Skip to content

borghei/Claude-Skills / project-management/execution/brainstorm-okrs

brainstorm-okrs

Verified

Draft and score quarterly OKRs: a qualitative objective, measurable key results and a counter-metric. Use when setting or reviewing a team's OKRs before they are committed, or when KRs read like a task list.

references/okr-workflow-and-examples.md

View on GitHub
# OKR Workflow, Examples & Tooling

> Read this when you need the full step-by-step OKR generation workflow, a worked quarterly example, the mistakes/KPI reference tables, troubleshooting, success criteria, or the `okr_validator.py` flag reference.

## Workflow

### 1. Identify the Theme

The agent asks: "What is the single most important thing this team needs to change this quarter?" The answer becomes the theme. Every OKR must connect back to this theme.

**Validation checkpoint:** If the user provides more than one theme, the agent pushes back. One theme per team per quarter. Multiple themes means no focus.

### 2. Generate 3 Distinct OKR Sets

For each set, the agent produces:

1. **Objective** -- One qualitative, inspirational statement (no numbers)
2. **Key Result 1** -- Primary metric proving progress
3. **Key Result 2** -- Secondary metric capturing a different dimension
4. **Key Result 3** -- Counter-metric preventing gaming of KR1 and KR2
5. **Rationale** -- 2-3 sentences on why this set matters and how it connects to the theme

**Objective quality criteria:**
- Qualitative (numbers belong in key results)
- Inspirational (team would be excited to achieve it)
- Time-bound (achievable within one quarter)
- Actionable (team can directly influence the outcome)

**Key result quality criteria:**
- Measurable (has a metric with a number)
- Outcome-focused (measures results, not activities)
- Set at 60-70% confidence (not sandbagging, not demoralizing)
- Limited to 3 per objective

### 3. Apply the Counter-Metric Test

For every pair of key results, the agent asks: "Could we hit these numbers by doing something harmful?" If yes, it adds a counter-metric.

Example: If KR1 is "Increase sign-ups by 40%", a counter-metric is "Maintain activation rate above 60%." Without it, the team could game KR1 by lowering sign-up barriers so far that unqualified users flood in.

### 4. Validate with Tool

```bash
python scripts/okr_validator.py --input okrs.json
```

The validator scores each OKR set and surfaces quality issues: disguised tasks, missing metrics, output-framed key results, or missing counter-metrics.

**Validation checkpoint:** Any OKR set scoring below 80 (`okr_validator.py` exits 1) must be revised before committing.

## Example: Quarterly OKR Generation

**Input:** Theme is "retention" for a SaaS product team.

**Output:**

```
OKR Set 1:
  Objective: "Become the product teams can't imagine leaving"
  KR1: Reduce monthly churn from 4.2% to 2.5%
  KR2: Increase 90-day retention cohort from 68% to 82%
  KR3 (counter): Maintain NPS score above 45 (prevent forced lock-in tactics)
  Rationale: Churn is the top revenue leak. Improving retention directly
  increases LTV and reduces pressure on acquisition spend.

OKR Set 2:
  Objective: "Make our onboarding so good that users hit value in their first session"
  KR1: Increase Day-1 activation rate from 34% to 55%
  KR2: Reduce time-to-first-value from 12 minutes to under 4 minutes
  KR3 (counter): Maintain support ticket volume below 200/week (don't hide complexity)
  Rationale: Users who activate on Day 1 retain at 3x the rate. Onboarding
  is the highest-leverage retention lever.

OKR Set 3:
  Objective: "Turn our power users into vocal advocates"
  KR1: Increase referral-sourced signups from 8% to 20% of new users
  KR2: Grow active community members from 500 to 2,000
  KR3 (counter): Maintain power user retention above 95% (don't distract them)
  Rationale: Advocacy compounds. Referred users have 37% higher retention
  than paid-acquisition users.
```

```bash
$ python scripts/okr_validator.py --input okrs.json

OKR Validation Results
======================
Set 1: 92/100 - PASS
  Objective: Qualitative, inspirational, time-bound
  KR1: Measurable, outcome-focused, stretch target
  KR2: Measurable, different dimension from KR1
  KR3: Valid counter-metric for churn reduction

Set 2: 88/100 - PASS
  Objective: Qualitative, inspirational, time-bound
  KR1: Measurable, outcome-focused
  KR2: Measurable, tracks different dimension
  KR3: Valid counter-metric
  Note: "under 4 minutes" - verify baseline measurement exists

Set 3: 85/100 - PASS
  Objective: Qualitative, inspirational
  KR1: Measurable, outcome-focused
  KR2: Measurable, but "active" needs precise definition
  KR3: Valid counter-metric
```

