elgarde / research

Email authentication in Portuguese real estate: agencies and builders, November 2026

We read the SPF and DMARC records of 5,038 estate agencies and 7,978 builders in Portugal. 40.6% of agencies, and 74.9% of builders with their own domain, leave the domain open to sender spoofing.

A study initiated by our partners at flatin.pt, who consulted us as their security specialists. flatin.pt publishes a version for the public and the press; this is the full technical version.

Summary

  • Of 5,038 estate agencies whose website we confirmed by their AMI licence number, 40.6% have an email domain exposed to sender spoofing, 48.8% are protected and 9.5% use only free webmail. Counting only the 4,503 that send email from their own domain, 45.4% are exposed.
  • Of 7,978 builders holding a classe 3 or higher licence with an email in the IMPIC register, 40.1% use only free webmail and 14.7% are protected. Of the 4,691 with their own domain, 74.9% are exposed. Free webmail plus exposed domains add up to 84.2% (6,715 companies).
  • Exposure grows as companies get smaller: 78.4% in classe 3-4 against 59.3% in classe 7-9 (counting only those sending from their own domain).
  • A single Portuguese website and CRM vendor for estate agencies hosts the mail of 1,333 agencies and publishes DMARC p=reject by default. Those agencies make up 52.5% of all protected agencies. Without them, protection among the remaining 3,170 falls to 36.8%.
  • The mail platform matters little. Among independent agencies, the two largest international business mail platforms have 81.3% and 86.1% of agencies exposed.
  • Of the 11 networks with 8 or more verified offices, 7 protect their main domain. 521 offices sit under a protected network domain and 158 under an exposed one.
  • 25.7% of agencies that publish a list of their consultants have at least one consultant on free webmail. On agencies’ own sites, 18.5% of consultant addresses are free webmail; on network pages, 0.9%.

Context: what is known about email fraud in real estate

No statistics exist, in Portugal or in the EU, on how much fraud relies on spoofing a company’s exact domain. Public series measure broader categories:

  • FBI IC3, Internet Crime Report 2025 (US): 24,768 business email compromise complaints and USD 3.05 billion in losses; in the real estate category, 12,368 complaints and USD 275 million.
  • Joint EBA and ECB report on payment fraud (December 2025, 2024 data): payer manipulation accounts for 74% of the value of credit transfer fraud (65% in 2023).
  • Action Fraud, quoted by The Negotiator (England and Wales, April 2024 to March 2025): 143 conveyancing fraud reports, GBP 11.7 million, an average of GBP 78,393 per case.
  • PSP (Portuguese police), via Lusa and Observador (April 2026): 4,553 rental-related crimes in three years. The GNR recorded 725 fraud cases in property purchases or rentals in 2025. These cases mostly involve fake listings.
  • Cybercrime Office of the Portuguese Prosecutor General (2023): rental scams use free mailboxes and messages sent in the name of platforms.
  • Caixa Geral de Depósitos, “Fraude imobiliária” (April 2026): describes a fake email “from the agency” sending the buyer new bank details.
  • Banco de Portugal, via RTP (2026): the payee confirmation service has prevented EUR 6.5 million of fraud since May 2024.
  • CNCS (Portuguese National Cybersecurity Centre), Society Report 2025: without SPF, DKIM and DMARC the receiving server cannot verify that the sender’s domain is authentic.

Most payment diversion fraud uses compromised real mailboxes, lookalike domains or free accounts. Domain authentication closes one path among several, the cheapest one for the attacker, and costs only configuration. In the UK public sector, the NCSC reports more than 80 million spoofed emails blocked in 30 days after moving to strict DMARC.

Methodology

Sources and dates

Step Date
IMPIC AMI licence register (public lookup) 13.11.2026
Search and verification of agency websites 14.11.2026
DNS reads, agencies and builders 14.11.2026
Second read of _dmarc records 19.11.2026

Sources: IMPIC public register of AMI (estate agency) licences; IMPIC public and private works contractor licences (alvarás), classe 3 to 9, with the email from each record; web search and agency websites; public DNS (resolvers 1.1.1.1 and 8.8.8.8).

