We scored 73 domains belonging to companies and bodies that sell, standardise or operate digital identity: enterprise single sign-on, identity governance and privileged-access vendors; the zones that serve real sign-in traffic; MFA and hardware-key makers; developer and customer-identity platforms; password managers; identity-verification and national eID providers; and the standards bodies behind FIDO and OpenID. Each was scored on NODA's dual-axis methodology: a Health score for whether the domain's DNS works right now, and a Best Practice score for whether it meets the standards that limit future risk, across 41 checks in five weighted categories.
A note on scope: the list was compiled from well-known vendors and bodies in each sub-group, not from a canonical index of the identity industry, so treat it as a representative sample rather than a census. Only each organisation's apex zone was scored. All 73 domains were checked live on November 7, 2026, between 14:00 and 14:09 UTC, under NODA-v1.0.2. DNS configurations drift, so exact figures will shift over time.
This piece reports the output of a deterministic, published and open technical methodology (NODA-v1.0.2) applied identically to every domain listed. It describes DNS configuration state as observed on the check date only. It is not a judgment of any organisation's products, overall security posture, engineering quality or fitness as a vendor, and nothing here should be read as investment guidance or a claim of financial materiality regarding any publicly traded company named. A score reflects a specific, fixable technical state on a given day, not a permanent verdict. Every domain was scored under the same public rule set, and anyone can run the same 41 checks against their own domain at noda.marinadns.io.
The numbers
| Metric | Value |
|---|---|
| Domains scored | 73 |
| Average Health score | 97.7 / 100 (median 98) |
| Health range | 91 – 100 |
| Average Best Practice score | 79.3 / 100 |
| Median Best Practice score | 79 / 100 |
| Best Practice range | 67 – 100 |
| Zones with a DS record at the parent (DNSSEC signed) | 24 of 73 (32.9%) |
| DMARC enforced on all mail | 61 of 73 (83.6%) |
On Health these vendors do very well: all 73 domains are in the Exceptional tier. None runs an open recursive resolver, and none allows a zone transfer to an arbitrary client. The variation sits on the Best Practice axis, where only 7 domains reach Exceptional, 41 are Good and 25 are in Needs Improvement.
The scores bunch at a few values, and the line between those clusters is mostly DNSSEC. Every signed zone scores 82 or higher, and all 37 domains below 82 are unsigned. Signing isn't the only route to a high Best Practice score: pingidentity.com (90) and roboform.com (90) score 90 or above without it, on the strength of everything else. That cuts the other way too. Under NODA, SPF and DMARC count toward Health, not Best Practice, which is why clear.me reaches 87 on Best Practice while publishing neither record; its missing mail authentication shows up in its Health score of 91, the lowest in the cohort.
Signed, half-signed and unsigned
DNSSEC lets a validating resolver confirm that an answer really came from the zone's owner and wasn't changed on the way. It only works if the chain of trust is unbroken: the zone has to publish keys, and its parent has to publish a matching DS record. Without the DS record, validators have no way of knowing the keys are meant to be trusted.
24 of 73 zones (32.9%) have a complete chain. Another 4 publish DNSKEY records with no DS at the parent, which leaves their signatures invisible to validators: authelia.com, fidoalliance.org, saviynt.com and workos.com. The remaining 45 (61.6%) aren't signed at all. Every one of the 28 zones that has keys uses an allowed signing algorithm, and every published algorithm is covered by a signature, so where these vendors sign, they sign properly. The gap is that most don't.
The unsigned list includes some of the best-known names in workforce identity: okta.com, auth0.com, pingidentity.com, onelogin.com, sailpoint.com and rsa.com, and Cisco's duo.com and duosecurity.com. Ownership doesn't predict the outcome either: forgerock.com, the brand Ping Identity acquired, is signed; pingidentity.com isn't.
