| Group | Percentage |
|---|---|
| of signatories past the two-month mark had the signed declaration on their site, as far as we could find | 14% |
| had nothing about the pledge that we could find on their own site | 54% |
| had not enforced DMARC on their main domain | 23% |
| of the cyber-security specialists on the list had not enforced DMARC | 27% |
Source: Digital Royalty analysis, data collected 29 October 2026
Download data
Comment
“Most of these firms sell cyber security or IT and more than a quarter of the security specialists still haven't set DMARC to block anything, it's free and it stops people sending email pretending to be you. The pledge says to put the signed declaration on your website and of the ones that should have by now, only about one in seven have. I'd expect the people selling security advice to have done it on their own sites first.”
What we found
The Cyber Resilience Pledge is a voluntary government pledge. Organisations sign up to three actions: make cyber a board responsibility, use NCSC’s Early Warning service, and require Cyber Essentials across their supply chains. The declaration they sign also says: “within 2 months, we will publish the signed pledge declaration on our company website”.
On 29 October 2026, a little under four months after the pledge launched on 7 July, we looked at the public websites of the 132 signatories that had been on the GOV.UK list for at least two months and whose site we could identify.
| What we found on the signatory’s own site | Signatories | Share |
|---|---|---|
| The signed declaration | 19 | 14% |
| A mention of the pledge, without the declaration | 42 | 32% |
| Nothing about the pledge that we could find | 71 | 54% |
A mention can be a news post, a blog, a trust page or a line in an annual report. The declaration’s text has four checkable lines: cyber governance training, the Cyber Governance Code of Practice, Early Warning and the Cyber Essentials Supplier Check tool. If we also count a page that restates at least three of them, the 19 becomes 29 (22%).
“Nothing we could find” is the most we can say. A declaration can sit in an image or on a page that search engines and the site’s own search both miss, so treat 54% as an upper bound. The limits below say how often our checks caught a miss. GOV.UK’s questions and answers say: “As this is a voluntary pledge, there is no formal assurance mechanism.” They add that DCMS can review some actions, and that not completing an action may in some cases lead to a company withdrawing or being removed.
This is our second look. 115 signatories had reached two months when we first checked on 9 October. 16 of them had the declaration on their site then and 18 did on 29 October, while the number with nothing we could find went from 65 to 59. We can’t tell whether those pages are new or whether we missed them the first time.
Results by type of signatory
Signatories past the two-month mark. We assigned the types by hand.
| Type | Signatories | Declaration found | Declaration or a mention found |
|---|---|---|---|
| Cyber-security specialists | 47 | 8 (17%) | 22 (47%) |
| IT services, consultancies and software | 40 | 5 (13%) | 17 (43%) |
| Other companies | 39 | 4 (10%) | 19 (49%) |
| Trade bodies, public and membership bodies | 6 | 2 (33%) | 3 (50%) |
The last row is too small to read anything into. Size shows no steady pattern. Using the last accounts type at Companies House, we found nothing about the pledge for 37 of 71 large organisations (52%) and 26 of 43 small ones (60%). On 9 October, with fewer signatories past two months, large organisations had the higher share. The bands also rest on name matching, and “large” includes medium-sized filers.
Another 26 measured signatories hadn’t reached two months by 29 October. 2 of them had already published the declaration and 4 mentioned the pledge.
Email and disclosure basics
The pledge doesn’t ask for any of the following, so nothing in this section is a breach of it. We measured them because they are NCSC’s own advice and the pledge is about strengthening cyber defences. NCSC says of DMARC: “Having a DMARC policy of ‘reject’ on all of your domains is the best way to prevent spoofing of your email.” Its vulnerability disclosure toolkit says security.txt “should be published to all your domains and subdomains”.
We could measure 158 signatories, each on its main domain.
| Check | Result | Share |
|---|---|---|
| DMARC enforced (quarantine or reject, at 100%) | 121 of 158 | 77% |
| DMARC not enforced | 37 of 158 | 23% |
| Valid security.txt | 13 of 141 readable sites | 9% |
| MTA-STS in enforce mode | 27 of 158 | 17% |
| HSTS on the homepage | 99 of 158 | 63% |
| Valid TLS certificate | 156 of 158 | 99% |
| DMARC enforced, valid security.txt and MTA-STS enforced, all three | 3 of 158, and up to 8 if the 5 otherwise qualifying sites with unreadable security.txt have a valid file | 2% to 5% |
DMARC tells receiving mail servers what to do with messages that fail authentication checks. Of the 37 domains without enforcement, 27 publish p=none, which tells receiving servers to take no action on failures, 8 have no record, and 2 enforce on only part of their mail. Mail providers run filters of their own, so this doesn’t mean forged mail from those domains always gets through. p=none is also the first stage NCSC describes on the way to reject, so some of the 27 may be part-way there.
| Type | Signatories | DMARC enforced | Valid security.txt (of readable) | MTA-STS enforced |
|---|---|---|---|---|
| Cyber-security specialists | 55 | 40 (73%) | 4 of 48 (8%) | 11 (20%) |
| IT services, consultancies and software | 54 | 44 (81%) | 7 of 50 (14%) | 6 (11%) |
| Other companies | 41 | 32 (78%) | 2 of 35 (6%) | 8 (20%) |
| Trade bodies, public and membership bodies | 8 | 5 (63%) | 0 of 8 | 2 (25%) |
Cyber-security specialists had the second lowest DMARC enforcement of the four groups. Of the 15 of 55 without it, 9 publish p=none and the other 6 have no record or enforce on only part of their mail. Among small cyber-security firms the figure is 11 of 37 (30%).
