# Results, probabilities and confidence

Source: https://iachecker.com/docs/results-and-confidence
Status: Available now
Last verified: 2026-08-08

The result page is designed as an evidence hierarchy: verdict first, probability second, contributing signals third. Confidence describes evidence depth rather than certainty about authorship.

## Result anatomy

*What each result element means*

| Element | Meaning |
| --- | --- |
| Verdict | Plain-language direction based on the combined reading |
| AI probability | Normalized strength of AI-leaning signals in this analysis |
| Confidence | How much suitable evidence was available |
| Signal pills | Named checks and their measured contribution |
| Details | Model, report identifier, format, timing and technical context |

## Compare like with like

Scores from a 20-word excerpt and a 1,000-word article do not have the same evidence depth. Likewise, an original PNG with metadata and a recompressed screenshot offer different source signals.

- Use similar lengths when comparing text samples.
- Keep image format and resolution comparable.
- Record whether a sample was copied, translated, resized or recompressed.
- Use the same detector version when tracking changes over time.

## Turn a result into a review

A useful workflow connects the signal to the underlying material. For text, inspect the highlighted passages and source history. For images, compare provenance, visible labels, model readings and file-level evidence.

> **Document the context**
>
> When a decision matters, save the source, detector version, date and review notes alongside the report instead of keeping only a screenshot of the percentage.

## Documentation index

- [IA Checker documentation](https://iachecker.com/docs)
- [Assistant index](https://iachecker.com/llms.txt)
- [Privacy Policy](https://iachecker.com/privacy)