Agencies: from register to risk base

The register had 12,238 rows and 12,237 distinct licences (one licence appears twice with different addresses): 10,579 companies and 1,658 sole traders (ENI). We searched for all 12,237, with 20,970 search queries.

Search outcome Licences %
Verified (AMI number found on the site) 5,038 41.2%
Found by name only, no AMI number 440 3.6%
Social networks or portals only, no own site 4,234 34.6%
Not found 2,455 20.1%
Fetch error 70 0.6%

A licence counts as verified when its AMI number appears on the page found: on the agency’s own site (4,382) or on a network office page (656). Verification rate is 44.2% for companies and 22.0% for ENI. The domain analysed is the domain of the main email published on the verified page (4,739 cases); with no published email we used the site’s domain (299 cases). Shared hosts of website builders were excluded. We read 4,759 domains.

Of the licences with no own site, 2,323 appear only on social networks and 1,911 only on property portals (when both, social networks win). This is a lower bound: the 2,455 licences not found may also lack a site.

Builders

IMPIC lists of classe 3 to 9 licences (public and private works) have 8,255 rows and 8,158 distinct companies. 7,978 have an email in the register; 180 do not. By email, 3,625 use a domain matching the company name, 1,153 a domain that does not match the name (group, accountant, IT provider or abbreviation, scanned as is) and 3,200 free webmail. We read 4,736 domains, including flatin.pt’s developer base.

What we checked on each domain

  • MX: whether the domain receives mail and where it is hosted.
  • SPF: presence, number of records, final term (-all, ~all, ?all, +all or none), number of DNS lookups.
  • DMARC: presence, number of records, p, sp, pct, report recipients (rua).
  • DKIM: 18 common selectors, including the default names used by the large mail and bulk-sending platforms and generic ones such as default, selector1, selector2, k1, s1, mail and dkim.
  • MTA-STS, BIMI and the DNSSEC AD flag.

DKIM keys cannot be listed through DNS. We only probed the selectors above, so “not found” does not mean the domain does not sign with DKIM. DKIM does not enter the risk class.

Risk classes

Class Definition
High No DMARC, or no SPF, or SPF ending in ?all, +all or with no all term, or DMARC without a valid p tag
Medium DMARC p=none, or quarantine/reject with pct<100, or enforcing DMARC with SPF over 10 lookups
Low (protected) DMARC quarantine or reject at 100% and a valid SPF
Free webmail only Only a free webmail address: no own domain to protect
No MX Own domain with no MX record: receives no mail, but can be spoofed if unprotected

Exposed = high + medium. Protected = low. “Own-domain” percentages exclude free webmail and domains without MX.

Exclusions, unit of count and privacy

  • An agency is one verified AMI licence; a builder is one company (NIPC) with classe 3 or higher. Several agencies can share one domain (network offices).
  • Unverified licences and builders without an email in the register are outside the risk base.
  • The study is passive: we sent no email and contacted no server beyond public DNS.
  • We stored domains only. Consultant addresses were pseudonymised at fetch time and only their domains were counted. We publish no names of companies, networks or vendors, and no domains.

