothing is sent to CraftUp servers, and analytics carry only structural fields such as counts and selected options, never your Objective, metric, target or notes. Share links encode the workspace in the URL, so review confidential content before sharing.
After you draft the OKRs
If the measurement structure is unclear, work out the metric and its drivers first. If the OKRs are sound and the investments need sequencing, take them to the roadmap. If one initiative is really an experiment, design it with the experiment owners instead of calling the work a Key Result.
Learn the reasoning behind the tool
This tool owns the Objective set and the measurement contracts. The courses, guides and glossary entries below deepen the underlying discovery, metrics and roadmap thinking.
Keep OKRs moving between reviews
Use CraftUp to turn measurement contracts and check-in decisions into clear next steps, without turning goal-setting into performance management.
Last updated: 2026-09-24
- 2026-09-24: Rebuilt as an evidence-safe OKR builder: blank-first workspace, KR Measurement Contracts and no fabricated targets.
- 2026-09-24: Added the OKR Quality & Alignment Audit with grouped findings and no aggregate score.
- 2026-09-24: Added the Check-In Learning Ledger, target/KR change history, and v1 workspace migration with review flags.
- 2026-09-24: Added v2 normalized storage/share, share-size and privacy handling, and structural-only analytics.