29 of the 132 signatories past the two-month mark (22%) had neither a declaration we could find nor enforced DMARC. DMARC isn’t a pledge item, so this is a count of both results and no more.
Our interest, and our own domain
Digital Royalty is a small UK web and software studio. We set up DNS and email for clients, sell support retainers that include security work, and run a free email deliverability check, so we have a commercial interest in organisations caring about these settings. Nobody commissioned this study. We haven’t signed the pledge, and we made none of the promises measured here.
We ran the same checks on our own domain on 29 October. DMARC is enforced (p=quarantine, 100%), security.txt is valid and HSTS is on. We don’t run MTA-STS, so we wouldn’t be among those with all three.
To look up any domain’s records, including your own, use our DNS lookup at _dmarc. followed by the domain name.
Method
How we did this
Data collected 29 October 2026
The list. We took the signatory list from GOV.UK, as updated on 6 October 2026: 167 organisations with the date each was added, plus one later addition, making 168. We found each organisation’s main domain from a UK Google search on its name, picked by hand, and checked the uncertain ones by fetching the homepage. 10 had no domain we could confirm and are left out, which leaves 158.
Who counts as past two months. An organisation counts if the date it was added to the list, plus two months, falls on or before 29 October. That covers signatories added between 7 July and 21 August: 138 organisations, 132 of them with a domain we could identify. Organisations sign on or before the date they are added, so each one has had at least two months.
Finding the declaration. For each signatory we searched its own domain in Google for the pledge by name, read its sitemap and homepage for pledge links, and used its own site search when nothing else turned up. We then fetched every candidate page and PDF. A page counts as the declaration if it carries the template wording, “hereby pledge”. It counts as a mention if it names the Cyber Resilience Pledge without that text, or if our fetch was blocked but Google’s own snippet for the page names the pledge. On 29 October we also re-fetched every page found on 9 October, because Google had stopped returning some pages that were still live. A page returning 404 or 410 doesn’t count.
Measuring the basics. We read DMARC, MTA-STS and TLS-RPT records from public DNS, then checked the HTTPS homepage, its certificate and HSTS header, and security.txt at the standard locations under RFC 9116. A valid security.txt has a contact and an expiry date that hasn’t passed. We couldn’t get a definite answer from 17 sites (bot walls and server errors), so they are left out of that figure and not counted as missing.
Checks. All 158 DMARC results agreed when re-resolved through a second DNS provider with a separate parser. 20 random security.txt results agreed when re-fetched with a different tool. A script ran the site’s own search on 20 random “nothing found” signatories, 16 of which had a usable search, and found one page we had missed (behind a faulty certificate chain). We fixed that, re-ran, and a fresh 20 found no misses, 15 of them searchable. On 4 November 2026 we re-read DMARC for all 158 and every result was unchanged. The GOV.UK list showed no new signatories past two months. We did not repeat the declaration or security.txt checks that day.
Limits.
- “Nothing found” doesn’t prove absence. One miss in 31 searchable checks on this run is a small sample, and that check only reads links with the pledge in their address, so it can miss a page too.
- We looked at one main domain per organisation. Large groups run many domains, and an unenforced main domain may sit beside enforced brand domains. A declaration could also sit on a parent or group site.
- We did not test the pledge’s three actions, which can’t be seen from outside. A missing declaration says nothing about whether board training, Early Warning or Cyber Essentials work has been done.
- The signing date isn’t public, so we use the date added to the list.
- The list has grown. It now has 242 named rows and its own total row says 235. Names added since 6 October aren’t in these figures.
- Types are our own hand classification, and size bands rest on Companies House name matching, with 23 organisations unmatched, 21 wrong matches rejected and 18 plainly large organisations labelled by hand.
- This is a snapshot of one date.
The data. We publish the aggregated tables here. We name no organisation on this page and we don’t publish a row-level file, because the date added and the type together can point to a single organisation on the public list. The GOV.UK list and every DNS record are public, so anyone can repeat the DMARC check. Journalists can ask for the full data, under embargo, and the method notes at press@digitalroyalty.co.uk, and more studies are on our research page.
FAQ
Does this mean these organisations broke the pledge?
We don’t say that. The pledge is voluntary and GOV.UK says it has no formal assurance mechanism. We report what we could and couldn’t find on public websites on one date, and the declaration may exist somewhere we didn’t reach.
Is DMARC part of the Cyber Resilience Pledge?
No. Nor are security.txt or MTA-STS. They are NCSC guidance, and we report them separately from the pledge’s own commitment.
Why don’t you name the organisations?
The finding is about the list as a whole, and a single missing page could have an innocent explanation. If a journalist names an organisation from the data, it’s for them to ask that organisation for comment.
Can I use these figures?
Yes, with a credit to Digital Royalty and a link to this page. Please keep the data date of 29 October 2026 and the wording “we could not find”, because we can’t show that a declaration doesn’t exist.