Results: agencies

Segment Analysed High risk Medium risk Low risk (protected) Free webmail only No email on domain With own domain % exposed* % protected*
Verified agencies (all) 5,038 962 (19.1%) 1,082 (21.5%) 2,459 (48.8%) 480 (9.5%) 55 (1.1%) 4,503 45.4% 54.6%
Independents (own or free domain) 4,332 928 (21.4%) 938 (21.7%) 1,932 (44.6%) 480 (11.1%) 54 (1.2%) 3,798 49.1% 50.9%
Offices on a network domain 706 34 (4.8%) 144 (20.4%) 527 (74.6%) 0 (0.0%) 1 (0.1%) 705 25.2% 74.8%
Companies (NIPC) 4,674 882 (18.9%) 1,012 (21.7%) 2,325 (49.7%) 404 (8.6%) 51 (1.1%) 4,219 44.9% 55.1%
Sole traders (ENI) 364 80 (22.0%) 70 (19.2%) 134 (36.8%) 76 (20.9%) 4 (1.1%) 284 52.8% 47.2%
Verified on own site 4,382 928 (21.2%) 964 (22.0%) 1,956 (44.6%) 480 (11.0%) 54 (1.2%) 3,848 49.2% 50.8%
Verified on a network page 656 34 (5.2%) 118 (18.0%) 503 (76.7%) 0 (0.0%) 1 (0.2%) 655 23.2% 76.8%
With a published e-mail 4,739 928 (19.6%) 1,045 (22.1%) 2,239 (47.2%) 480 (10.1%) 47 (1.0%) 4,212 46.8% 53.2%

* Share calculated only among companies that send email from their own domain (excluding free-webmail-only and no email on the domain).

Offices on a network domain are far better protected (74.8% counting only those sending from their own domain) than independents (50.9%). Among independents, companies and ENI are close (51.2% and 47.2% protected), but ENI use free webmail much more often: 20.9% against 8.6% for companies.

Consultants on free webmail

We counted a consultant list when the verified page carried at least two personal addresses. 685 agencies (13.6% of verified) publish such a list.

Where Lists ≥1 free-mail consultant % Free addresses Company addresses % free
Network pages 265 31 11.7% 33 3,843 0.9%
Own sites 420 145 34.5% 257 1,132 18.5%
Total 685 176 25.7% 290 4,975 5.5%

A free webmail address does not need to be spoofed: anyone who wants to imitate it can open a free account with a similar name, and no DNS record prevents that.

Results: agency networks

We identified 19 network brands with 706 verified offices. The 11 networks with 8 or more offices (696 offices) were analysed separately. The main domain is protected in 7, at p=none in 2, in the high class in 1 and without MX in 1.

Status of offices in the 11 networks Offices
Under a protected network domain 521 (74.9%)
Under an exposed network domain 158
On their own or another domain 17

The network domain’s configuration decides for all offices at once, in both directions. Offices with no published email were counted under the domain of the network page on which they were verified.

Results: builders, classe 3 and higher

Segment Analysed High risk Medium risk Low risk (protected) Free webmail only No email on domain With own domain % exposed* % protected*
All classe ≥3 with e-mail 7,978 2,041 (25.6%) 1,474 (18.5%) 1,176 (14.7%) 3,200 (40.1%) 87 (1.1%) 4,691 74.9% 25.1%
Classe 3-4 5,017 1,200 (23.9%) 773 (15.4%) 543 (10.8%) 2,449 (48.8%) 52 (1.0%) 2,516 78.4% 21.6%
Classe 5-6 2,456 706 (28.7%) 551 (22.4%) 437 (17.8%) 728 (29.6%) 34 (1.4%) 1,694 74.2% 25.8%
Classe 7-9 505 135 (26.7%) 150 (29.7%) 196 (38.8%) 23 (4.6%) 1 (0.2%) 481 59.3% 40.7%
Domain = company name 3,625 1,590 (43.9%) 1,097 (30.3%) 889 (24.5%) 0 (0.0%) 49 (1.4%) 3,576 75.1% 24.9%
Domain ≠ name (group, accountant, IT firm) 1,153 451 (39.1%) 377 (32.7%) 287 (24.9%) 0 (0.0%) 38 (3.3%) 1,115 74.3% 25.7%
“Promo”/”imobili” in name 333 72 (21.6%) 63 (18.9%) 39 (11.7%) 156 (46.8%) 3 (0.9%) 174 77.6% 22.4%

* Share calculated only among companies that send email from their own domain (excluding free-webmail-only and no email on the domain).