| Sub-group | n | Signed | Keys, no DS | Unsigned | Median BP | Median Health |
|---|---|---|---|---|---|---|
| Enterprise SSO / IAM / PAM | 18 | 4 | 1 | 13 | 74 | 99 |
| Auth-serving zones | 5 | 1 | 0 | 4 | 74 | 98 |
| MFA / hardware / FIDO | 14 | 4 | 0 | 10 | 75 | 97 |
| Developer / CIAM | 11 | 2 | 2 | 7 | 75 | 97 |
| Password managers | 9 | 5 | 0 | 4 | 86 | 99 |
| Identity verification / eID | 12 | 8 | 0 | 4 | 86 | 99 |
| Standards bodies | 4 | 0 | 1 | 3 | 78.5 | 96.5 |
The identity-verification and national eID group does best: 8 of 12 are signed, including BankID, MitID, login.gov, Onfido, Veriff, IDnow and Signicat. The standards bodies do worst: 0 of 4 have a complete chain of trust. The FIDO Alliance publishes keys but has no DS record at its parent, and the OpenID Foundation's zone isn't signed.
Following the Microsoft login path
An apex score doesn't always describe the hostname users actually sign in through, so we traced both Microsoft sign-in hostnames through a validating resolver (1.1.1.1), checked every zone in the CNAME chain for a DS record, and noted whether the final answer came back authenticated (the AD flag). As a control, login.gov returns the AD flag.
| Sign-in host | Chain | Result |
|---|---|---|
login.microsoftonline.com (Microsoft Entra ID) | login.microsoftonline.com → login.mso.msidentity.com → ak.privatelink.msidentity.com → www.tm.a.prd.aadg.akadns.net → A. Zones microsoftonline.com and msidentity.com: no DS. | AD flag not set |
login.live.com (Microsoft account) | login.live.com (zone live.com, signed) → login.msa.msidentity.com → … → A. Zone msidentity.com: no DS, no DNSKEY. | AD flag not set |
Signing live.com isn't enough on its own. Both chains pass through msidentity.com, which publishes no keys at all, so neither final answer validates. One unsigned zone anywhere in a CNAME chain is enough to break it. (Traced November 5 and re-checked November 7, 2026.)
Mail: mostly enforced, with notable exceptions
Account recovery, password resets and magic links all arrive by email. If a vendor's domain can be spoofed in email, attackers can impersonate that vendor to its own customers however strong its MFA product is. This is where these vendors are strongest.
47 domains publish p=reject and 16 publish p=quarantine. Counting only policies applied to all failing mail, 61 of 73 (83.6%) enforce DMARC. 2 quarantine records don't apply to all mail: forgerock.com sets pct=0 and kantarainitiative.org sets pct=25. 8 domains are in monitoring mode with p=none: fidoalliance.org, idsalliance.org, live.com, onlykey.io, openid.net, signicat.com, token2.com and zitadel.com, among them the FIDO Alliance, the OpenID Foundation, the IDSA and Microsoft's live.com.
2 domains publish no DMARC record, and both receive mail: clear.me and solokeys.com. clear.me publishes no SPF record either.
On SPF, 30 domains end with -all and 39 with ~all. With an enforced DMARC policy, ~all is a common and defensible choice, so we don't count it as a weakness. 3 domains hand off to a hosted record with redirect=; NODA follows the redirect and scores all of them as passing, but because their final policy lives in a record hosted elsewhere, we don't count them as either. We found DKIM keys for 49 of the 71 domains that receive mail. NODA can only test common selector names, so not finding a key doesn't mean a domain doesn't sign (§6.3), and we make no claim about DKIM absence.
Top and bottom
Highest Best Practice scores (92 and above):
| Domain | Sub-group | Best Practice | Health | DNSSEC |
|---|---|---|---|---|
| thalesgroup.com | MFA / hardware / FIDO | 100 | 100 | Signed |
| bankid.com | Identity verification / eID | 97 | 100 | Signed |
| mitid.dk | Identity verification / eID | 97 | 98 | Signed |
| imprivata.com | Enterprise SSO / IAM / PAM | 94 | 97 | Signed |
| proton.me | Password managers | 92 | 100 | Signed |
Lowest Best Practice score (67); all 6 domains tied at that score are listed:
| Domain | Sub-group | Best Practice | Health | DNSSEC |
|---|---|---|---|---|
| delinea.com | Enterprise SSO / IAM / PAM | 67 | 99 | Unsigned |
| fusionauth.io | Developer / CIAM | 67 | 99 | Unsigned |
| hidglobal.com | MFA / hardware / FIDO | 67 | 97 | Unsigned |
| rsa.com | Enterprise SSO / IAM / PAM | 67 | 99 | Unsigned |
| secureauth.com | Enterprise SSO / IAM / PAM | 67 | 99 | Unsigned |
| strongdm.com | Enterprise SSO / IAM / PAM | 67 | 99 | Unsigned |
Every domain in the bottom group scores 97 or higher on Health. These are zones that work fine today, with headroom against recommended practice rather than active operational risk.
