Three reasons:

1. LLM variance. Each analysis runs multiple parallel AI calls and takes the median result to improve accuracy. But there is always some natural variation between runs. A difference of 2-5 points is normal and expected.

2. Live data changes. Your score pulls real-time signals from 50+ sources including Google Trends, Reddit, and competitor databases. These change daily. A re-scan reflects current market conditions.

3. Your description changed. Even small wording changes affect how the analysis interprets your idea, target market, and competition.

A small score difference between scans is not a bug. If you see a large swing (10+ points) between two scans of the same tier, contact us at support@preuve.ai and we will investigate.

The free Reality Check is an estimate. It is cross-checked across multiple AI runs, but it scores your idea at global-market level in under a minute.

The deep analysis re-scores your idea from scratch with 10 AI agents across 50+ live sources, scoped to the exact market you selected. It prices in evidence a quick pass cannot see: funded competitors, the real size of your regional market, and unit-economics benchmarks.

Because it is more precise, the verified score can land higher or lower than your free estimate. A lower score is not a penalty. It means the deep analysis found risks worth knowing before you build, and the report's risk and competitor sections show exactly what it found.

Region matters too. Free scans run in Global mode, while deep analyses can target a specific country. A market that looks crowded globally can be open in your region, and vice versa, so the geo selection alone can move the score meaningfully.

The free Refine option is available on paid deep analysis reports. It re-runs the logic on your existing data without pulling fresh competitor or community data, so any score change reflects your positioning edits only.