Skip to content

rampstackco/claude-skills / skills/okr-design

okr-design

Verified

OKR design as actually shipped, not as conference-talk theory. Outcome statements that drive decisions, key results that measure the right thing, scoring discipline, mid-quarter recalibration, and the difference between sandbagged OKRs (always 100%) and aspirational OKRs (always 30%) and stretch OKRs (genuine ambition with quarterly accountability). Triggers on OKR design, OKR setting, key result design, OKR scoring, mid-quarter recalibration, OKR cascading, outcomes vs outputs, quarterly planning, goal setting. Also triggers when a team's OKRs are always hit and producing no learning, when OKRs are demoralizing because they were set as fantasy, or when the team uses OKR vocabulary but the practice has decayed.

references/scoring-discipline.md

View on GitHub
# Scoring discipline

The 0.0-1.0 scale. The 60-70% target. Scoring honesty. What 100% and 30% mean. The compensation question.

End-of-quarter scoring is where OKR practice succeeds or decays. Honest scoring produces signal: which OKRs the team hit, which they missed, what the gap reveals about the work. Dishonest scoring produces theater: 100% scores claimed regardless of outcome, generous rounding, partial credit for effort rather than outcome.

---

## The 0.0-1.0 scale

Each key result scores from 0.0 (no progress) to 1.0 (fully achieved). The objective scores as the average of its key results.

**Key result scoring examples.**

- KR: "Increase first-week activation rate from 32% to 45%."
- Outcome: rate reached 41%.
- Score: (41 - 32) / (45 - 32) = 9/13 = 0.69.

- KR: "Reduce median time-to-first-value from 4.2 days to under 2 days."
- Outcome: reduced to 2.6 days.
- Score: (4.2 - 2.6) / (4.2 - 2.0) = 1.6/2.2 = 0.73.

- KR: "Pass SOC 2 Type II audit with zero high-severity findings."
- Outcome: passed with zero high-severity findings.
- Score: 1.0 (binary KR; achieved).

**Objective scoring.** Average of the key result scores.

- Objective with 3 KRs scoring 0.69, 0.73, 1.0: objective scores (0.69 + 0.73 + 1.0) / 3 = 0.81.

---

## The 60-70% target

Stretch OKRs are designed so that the average score across the team's OKRs is 0.6-0.7.

**Why this target.**

- Average above 0.85 suggests sandbagging. The team set OKRs they were already on track to deliver.
- Average below 0.4 suggests fantasy. The team set OKRs that no realistic effort path would have produced.
- Average between 0.6-0.7 suggests stretch. The team committed to ambition and delivered most of it.

**The trajectory check.** Track the team's average score over multiple quarters. Drift toward 0.85+ signals sandbagging is creeping in; drift toward 0.4 signals fantasy. The target zone is 0.55-0.75 over a multi-quarter window.

**Variance is normal.** Some quarters will hit 0.8 average; some will hit 0.5. The variance reflects the genuine uncertainty in stretch OKRs. Programs that hit 0.65 every single quarter are usually managing scores rather than honestly scoring outcomes.

---

## Scoring honesty

**Score the actual outcome, not the effort.** A team that worked hard but did not move the metric scores low. A team that got lucky and moved the metric without much effort scores high. The score reflects outcome.

**Do not round up.** A KR at 0.55 scores 0.55, not 0.6. Honest scoring produces honest signal. Rounding up across multiple KRs accumulates into systemic over-scoring.

**Acknowledge what changed.** Some misses reflect mid-quarter priority shifts. Surface these without using them as excuses. "We deprioritized this in week 5 to address the security incident" is honest context; "we hit our priorities even though we missed this OKR" is hand-waving.

**Scoring sessions are honest, not adversarial.** The scoring conversation is for capturing reality, not for blame. Honest low scores are fine; effort recognition happens elsewhere.

---

## What 100% scores mean

100% scores warrant scrutiny.

**Possibilities.**

- The OKR was sandbagged. The team set targets they were already on track to hit.
- The team had a great quarter. Sometimes hitting 100% reflects unusual progress; informative.
- The team is rounding up. The actual outcome was 0.85 but got rounded.
- Binary KRs that hit 1.0. A milestone-based KR that achieves the milestone is 1.0; this is not sandbagging.

**The investigation.**

- Compare the target to the historical trajectory. Was the target above what the metric would have done without focused effort?
- Compare the target to other teams setting similar OKRs. Was this target genuinely stretch?
- Look at the work the team did. Did the team push beyond their normal capacity, or did they hit the target on standard execution?

**The interpretation.** Single-quarter 100% scores are not necessarily problematic. Repeated 100% scores across many KRs across multiple quarters are a sandbagging signal. The next OKR cycle should set targets above the prior quarter's actuals to prevent sandbagging compounding.

---

## What 30% scores mean

30% scores warrant scrutiny.

**Possibilities.**

- The OKR was fantasy. Targets had no realistic effort path; the team could not have hit them.
- Unexpected disruption. A market event, security incident, or priority shift consumed capacity that would have gone to the OKR.
- Mid-quarter deprioritization. The team chose to deprioritize this OKR for other work.
- Underperformance. The team did not execute well; the score reflects the execution gap.