All 73 domains
Each score carries a percentile within this cohort: P80 means the domain scored higher than about 80% of the 73 domains, with ties counted as half (mid-rank), computed separately per axis. Health scores cluster tightly near the top, so a point or two there can move a domain a long way in percentile terms; the Best Practice percentile is the more informative of the two.
| Domain | Sub-group | Best Practice | Health | DNSSEC | DMARC | SPF |
|---|---|---|---|---|---|---|
| thalesgroup.com | MFA / hardware / FIDO | 100 P99 | 100 P97 | Signed | p=quarantine | -all |
| bankid.com | Identity verification / eID | 97 P97 | 100 P97 | Signed | p=reject | -all |
| mitid.dk | Identity verification / eID | 97 P97 | 98 P49 | Signed | p=reject | -all |
| imprivata.com | Enterprise SSO / IAM / PAM | 94 P95 | 97 P30 | Signed | p=reject | ~all |
| proton.me | Password managers | 92 P94 | 100 P97 | Signed | p=quarantine | ~all |
| roboform.com | Password managers | 90 P92 | 99 P74 | Unsigned | p=quarantine | -all |
| pingidentity.com | Enterprise SSO / IAM / PAM | 90 P92 | 95 P9 | Unsigned | p=quarantine | ~all |
| signicat.com | Identity verification / eID | 89 P88 | 98 P49 | Signed | p=none | ~all |
| bitwarden.com | Password managers | 89 P88 | 97 P30 | Signed | p=reject | ~all |
| id.me | Identity verification / eID | 89 P88 | 97 P30 | Signed | p=reject | ~all |
| cyberark.com | Enterprise SSO / IAM / PAM | 89 P88 | 96 P14 | Signed | p=reject | -all |
| clear.me | Identity verification / eID | 87 P84 | 91 P1 | Unsigned | No record | No record |
| dashlane.com | Password managers | 86 P75 | 99 P74 | Signed | p=reject | -all |
| keepersecurity.com | Password managers | 86 P75 | 99 P74 | Signed | p=reject | redirect= |
| login.gov | Identity verification / eID | 86 P75 | 99 P74 | Signed | p=reject | ~all |
| onfido.com | Identity verification / eID | 86 P75 | 99 P74 | Signed | p=quarantine | -all |
| veriff.com | Identity verification / eID | 86 P75 | 99 P74 | Signed | p=reject | -all |
| yubico.com | MFA / hardware / FIDO | 86 P75 | 99 P74 | Signed | p=reject | ~all |
| entrust.com | MFA / hardware / FIDO | 86 P75 | 98 P49 | Signed | p=reject | -all |
| zitadel.com | Developer / CIAM | 86 P75 | 98 P49 | Signed | p=none | ~all |
| descope.com | Developer / CIAM | 86 P75 | 97 P30 | Signed | p=reject | -all |
| keepassxc.org | Password managers | 86 P75 | 97 P30 | Signed | p=reject | ~all |
| nitrokey.com | MFA / hardware / FIDO | 86 P75 | 97 P30 | Signed | p=reject | -all |
| forgerock.com | Enterprise SSO / IAM / PAM | 86 P75 | 96 P14 | Signed | p=quarantine · pct=0 | -all |
| ftsafe.com | MFA / hardware / FIDO | 85 P66 | 98 P49 | Unsigned | p=quarantine | -all |
| 1password.com | Password managers | 83 P62 | 99 P74 | Unsigned | p=quarantine | ~all |
| beyondtrust.com | Enterprise SSO / IAM / PAM | 83 P62 | 99 P74 | Signed | p=reject | ~all |
| idnow.io | Identity verification / eID | 83 P62 | 99 P74 | Signed | p=reject | -all |
| lastpass.com | Password managers | 83 P62 | 99 P74 | Unsigned | p=reject | redirect= |
| silverfort.com | Enterprise SSO / IAM / PAM | 83 P62 | 99 P74 | Unsigned | p=reject | -all |
| openid.net | Standards bodies | 83 P62 | 98 P49 | Unsigned | p=none | -all |
| live.com | Auth-serving zones | 82 P54 | 98 P49 | Signed | p=none | ~all |
| workos.com | Developer / CIAM | 82 P54 | 97 P30 | Keys, no DS | p=reject | ~all |