Free webmail falls from 48.8% in classe 3-4 to 4.6% in classe 7-9. Even among the largest, 59.3% of own domains are exposed. Whether the domain carries the company’s name or a third party’s barely changes the result: 75.1% and 74.3% exposed.

flatin.pt developer base

flatin.pt keeps a base of 299 developers, 219 of them with a domain we could analyse. The figures are indicative only: most domains come from the developer’s website rather than a registry email, so free webmail use is invisible, and the sample is not random.

Segment Analysed High risk Medium risk Low risk (protected) Free webmail only No email on domain With own domain % exposed* % protected*
flatin.pt developer base 219 61 (27.9%) 84 (38.4%) 65 (29.7%) 2 (0.9%) 7 (3.2%) 210 69.0% 31.0%
Domain from IMPIC e-mail 12 4 (33.3%) 4 (33.3%) 2 (16.7%) 2 (16.7%) 0 (0.0%) 10 80.0% 20.0%
Website domain (proxy) 207 57 (27.5%) 80 (38.6%) 63 (30.4%) 0 (0.0%) 7 (3.4%) 200 68.5% 31.5%

* Share calculated only among companies that send email from their own domain (excluding free-webmail-only and no email on the domain).

The vendor decides more than the company

A first classification by mail provider showed a large gap between the two largest international business mail platforms: 48.5% and 84.4% of agencies exposed. The arithmetic is right but the gap comes from composition: 220 network offices are hosted on one of them, under protected network domains (98.6% protected). Counting independents only, the two platforms have 81.3% and 86.1% exposed; by distinct domains, 81.0% and 85.6%. Neither publishes DMARC by default; the second platform’s domains lack SPF more often.

The same analysis surfaced the main factor. A single Portuguese website and CRM vendor for estate agencies hosts the mail of 1,333 agencies (29.6% of agencies sending from their own domain) and publishes DMARC p=reject by default. 96.9% of those agencies are protected, and they make up 52.5% of all protected agencies in the country. Without them, among the remaining 3,170 agencies, protection drops to 36.8% and exposure rises to 63.2%.

Agencies by mail hosting type (counting only those sending from their own domain)

Mail hosting type Analysed Share Protected Exposed
One Portuguese real-estate website/CRM vendor 1,333 29.6% 96.9% 3.1%
Portuguese hosting or ISP 1,287 28.6% 36.0% 64.0%
Own or unidentified server 638 14.2% 52.7% 47.3%
International business mail platform (1) 536 11.9% 51.5% 48.5%
International business mail platform (2) 366 8.1% 15.6% 84.4%
International hosting companies 225 5.0% 9.3% 90.7%
Security gateway in front of the mailbox 19 0.4% 21.1% 78.9%
Other 99 2.2% 10.1% 89.9%

Controlled comparison: independents, network offices and distinct domains

Mail hosting type Independents % exposed Network offices % exposed Distinct domains % exposed
One Portuguese real-estate website/CRM vendor 1,333 3.1% 0 n/a 1,333 3.1%
Portuguese hosting or ISP 978 72.3% 309 37.9% 976 72.0%
Own or unidentified server 532 56.8% 106 0.0% 532 56.6%
International business mail platform (1) 316 81.3% 220 1.4% 311 81.0%
International business mail platform (2) 324 86.1% 42 71.4% 333 85.6%
International hosting companies 224 90.6% 1 100.0% 220 90.5%

DMARC policy by hosting type, agency domains

Mail hosting type Domains No DMARC p=none p=quarantine p=reject No SPF
One Portuguese real-estate website/CRM vendor 1,333 0.5% 2.5% 0.2% 96.8% 0.0%
Portuguese hosting or ISP 976 29.7% 41.5% 21.2% 7.4% 0.2%
Own or unidentified server 532 30.5% 24.6% 41.5% 3.4% 1.1%
International business mail platform (1) 311 44.4% 35.4% 10.0% 10.3% 0.0%
International business mail platform (2) 333 43.8% 39.6% 9.9% 6.0% 15.9%
International hosting companies 220 44.5% 45.5% 4.1% 5.9% 2.7%

