Document Trust Grade

An open grading scale for signed documents

Document Trust Grade v0.1.0

How strongly this file proves, on its own, who signed it and that it has not changed — and how long it will keep proving it, without the signing platform.

Version 0.1.0 · pre-release, open for comment · published 2026-08-31

How it works

Three questions in strict order: integrity (has it changed?), then trust (does the identity chain to a recognised trust regime?), then durability (will it still verify years from now without anyone's help?). This is a ladder of necessary conditions, not a points total — a file receives the highest grade whose conditions it meets in full, and one failed gate caps it regardless of everything else.

The scale

A+

Beyond the standard

The supplier went further than the standard asks.

All of

  • Everything required for A.
  • ETSI baseline B-LTA — carries archive timestamps, so the proof can be refreshed as cryptographic algorithms age and outlives the certificate itself.
  • No post-signing modification permitted or detected.
  • At least one marker from the closed list below.
A

The ETSI standard, met in full

This is what a properly signed document looks like. It proves who signed it, when, and that nothing has changed — on its own, offline, without the platform that made it.

All of

  • Validation outcome TOTAL_PASSED — intact, chain valid, not revoked.
  • Chains to a trust regime whose ceiling is A: an EU/EEA Trusted List, or a third-country list recognised through an eIDAS Article 14 mutual-recognition agreement.
  • Qualifies as a qualified electronic signature (QES) or qualified electronic seal (QESeal).
  • ETSI baseline B-LT or B-LTA — certificates and revocation data embedded, so the file verifies offline without contacting anyone.
  • The signature covers the whole document.
B

Trusted and time-anchored

An independent third party has proven when this was signed, and it will keep verifying for years.

All of

  • Clears everything required for C, and carries a qualified timestamp (ETSI B-T or better).
  • Misses at least one A requirement — either the trust regime's own ceiling is B, or it is qualified but not yet self-contained (B-T rather than B-LT/B-LTA).
C

Trusted, but decaying

Verifies today. There is no independent proof of when it was signed, and once the signing certificate expires or its revocation data goes offline, it may stop verifying entirely.

All of

  • Intact, and chains to a recognised trust regime.
  • ETSI baseline B-B — no qualified timestamp, no embedded validation material.
D

Provable file, unprovable signer

You can prove this file has not changed. You cannot prove who signed it.

All of

  • The signature verifies and the document is intact.
  • The certificate chains to no trust regime we recognise — self-signed, a private CA, or an unknown issuer.
E

A claim, without proof

Something was signed. This file cannot prove it. The proof sits with the signing platform, not with you. Nothing here is wrong — it is simply not self-proving.

All of

  • The file carries a signing claim — an attached evidence package, a platform audit trail, or a visual signature.
  • There is no cryptographic signature over the document that can be independently verified, so integrity cannot be checked at all.
F

Fails

This document does not prove what it claims. Do not rely on it without going back to the source.

Any of

  • The signed content has been modified since signing.
  • The signature is cryptographically invalid.
  • The signing certificate was revoked at the time of signing.
—

Not graded

No letter is issued when there is no signature and no evidence package to assess, or the file cannot be read. This is never shown as F. F means we checked and the document failed; absence of proof is a different statement from disproof.

Trust regimes and their ceilings

A validator applying the scale checks every trust source it carries and names the one that vouched for the document, rather than refusing documents outside a preferred list. What differs between regimes is what a positive result legally means — and that sets the highest grade the document can reach.