| fidoalliance.org | Standards bodies | 82 P54 | 96 P14 | Keys, no DS | p=none | ~all |
| saviynt.com | Enterprise SSO / IAM / PAM | 82 P54 | 96 P14 | Keys, no DS | p=quarantine | ~all |
| authelia.com | Developer / CIAM | 82 P54 | 95 P9 | Keys, no DS | p=reject | -all |
| okta.com | Enterprise SSO / IAM / PAM | 79 P50 | 98 P49 | Unsigned | p=reject | -all |
| authy.com | MFA / hardware / FIDO | 75 P42 | 97 P30 | Unsigned | p=reject | ~all |
| frontegg.com | Developer / CIAM | 75 P42 | 97 P30 | Unsigned | p=reject | ~all |
| goteleport.com | Enterprise SSO / IAM / PAM | 75 P42 | 97 P30 | Unsigned | p=reject | -all |
| kantarainitiative.org | Standards bodies | 75 P42 | 97 P30 | Unsigned | p=quarantine · pct=25 | ~all |
| nordpass.com | Password managers | 75 P42 | 97 P30 | Unsigned | p=reject | -all |
| onespan.com | MFA / hardware / FIDO | 75 P42 | 97 P30 | Unsigned | p=reject | ~all |
| ory.sh | Developer / CIAM | 75 P42 | 97 P30 | Unsigned | p=reject | ~all |
| supertokens.com | Developer / CIAM | 75 P42 | 97 P30 | Unsigned | p=reject | -all |
| idsalliance.org | Standards bodies | 75 P42 | 96 P14 | Unsigned | p=none | ~all |
| onlykey.io | MFA / hardware / FIDO | 75 P42 | 94 P4 | Unsigned | p=none | ~all |
| solokeys.com | MFA / hardware / FIDO | 75 P42 | 94 P4 | Unsigned | No record | ~all |
| okta-emea.com | Auth-serving zones | 74 P33 | 100 P97 | Unsigned | p=reject | -all |
| oktapreview.com | Auth-serving zones | 74 P33 | 94 P4 | Unsigned | p=quarantine | ~all |
| onelogin.com | Enterprise SSO / IAM / PAM | 73 P31 | 100 P97 | Unsigned | p=reject | -all |
| auth0.com | Enterprise SSO / IAM / PAM | 71 P19 | 99 P74 | Unsigned | p=quarantine | -all |
| duo.com | MFA / hardware / FIDO | 71 P19 | 99 P74 | Unsigned | p=reject | ~all |
| duosecurity.com | Auth-serving zones | 71 P19 | 99 P74 | Unsigned | p=reject | ~all |
| hypr.com | MFA / hardware / FIDO | 71 P19 | 99 P74 | Unsigned | p=reject | ~all |
| itsme-id.com | Identity verification / eID | 71 P19 | 99 P74 | Unsigned | p=reject | -all |
| jumio.com | Identity verification / eID | 71 P19 | 99 P74 | Unsigned | p=reject | ~all |
| jumpcloud.com | Enterprise SSO / IAM / PAM | 71 P19 | 99 P74 | Unsigned | p=quarantine | redirect= |
| keycloak.org | Developer / CIAM | 71 P19 | 99 P74 | Unsigned | p=quarantine | ~all |
| stytch.com | Developer / CIAM | 71 P19 | 99 P74 | Unsigned | p=quarantine | -all |
| transmitsecurity.com | Enterprise SSO / IAM / PAM | 71 P19 | 99 P74 | Unsigned | p=reject | ~all |
| withpersona.com | Identity verification / eID | 71 P19 | 99 P74 | Unsigned | p=reject | ~all |
| sailpoint.com | Enterprise SSO / IAM / PAM | 71 P19 | 98 P49 | Unsigned | p=reject | ~all |
| clerk.com | Developer / CIAM | 71 P19 | 97 P30 | Unsigned | p=reject | ~all |
| protectimus.com | MFA / hardware / FIDO | 71 P19 | 97 P30 | Unsigned | p=reject | ~all |
| token2.com | MFA / hardware / FIDO | 71 P19 | 95 P9 | Unsigned | p=none | ~all |
| microsoftonline.com | Auth-serving zones | 71 P19 | 94 P4 | Unsigned | p=reject | -all |
| delinea.com | Enterprise SSO / IAM / PAM | 67 P4 | 99 P74 | Unsigned | p=reject | -all |
| fusionauth.io | Developer / CIAM | 67 P4 | 99 P74 | Unsigned | p=reject | -all |
| rsa.com | Enterprise SSO / IAM / PAM | 67 P4 | 99 P74 | Unsigned | p=reject | ~all |
| secureauth.com | Enterprise SSO / IAM / PAM | 67 P4 | 99 P74 | Unsigned | p=quarantine | ~all |
