Methodology

1. How an analysis works

Licenscope opens the public pages of a website in an isolated browser, follows links within the same domain and inventories the resources loaded: fonts, scripts, stylesheets, images, SVG, videos and documents.

Each resource is then matched against known sources: font file metadata, library signatures and provider hosts. Licence rules are applied to these elements and a status is proposed.

2. What is reachable and what is not

Only publicly observable resources are analysed. Pages behind a login, a paywall or a CAPTCHA are not crawled, and no protection is bypassed.

No form is submitted, no account is used and no video is downloaded. The site's robots.txt is respected.

3. Crawl limits

Each analysis is bounded: number of pages (5 in the free discovery scan, up to 200 depending on the plan), depth, number of requests, response size and duration. When a limit is reached, the analysis states it in its scope and in the report.

An analysis that reaches the plan's page cap can therefore be stopped by the maximum duration. Pages not visited are then not analysed, and the report says so.

4. Identified, licence known, rights held: three distinct notions

A resource is identified when Licenscope recognises its component or foundry with sufficient confidence. This says nothing about the licence that applies to your website.

A licence is known when the reference records its version and general terms. The terms of a version, a product or a contract may differ from the standard licence.

Rights are held only when you have a contract, an invoice or a licence, and a supporting document has been reviewed. Licenscope cannot know whether you bought a licence.

5. Commercial components

A component whose name suggests a commercial offer is not classified as such on the name alone: its terms may depend on the version, product or contract. Licenscope then reports it as 'to verify' and points to the official terms.

A commercial licence without a supporting document is never presented as infringement.

6. Cases that cannot be analysed

Dependencies bundled into a combined file are only visible when they expose a banner or a global object. Image rights are not identified; only apparent provenance is shown. Renamed fonts or fonts without metadata remain 'to identify'. The site's source code, password-protected files and content behind an account are not analysed.

7. Confidence levels

The percentage shown is a technical confidence in the identification, from 0 to 100%. It measures neither the validity of a licence nor the compliance of a use. An identification at 60% is a lead, not a certainty.

8. Rules for classifying alerts

Each finding cites the rule, its version and its documentary source. Statuses are: detected, to identify, to verify, confirmed anomaly, documented compliance, remediated, not applicable.

Only a human decision, taken with a reviewed document, can move a resource to 'documented compliance'. Rules produce at most one automatic status. No rule alone establishes a legal obligation.

The absence of an alert only means that no rule was triggered on the resources observed. It does not guarantee that there is no licence issue.

9. Changes and corrections to the reference

The reference is versioned: a modified rule gets a new version number and the previous one remains available. Earlier reports keep the version that produced them.

A reported qualification error is checked against the official source, corrected if needed, then covered by a test. Reports are made from the resource sheet or from the help centre.