RegimeWhat it isLegal weightCeiling
EU/EEA qualifiedMember-state trusted lists under eIDAS Article 22, where the service is granted for qualified certificates. Published by each member state, aggregated by the European Commission.Presumption under eIDAS Article 25 (signatures) / Article 35 (seals).A+
Recognised third countryTrusted lists of countries holding an eIDAS Article 14 mutual-recognition agreement — currently Ukraine and Moldova.Legally equivalent to EU-qualified, by agreement.A+
EU/EEA non-qualifiedOn a member-state trusted list, but the service is not granted for qualified certificates.Supervised; no presumption.B
National regulatedA national trusted list operating outside eIDAS recognition — the Swiss list (ZertES, Swiss Accreditation Service) and the UK list (Information Commissioner's Office). Real supervision, real audit, no EU mutual recognition.Regulated nationally; no eIDAS presumption.B
Commercial programmeThe Adobe Approved Trust List — around 300 certificate authorities admitted under Adobe's published technical requirements and independent audit. This is what makes Adobe Acrobat show a green tick.Contractual, not statutory; no presumption.B
Not recognisedChains to no list we carry: self-signed, a private certificate authority, or an unknown issuer.None.D

The A+ markers

This list is closed and published, so A+ is achievable rather than a matter of taste. A document needs everything required for A, plus the conditions listed under A+, plus at least one of these:

Rules for every grade

A document's grade is the lowest of its signatures' grades — a contract is only as good as its weakest party's proof. Each signature also carries its own grade, so the roll-up is never opaque.

A grade is never shown as a bare letter. It always appears with the assurance level, the ETSI form, and the trust regime that vouched for it — the letter carries the legibility, the rest carries the defensibility.

A real document, walked through the ladder

A PDF sealed by an e-signing platform, validated by Tindom. It clears every gate up to the top of the ladder and is then held at B by two separate things — which is the ordinary case, not a failure.

IntegrityIntact — the signed content has not been modified since sealing.
RevocationNot revoked, checked live against the issuer's OCSP responder at validation time.
Trust regimeEU Trusted List (Norway), via a trust service granted for qualified certificates. Regime ceiling: A+.
Signature scopeCovers the whole document.
DurabilityPAdES-BASELINE-T. A timestamp proves when it was sealed, but the certificates and revocation data are not embedded — verifying it later means fetching them from the issuer, which will not be possible forever. A needs B-LT.
Assurance levelAdvanced electronic seal supported by a qualified certificate — not a full qualified electronic seal. A requires QES or QESeal.

Grade B. Trusted and time-anchored, and it will keep verifying for years — but two separate things stand between it and A, and either one alone would be enough to hold it there. Both are configuration changes on the signing platform's side, not re-architecture.

How to raise a grade

The scale is only useful if it says what to change. Each entry below is a ceiling a document can hit, what it means, and what would move it up.

Timestamped, but the file does not carry what is needed to verify it on its own later (B-T).

Embed the certificate chain and revocation data at signing time — B-T becomes B-LT. Usually a setting in the signing platform.

The signature alone, with no independent proof of when it was made (B-B).

Add a qualified timestamp at signing — B-B becomes B-T, and the document stops depending on a claimed signing time.

An advanced signature or seal, possibly on a qualified certificate, but not a qualified signature or seal in the eIDAS sense.

Issue on a qualified certificate held on a qualified signature creation device (QSCD). This is a change of product with the trust service provider, not a file-format change.

Self-contained, but without archive timestamps the proof cannot be refreshed as cryptographic algorithms age (B-LT).

Apply an archive timestamp — B-LT becomes B-LTA, and the evidence outlives both the certificate and the algorithms in use today.

The certificate chains to nothing we recognise — self-signed, a private authority, or an unknown issuer.

Sign with a certificate from a provider on one of the trust lists above. An internal certificate authority cannot reach beyond D, however well run.

The file records that signing happened, but carries no cryptographic signature we can check.

Embed a real signature in the document itself (PAdES for PDFs) rather than keeping the proof on the platform. This is the single largest step available — it moves a document from E to a graded rung.

Content sits outside what the signature actually covers.

Sign the whole document. A partial signature vouches only for the part it covers, whatever the rest of the file appears to say.

What introduced this version

0.1.0 · 2026-08-31 · initial release

First published version of the scale. Pre-release under semantic versioning: the thresholds have not yet been calibrated against a real distribution of documents, so they may still move.

Grades affected: Not applicable — no earlier version exists.

All versions and the full changelog