| strongdm.com | Enterprise SSO / IAM / PAM | 67 P4 | 99 P74 | Unsigned | p=reject | ~all |
| hidglobal.com | MFA / hardware / FIDO | 67 P4 | 97 P30 | Unsigned | p=reject | -all |
A note on our own configuration
NODA and MarinaDNS aren't exempt from the standard this article applies to everyone else. As of the check date, marinadns.io and noda.marinadns.io are not DNSSEC-signed either; marinadns.io publishes DMARC at p=reject. The fix is the one this article recommends to everyone in it, and it's on our own list.
What to make of it
The companies in this study build products that decide who gets in. The DNS under their own domains mostly works, and their mail policies are mostly in good shape. What's missing in most cases is DNSSEC, the one control that lets everyone else verify that their DNS answers are genuine. That's not proof of compromise, and DNSSEC isn't the only defence against DNS spoofing. But a vendor whose business is proving identity is, in most cases here, publishing its own domain's records without proof of origin.
Turning it on isn't technically hard. Every signed zone in the sample passes both of NODA's signature-quality checks, and for the 4 zones that already publish keys, the missing step is a single DS record at the parent, usually set through the registrar. Keeping it healthy is the harder part: previously we discussed why DNSSEC needs ongoing attention, and why that keeps adoption low.
What to do about it
- Sign the zones your users actually touch, including every zone in the CNAME chain behind your sign-in hostnames, not only the marketing apex.
- If you already publish keys, finish the chain: publish the DS record at your registrar, or use CDS/CDNSKEY if your registry scans for it.
- Move DMARC from
p=noneto enforcement once reports show your legitimate senders aligned, and droppctbelow 100 only as a short ramp. - Publish DMARC and SPF even on domains that only receive mail, so nobody can send as you.
- If you're evaluating an identity vendor, check its domain: the same checks are public and take seconds.
Curious where your own domain lands? Run it through NODA and see the same 41 checks scored live.
⇩ Download full report (PDF, 73 domains)
Population: 73 apex domains in seven sub-groups; all 73 were assessed without error. Scoring: NODA-v1.0.2; Health and Best Practice are computed separately and never combined. We recomputed every score independently from the published Appendix A weights and matched all 73 stored values. Cross-check: DS, DNSKEY, DMARC and SPF were re-queried for every domain with raw dig against 1.1.1.1 and agreed with NODA on all 73. Version: a first run on November 5, 2026 used NODA-v1.0.1; every figure here comes from the November 7 re-run under v1.0.2, which fixes DMARC and SPF parsing and ASN mapping in the reference engine. DNSSEC findings were identical in both runs. DKIM: a key NODA can't find at the selectors it checks is recorded as Informational, not as absent (§6.3). Scope: apex zones at a single point in time; a score describes public DNS on the check date, not the security of anyone's products. Data: NODA bulk batch identity_provider_v102_2026-11-07.
Fix the gaps this report found
- DNSSEC Checker — see whether your zone is signed and the chain of trust validates
- SPF / DKIM / DMARC Checker — audit all three mail-authentication records in one pass
- DMARC Record Generator — build the record that moves you from p=none to enforcement