How the Sovereign Compass works
Describe your need
Ten questions on data class, regulation, location, access, operations, AI, exit, evidence standard and sustainability. You can adjust the weighting of the eight objectives.
The evidence engine measures
Your answers become a target evidence level per criterion plus knock-out criteria. Every provider is measured against it with evidence that has been validated against its source.
AI summarises
An AI writes the summary based on the finished result. Every number in the text is checked against the result; otherwise the template text applies.
Result by email
As soon as the run is finished, the link to your result arrives by email: top 3 with fit, gaps and evidence questions, plus the list of all sources.
You see the three best-fitting providers in full, all others together with the free expert check.
Methodology in brief
Evidence level 0–4
0 = not given or refuted · 1 = claimed only · 2 = partially evidenced · 3 = evidenced (documentation, contract, certificate scope) · 4 = fully evidenced and externally audited or technically enforced. We rate the quality of the evidence, not the quality of the promise.
Source rule
Every cell has exactly one public source – provider documentation, register, certifier or European Commission – and a check date. What only the provider claims is marked “provider statement only” and can be capped on request.
Validation against the source
Every evidence item is checked mechanically against its source: the page must be reachable, and every number, date and identifier in the evidence must appear there. Only validated evidence is shown and scored in full; everything else counts as a claim (grade 1). Any later change to text, source or grade voids the validation.
Distinction from the official SEAL
The official EU SEAL comes from the European Commission's Cloud III award (2026) and applies to the tendered service. We show it as a badge – we do not sell it as a score of our own; it only counts as evidence in the criterion 'EU labels & schemes'. Our scale is called evidence level, our aggregate sovereignty fit.
AI only for the summary
The evidence engine computes fit, status and order traceably from the evidence levels. The AI only writes the summary of the finished result. Every number in the text must appear in the result and only the visible providers are named; if a part fails, a template text replaces it. The result shows which text comes from the AI.
Frequently asked questions about the Sovereign Compass
How does the Sovereign Compass differ from the Cloud Compass?
The Cloud Compass rates workload fit – cost, security, availability, sovereignty and compliance – AI-assisted and with a report. The Sovereign Compass takes only the sovereignty dimension, but in depth: 23 criteria, a target evidence level per criterion derived from your need, and a traceably computed fit per provider. The AI only writes the summary; the result arrives by email, with the list of all sources.
Is the result a ranking of the most sovereign providers?
No. A sovereignty ranking answers “Who is the most sovereign?”, not “What does my workload need?”. The Sovereign Compass asks for the need first and then measures hyperscalers in EU regions and EU/CH providers against the same yardstick. The order in the result follows solely from the fit with your profile – met, met with an evidence gap, or knock-out. You see the first three providers straight away, all others after a short contact request.
What does the evidence level 0–4 mean?
0 = not given or refuted, 1 = claimed only, 2 = partially evidenced, 3 = evidenced by public documentation, contract text or certificate scope, 4 = fully evidenced and externally audited or technically enforced. We rate the quality of the evidence – not how sovereign a provider calls itself.
Is the official EU SEAL part of the rating?
Not as a score of its own. We show the SEAL from the European Commission's Cloud III award (2026) as a badge for OVHcloud, STACKIT and Scaleway because it is an external, official classification. It applies to the tendered service and is not converted into the sovereignty fit – it only counts as one piece of evidence in the criterion 'EU labels & schemes' (SOV-7). The fit is computed solely from the evidenced levels per criterion.
Does “knock-out” mean the provider is insecure?
No. Knock-out means a criterion you marked as exclusionary is not evidenced at the required level – for instance third-country access for classified data. Security is a separate dimension: hyperscalers too can be run GDPR-compliant and strongly encrypted in EU regions. Sovereign does not mean secure, and not sovereign does not mean insecure.
How do you keep invented or outdated evidence out?
Every evidence item is checked against its source without AI: the page must be reachable, and every number, date and identifier in the evidence must appear there. Only validated evidence is shown and scored in full; everything else counts as a claim (grade 1). If text, source or grade change, the validation lapses until the check passes again.

