Documentation ยท Tracking & Analysis
Building Your Learnings Library
Transform individual experiment results into a searchable knowledge base that compounds over time and informs future experiment design.
Why a Learnings Library?
- Prevents teams from repeating failed experiments
- Builds institutional knowledge that survives team changes
- Enables pattern recognition across experiments
- Informs hypothesis development with evidence
- Powers the Evidence-First ideation mode and Smart ICE Editor learning calibration
Structure of a Learning
Each learning captures three key elements:
What Happened - The factual outcome of the experiment What It Means - Your interpretation and business implications What to Try Next - Actionable next steps based on the insight
Two ways to capture a learning
- From a completed experiment - Mark a Running experiment as Done on the board and record its outcome. The learning is saved against that experiment.
- Log a past learning - Click "Log a learning" in the Learnings header (or from the empty state). Capture a result from an experiment you ran anywhere, even before GrowthLab. It joins your library immediately, so the compounding loop starts in your first session instead of weeks later.
Either way, the learning is indexed and becomes citable evidence in future batches.
AI Learning Synthesis
Our AI can synthesize patterns across multiple learnings to surface:
- Recurring themes and meta-insights across experiments
- Contradictory findings that need resolution
- Validated principles you can rely on
- Knowledge gaps that need testing
The synthesis engine also powers the Related Learnings feature in the Evidence Panel, automatically connecting past insights to new experiment designs.
Tagging and Organization
Use tags to categorize learnings by:
- Funnel stage - Acquisition, Activation, Retention, Revenue, Referral
- Experiment type - Copy, UX, Pricing, Feature
- Audience segment - New users, Power users, Enterprise
- Impact direction - Positive, Negative, Neutral
- Custom tags - Team-specific categories
How Learnings Feed Back into the System
| Feature | How It Uses Learnings |
|---|---|
| Evidence-First Ideation | Cites relevant past learnings as evidence sources |
| Smart ICE Editor | Adjusts confidence scores based on past experiment success rates |
| Evidence Panel | Displays related learnings with outcome badges |
| AI Learning Synthesis | Identifies cross-experiment patterns |
Best Practices
- Write learnings immediately after analyzing results
- Focus on "so what?" not just "what happened"
- Tag consistently for searchability
- Review monthly for emerging patterns
- Use AI synthesis to identify themes you might miss