## Common OKR Mistakes

| Mistake | Example | Fix |
|---------|---------|-----|
| Disguised task | "Launch the mobile app" | Ask "why?" -- measure the outcome the launch enables |
| Too many OKRs | 5 objectives per team | Pick 1, maybe 2. More means no focus |
| 100% confidence | Target you know you will hit | Stretch to 60-70% confidence |
| Activity metric | "Publish 12 blog posts" | Measure impact: "Increase organic traffic by 30%" |
| Set and forget | Review only at quarter end | Weekly check-ins with confidence scoring |
| Top-down only | All OKRs from leadership | Combine top-down direction with bottom-up team insight |

## OKRs vs KPIs vs North Star Metric

| Concept | Purpose | Cadence | Example |
|---------|---------|---------|---------|
| North Star Metric | Single metric capturing core value delivery | Permanent | Weekly active users completing a workflow |
| KPIs | Health indicators across the business | Ongoing | Revenue, churn rate, response time |
| OKRs | Ambitious quarterly goals that move KPIs | Quarterly | "Become the fastest onboarding in our category" |

**Relationship:** OKRs are the lever pulled to move KPIs toward the North Star Metric. KPIs indicate business health. The NSM indicates core value delivery. OKRs define what changes this quarter.

## Tools

| Tool | Purpose | Command |
|------|---------|---------|
| `okr_validator.py` | Validate and score OKR sets | `python scripts/okr_validator.py --input okrs.json` |
| `okr_validator.py` | Run demo validation | `python scripts/okr_validator.py --demo` |

## Troubleshooting

| Symptom | Likely Cause | Resolution |
|---------|-------------|------------|
| OKR set scores below 80 consistently | Key results framed as tasks/outputs instead of outcomes, or objective contains numbers | Ask "So what?" for each KR until you reach a measurable outcome; remove numbers from objectives |
| Validator flags "output-oriented language" | KR description starts with verbs like "launch", "build", "implement", "ship" | Reframe: "Launch mobile app" becomes "Increase mobile-originated revenue from 0% to 15%" |
| Team sets 5+ objectives per quarter | Lack of strategic focus or inability to say no | Enforce 1 theme per team per quarter; use the Radical Focus constraint: one objective, maybe two |
| Key results hit 100% every quarter | Targets are sandbagged at 100% confidence | Stretch to 60-70% confidence; if you hit every KR, you are not being ambitious enough |
| Counter-metrics missing from OKR sets | Team did not apply the gaming test to KR pairs | For every pair of KRs, ask: "Could we hit these numbers by doing something harmful?" Add a counter-metric if yes |
| OKRs set and forgotten until quarter end | No weekly check-in rhythm established | Implement weekly confidence scoring (red/yellow/green) per KR; teams with weekly check-ins complete 43% more goals |
| Validator rejects input JSON | Schema mismatch: missing `okr_sets` key or `key_results` array per set | Ensure JSON has `okr_sets` array, each with `objective` string and `key_results` array containing `description`, `metric`, `target_value`, `current_value` |

## Success Criteria

- Each OKR set scores above 80/100 on the validator before committing to the quarter
- Maximum 1-2 objectives per team per quarter (focus over breadth)
- Every objective is qualitative and inspirational (no numbers in the objective itself)
- Each objective has exactly 3 key results: primary metric, secondary dimension, and counter-metric
- Key results are set at 60-70% confidence (stretch, not sandbagged)
- Weekly confidence check-ins are conducted, not just end-of-quarter reviews
- OKR retrospectives run at quarter end with structured review of what was learned

## Tool Reference

### okr_validator.py

Validates and scores OKR sets against quality criteria. Checks objectives for qualitative/inspirational language, key results for measurable outcomes, and structural completeness.

| Flag | Type | Default | Description |
|------|------|---------|-------------|
| `--input` | string | (required, mutually exclusive with --demo) | Path to JSON file containing OKR sets |
| `--demo` | flag | off | Run validation on built-in demo data (mix of good and bad OKRs) |
| `--format` | choice | `text` | Output format: `text` or `json` |

**Input JSON schema:**
```json
{
  "okr_sets": [
    {
      "objective": "string (qualitative, no numbers)",
      "key_results": [
        {
          "description": "string",
          "metric": "string (unit of measurement)",
          "target_value": "number",
          "current_value": "number (baseline)"
        }
      ]
    }
  ]
}
```
receiving-code-reviewobra/superpowersUse when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementationStars: 322kUpdated Claude CodeCodexVerifiedInstall
requesting-code-reviewobra/superpowersUse when completing tasks, implementing major features, or before merging to verify work meets requirementsStars: 322kUpdated Claude CodeCodexVerifiedInstall
using-git-worktreesobra/superpowersUse when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallbackStars: 322kUpdated Claude CodeCodexVerifiedInstall
verification-before-completionobra/superpowersUse when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions alwaysStars: 322kUpdated Claude CodeCodexVerifiedInstall