**The investigation.**

- Was the target reachable? At setting time, could the team identify effort paths that would produce the target?
- What changed during the quarter that affected the trajectory?
- What did the team actually work on? Did effort match the OKR, or did effort go elsewhere?

**The interpretation.** 30% scores are informative regardless of cause. Fantasy targets need recalibration in the next quarter. Disruption-driven misses inform what to plan around. Deprioritization-driven misses surface tension between OKRs and other work. Execution gaps surface team-capability or process issues.

**The non-recovery.** Repeated 30% scores quarter after quarter signal that OKR design is failing. Either the org is consistently setting fantasy targets, or the team is consistently mismatched to its commitments. Investigate.

---

## The compensation question

OKRs work best when not directly tied to compensation.

**The problem with compensation-coupling.** When OKRs determine bonuses or reviews, sandbagging incentives become severe. Teams set OKRs they know they can hit. The 60-70% target collapses to 95% target because nobody wants to risk their bonus on stretch.

**Why decoupling works.** OKRs become genuine accountability infrastructure. The team can stretch without risking compensation. Honest scoring is safe. Misses produce learning rather than punishment.

**Compensation alternatives.**

- Compensation tied to overall performance (manager assessment, peer feedback, role expectations) rather than to OKR scores specifically.
- OKR scores inform performance review as one input among many, not as the dominant signal.
- Bonus structures based on company performance rather than team-OKR scores.

**The transition for orgs already coupled.** Decoupling requires explicit communication. Teams need to trust that scoring honestly will not affect their bonuses. The transition can take 2-3 cycles before teams believe the decoupling is real.

---

## Scoring formats

How scoring sessions get run.

**The format.**

- Team meets at end of quarter (60-90 minutes).
- Each KR scored honestly.
- Each objective scored as the KR average.
- Discussion: what produced the scores, what worked, what did not.
- Output: published scores with brief context per OKR.

**The discussion content.**

- For high-scoring OKRs: was this stretch? What worked? What can be replicated?
- For low-scoring OKRs: what got in the way? What does the team need next quarter?
- For mid-scoring OKRs: where did the work land? What was the gap between target and actual?

**The output.** Scores published to the team and stakeholders. Brief context per OKR. The published record informs the next quarter's setting and the org's broader OKR practice.

---

## Scoring across multiple teams

When multiple teams cascade to a company OKR, scoring composes.

**The composition.**

- Each team scores its KRs independently.
- The company OKR's score reflects the cumulative outcome (not the average of team scores).
- A company OKR can score high even if some contributing team OKRs scored low, depending on which contributions mattered most.

**Worked example.** Company OKR: "Improve net new ARR through enterprise segment growth."

- Sales team KR (close 8 enterprise deals): scored 0.85 (closed 7).
- Product team KR (enterprise-readiness foundations): scored 0.7 (achieved most foundations; SOC 2 in progress).
- Marketing team KR (enterprise-segment marketing campaigns): scored 0.5 (delayed launch).

Company OKR score: depends on the actual ARR outcome, not the average of team scores. If ARR met the target despite the marketing delay (because sales over-delivered), the company OKR scores high.

**The discipline.** Score outcomes, not averages of inputs.

---

## Common scoring failures

**Rounding up.** Systemic over-scoring through small individual rounds.

**Effort-based scoring.** Scoring based on how hard the team worked rather than what was achieved.

**Compensation-coupled scoring.** Sandbagging incentives produce 95%+ scores.

**Avoidance of low scores.** Teams reluctant to score low; OKRs renamed or rewritten retroactively to avoid the miss.

**Aggregate-of-input scoring.** Company OKR scored as average of team scores rather than as the actual outcome.

**Scoring without context.** Scores published without the conditions that produced them. Lessons get lost.

**One-quarter-at-a-time view.** Scoring evaluated only against the current quarter; trajectory over multiple quarters not surfaced.

---

## Methodology-level choices that stay in the public skill

The 0.0-1.0 scale and worked examples. The 60-70% target. Scoring honesty. What 100% and 30% scores mean. The compensation question. Scoring formats. Scoring across teams. Common scoring failures.

## Implementation choices that stay internal

Specific scoring tools and dashboards. Specific scoring-session facilitation formats. Specific compensation structures. The team's own conventions for context publishing. These vary by org.
brainstormingobra/superpowersYou MUST use this before any creative work - creating features, building components, adding functionality, modifying behavior, or planning anything new, in software or out of it (a talk, a business, a renovation).Stars: 322kUpdated Claude CodeCodexVerifiedInstall
brand-guidelinesanthropics/skillsApplies Anthropic's official brand colors and typography to any sort of artifact that may benefit from having Anthropic's look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
frontend-designanthropics/skillsGuidance for distinctive, intentional visual design when building new UI or reshaping an existing one. Helps with aesthetic direction, typography, and making choices that don't read as templated defaults.Stars: 198kUpdated Claude CodeCodexVerifiedInstall
mcp-builderanthropics/skillsGuide for creating high-quality MCP (Model Context Protocol) servers that enable LLMs to interact with external services through well-designed tools. Use when building MCP servers to integrate external APIs or services, whether in Python (FastMCP) or Node/TypeScript (MCP SDK).Stars: 198kUpdated Claude CodeCodexVerifiedInstall