Builders by mail hosting type (counting only those sending from their own domain)

Mail hosting type Analysed Share Protected Exposed
One Portuguese real-estate website/CRM vendor 20 0.4% 95.0% 5.0%
Portuguese hosting or ISP 1,618 34.5% 22.3% 77.7%
Own or unidentified server 597 12.7% 26.0% 74.0%
International business mail platform (1) 1,679 35.8% 26.3% 73.7%
International business mail platform (2) 370 7.9% 14.9% 85.1%
International hosting companies 131 2.8% 9.9% 90.1%
Security gateway in front of the mailbox 119 2.5% 53.8% 46.2%
Other 157 3.3% 42.7% 57.3%
Mail hosting type Domains No DMARC p=none p=quarantine p=reject No SPF
One Portuguese real-estate website/CRM vendor 20 0.0% 5.0% 0.0% 95.0% 0.0%
Portuguese hosting or ISP 1,618 44.1% 33.0% 5.9% 16.8% 0.8%
Own or unidentified server 597 37.4% 34.5% 18.6% 9.4% 2.0%
International business mail platform (1) 1,679 42.1% 30.0% 14.9% 12.9% 0.7%
International business mail platform (2) 370 57.0% 27.6% 6.5% 8.6% 17.6%
International hosting companies 131 48.9% 39.7% 4.6% 6.9% 9.9%

Among builders the agency website vendor is marginal (20 companies, 95.0% protected). Outside it, no hosting type goes above 54% protection.

Results by region

Exposed and protected shares counting only those sending from their own domain; free webmail over all entities with a domain. Cells with a base under 30 are shown as insufficient sample. Districts normalised; all Madeira islands under Madeira and all Azores islands under Açores.

Agencies by district

Region Analysed With own domain % exposed* % protected* Free webmail only AMI licences % verified
Faro 579 515 53.6% 46.4% 10.2% 1,199 48.3%
Madeira 149 129 53.5% 46.5% 12.8% 324 46.0%
Açores 48 46 47.8% 52.2% 4.2% 92 52.2%
Coimbra 148 130 47.7% 52.3% 11.5% 339 43.7%
Porto 780 708 47.7% 52.3% 8.2% 2,033 38.4%
Lisboa 1,402 1,273 45.3% 54.7% 7.7% 3,801 36.9%
Viseu 87 78 44.9% 55.1% 6.9% 204 42.6%
Leiria 223 195 44.6% 55.4% 10.3% 476 46.8%
Setúbal 495 438 43.6% 56.4% 10.5% 1,169 42.3%
Castelo Branco 78 69 43.5% 56.5% 10.3% 144 54.2%
Braga 332 290 42.8% 57.2% 12.0% 841 39.5%
Évora 47 45 42.2% 57.8% 4.3% 90 52.2%
Viana do Castelo 98 82 40.2% 59.8% 15.3% 199 49.2%
Beja 40 33 39.4% 60.6% 17.5% 100 40.0%
Aveiro 255 232 37.9% 62.1% 8.6% 571 44.7%
Santarém 160 143 36.4% 63.6% 10.6% 369 43.4%
Vila Real 43 38 18.4% 81.6% 11.6% 101 42.6%
Bragança 29 23 insufficient sample insufficient sample insufficient sample 68 42.6%
Guarda 17 15 insufficient sample insufficient sample insufficient sample 61 27.9%
Portalegre 27 20 insufficient sample insufficient sample insufficient sample 52 51.9%

* Share calculated only among companies that send email from their own domain (excluding free-webmail-only and no email on the domain).

