Smart ICE Editor

Get data-driven, calibrated ICE scores that help you prioritize experiments objectively and avoid common biases.

The Prioritization Challenge

Teams frequently fall victim to cognitive biases that distort prioritization:

• Overconfidence bias - Assuming your ideas will definitely work
• Effort underestimation - Forgetting about hidden complexity
• Impact inflation - Overestimating potential gains without data
• HiPPO - Deferring to the highest paid person's opinion rather than evidence

The Smart ICE Editor helps you score more objectively by using data, benchmarks, and historical patterns.

How the Smart ICE Editor Works

The editor combines manual sliders (1–10 per dimension) with AI suggestions:

1. Open any experiment's ICE score section
2. Use sliders to set your initial scores manually
3. Click "Get AI Suggestions" for calibrated recommendations
4. Review the per-dimension reasoning provided by the AI
5. Click "Apply All" to accept all suggestions, or "Use X" per dimension to apply individually
6. Adjust based on your domain knowledge

What the AI Considers

Your Business Context

Industry, company stage, available resources, and current growth metrics from Company Settings.

Historical Experiment Data

Which experiments produced lifts, how accurate past estimates were, and success/failure patterns from your team's history.

Industry Benchmarks

Typical conversion improvements, implementation timelines, and success rates for similar experiment types.

Complexity Signal Analysis

Keywords and patterns in your hypothesis, experiment map, and description that indicate implementation difficulty.

Learning Calibration

💡 Tip: This is what makes the Smart ICE Editor unique - it learns from your past experiments.

When the AI detects relevant past learnings, each score dimension may show a learning influence panel indicating:

• Modifier - How much the score was adjusted (e.g., +2 or -1)
• Source learning - The specific past experiment insight that informed the adjustment
• Success rate - How often similar experiments succeeded

A Learning Calibration Summary appears at the top when historical data is available, explaining the overall influence of past experiments on the suggested scores.

Priority Tier

After analysis, the AI assigns a priority tier:

All documentation