# How the success probability is calculated

> The estimate comes from deterministic rules over ENISA's published requirements. It is never 0 or 100, a blocking rule caps it at 18, and the AI never decides.

- Canonical: https://help.enisa.ai/en/el-analisis/como-se-calcula-la-probabilidad
- Updated: 2026-08-08
- Language: en-US
- Versión en español: https://help.enisa.ai/el-analisis/como-se-calcula-la-probabilidad.md

The enisa.ai success estimate is calculated with deterministic rules over the requirements ENISA publishes: it is never 0 or 100, and if a blocking requirement fails it is capped at 18 at most — the AI does not decide the number. By the end you will know where it comes from, why we call it an estimate, and what can move it up or down.

## Why we say "estimate" and not "probability"

A true probability requires real outcome data behind it. Our number comes from a rules-based methodology, not from a history of approvals — so we present it as what it is: an **indicative estimate** out of 100, always accompanied by the reasons that explain it. It is never a promise, and the final decision is always ENISA's.

## Where the number comes from

1. **First, the rules.** Every requirement ENISA publishes is encoded as a deterministic rule, with its **verbatim quote** from the official source. Your report shows the call, the version of the rules and the date they were verified against enisa.es. Same inputs, same result, every time.

2. **Then, the adjustments.** On that base, factors such as traction, team and degree of innovation move the estimate — each within bounds, so that no soft factor ever outweighs a requirement.

3. **And limits that are code, not judgement.** The estimate is **never 0 and never 100**: there is always uncertainty in both directions. And if a blocking requirement fails, it is capped at **18 at most** — however good the rest of the story, a failed requirement rules.

## The verdict

- **Apto (eligible)** — you meet what can be checked and the project fits.
- **Apto con condiciones (eligible with conditions)** — it fits, with specific caveats the report details.
- **Todavía no apto (not eligible yet)** — what fails is a fixable practitioner signal; the report says what to change. It is not a final no.
- **No apto (not eligible)** — a requirement published by ENISA fails. There is no viable file today, and it is better to learn that for €3 than after weeks of work.

If information was missing to check some requirement when the verdict was calculated, the report marks it as a **provisional verdict**.

## What the AI does (and does not do)

The AI **writes the explanation**: the summary, the strengths, the risks and the next steps, from the already-decided result. **It never decides the verdict or the estimate.** That is deliberate: rules can be tested and audited; a model asked to "check the requirements" cannot.

## Troubleshooting

### My estimate is high — will I get the loan?

We do not know, and nobody honest can assure you of it. A high estimate means that, against the published requirements and what you told us, your file starts from a good place.

### My estimate is 18 or lower — should I give up?

First look at which requirement fails and what class it is. If it is a fixable practitioner signal, the report tells you what to change — and once changed, a new analysis can give a very different result.