Builders by district

Region Analysed With own domain % exposed* % protected* Free webmail only
Castelo Branco 131 59 93.2% 6.8% 53.4%
Évora 72 37 83.8% 16.2% 47.2%
Faro 468 205 82.9% 17.1% 55.6%
Coimbra 295 171 82.5% 17.5% 40.3%
Bragança 100 34 82.4% 17.6% 64.0%
Açores 112 71 80.3% 19.7% 35.7%
Setúbal 462 246 78.5% 21.5% 45.5%
Braga 799 501 77.8% 22.2% 35.7%
Leiria 519 288 77.1% 22.9% 42.8%
Guarda 91 47 76.6% 23.4% 47.3%
Madeira 210 92 73.9% 26.1% 55.7%
Porto 1,323 879 73.6% 26.4% 32.5%
Aveiro 461 301 72.8% 27.2% 34.5%
Lisboa 1,827 1,122 72.5% 27.5% 37.4%
Vila Real 122 54 72.2% 27.8% 55.7%
Viseu 274 153 71.9% 28.1% 42.7%
Santarém 260 142 67.6% 32.4% 44.6%
Viana do Castelo 192 97 67.0% 33.0% 49.0%
Beja 59 25 insufficient sample insufficient sample 57.6%
Portalegre 34 12 insufficient sample insufficient sample 64.7%

* Share calculated only among companies that send email from their own domain (excluding free-webmail-only and no email on the domain).

Agencies by municipality (municipalities with 30 or more agencies analysed)

Region Analysed With own domain % exposed* % protected* Free webmail only
Lagos (Faro) 58 53 69.8% 30.2% 8.6%
Matosinhos (Porto) 85 73 63.0% 37.0% 11.8%
Loulé (Faro) 148 140 60.7% 39.3% 4.7%
Portimão (Faro) 72 63 55.6% 44.4% 11.1%
Coimbra (Coimbra) 64 58 55.2% 44.8% 7.8%
Castelo Branco (Castelo Branco) 36 33 54.5% 45.5% 8.3%
Lisboa (Lisboa) 487 448 53.3% 46.7% 6.6%
Funchal (Madeira) 110 99 52.5% 47.5% 10.0%
Porto (Porto) 191 178 52.2% 47.8% 5.8%
Caldas da Rainha (Leiria) 43 41 51.2% 48.8% 4.7%
Albufeira (Faro) 73 65 50.8% 49.2% 8.2%
Setúbal (Setúbal) 77 72 50.0% 50.0% 6.5%
Lagoa (Algarve) (Faro) 33 31 48.4% 51.6% 6.1%
Braga (Braga) 117 104 48.1% 51.9% 11.1%
Guimarães (Braga) 54 46 47.8% 52.2% 14.8%
Loures (Lisboa) 72 67 47.8% 52.2% 5.6%
Viseu (Viseu) 41 36 47.2% 52.8% 9.8%
Almada (Setúbal) 89 81 45.7% 54.3% 6.7%
Oeiras (Lisboa) 124 115 45.2% 54.8% 5.6%
Sesimbra (Setúbal) 54 42 45.2% 54.8% 20.4%
Seixal (Setúbal) 101 89 44.9% 55.1% 11.9%
Cascais (Lisboa) 249 227 44.1% 55.9% 6.8%
Torres Vedras (Lisboa) 45 41 43.9% 56.1% 8.9%
Vila Nova de Gaia (Porto) 175 157 43.3% 56.7% 9.7%
Leiria (Leiria) 59 52 42.3% 57.7% 10.2%
Faro (Faro) 43 38 42.1% 57.9% 11.6%
Aveiro (Aveiro) 64 62 41.9% 58.1% 3.1%
Gondomar (Porto) 47 43 41.9% 58.1% 6.4%
Sintra (Lisboa) 165 145 41.4% 58.6% 10.9%
Vila Franca de Xira (Lisboa) 52 45 40.0% 60.0% 9.6%
Vila do Conde (Porto) 36 33 39.4% 60.6% 8.3%
Vila Nova de Famalicão (Braga) 51 46 39.1% 60.9% 7.8%
Maia (Porto) 59 57 38.6% 61.4% 3.4%
Barreiro (Setúbal) 36 34 38.2% 61.8% 2.8%
Figueira da Foz (Coimbra) 41 34 38.2% 61.8% 17.1%
Odivelas (Lisboa) 48 42 38.1% 61.9% 10.4%
Montijo (Setúbal) 40 35 37.1% 62.9% 10.0%
Palmela (Setúbal) 34 30 33.3% 66.7% 11.8%
Viana do Castelo (Viana do Castelo) 37 34 32.4% 67.6% 8.1%
Mafra (Lisboa) 44 38 31.6% 68.4% 13.6%
Santa Maria da Feira (Aveiro) 48 45 28.9% 71.1% 6.2%
Amadora (Lisboa) 67 60 23.3% 76.7% 9.0%
Barcelos (Braga) 33 28 insufficient sample insufficient sample 12.1%
Olhão (Faro) 30 27 insufficient sample insufficient sample 10.0%
Póvoa de Varzim (Porto) 31 25 insufficient sample insufficient sample 19.4%
Silves (Faro) 31 26 insufficient sample insufficient sample 16.1%
Valongo (Porto) 31 28 insufficient sample insufficient sample 6.5%

* Share calculated only among companies that send email from their own domain (excluding free-webmail-only and no email on the domain).

Technical oddities (distinct domains)

Indicator Agencies % Builders %
Distinct domains 3,845 4,554
Non-existent domain (NXDOMAIN) 26 0.7% 67 1.5%
No MX 54 1.4% 87 1.9%
No SPF 121 3.1% 199 4.4%
SPF with +all 1 0.0% 1 0.0%
SPF with ?all 23 0.6% 24 0.5%
SPF without an all term 6 0.2% 17 0.4%
SPF ~all 1,266 32.9% 2,013 44.2%
SPF -all 2,402 62.5% 2,132 46.8%
SPF over 10 DNS lookups 47 1.2% 45 1.0%
Multiple SPF records 16 0.4% 21 0.5%
No DMARC 917 23.8% 2,017 44.3%
DMARC p=none 956 24.9% 1,391 30.5%
DMARC p=quarantine 513 13.3% 516 11.3%
DMARC p=reject 1,454 37.8% 624 13.7%
DMARC without a p tag 4 0.1% 5 0.1%
DMARC with pct<100 7 0.2% 38 0.8%
Multiple DMARC records 11 0.3% 18 0.4%
DMARC with reports (rua) 1,903 49.5% 973 21.4%
DKIM found (common selectors) 2,749 71.5% 3,291 72.3%
MTA-STS 5 0.1% 20 0.4%
BIMI 13 0.3% 22 0.5%
DNSSEC validated 90 2.3% 187 4.1%
High only because of SPF (DMARC already strict) 12 0.3% 10 0.2%

A few cases: one agency domain publishes “p=quarentine”, a typo that receivers treat as an invalid policy; one builder domain has “p=ninguno”. One domain in each group has SPF with +all, which authorises any server in the world. 12 agency domains have strict DMARC but stay in the high class because of SPF; against exact-domain spoofing they are already protected by DMARC, but we kept the class rule unchanged.

On the second read on 19 November, 6 agency domains and 5 builder domains had changed their DMARC policy since 14 November. On that read, rua recipients of agency domains were: 1,315 domains send reports to the website vendor mentioned above, 204 to their own domain, 105 to a DMARC reporting service, 283 to another external address and 7 to a free webmail mailbox.

Limitations

  • The agency base covers agencies with a website showing the AMI number. The 4,234 licences with only social networks or portals and the 2,455 not found are outside it; their email posture was not measured.
  • We measure published configuration, not actual delivery. An exposed domain may still be caught by the receiver’s filters; a protected domain does not stop compromised mailboxes or lookalike domains.
  • DKIM was probed on common selectors only and is not part of the classification.
  • Builders’ email is the one in the IMPIC register, which may be outdated or belong to an accountant.
  • flatin.pt’s developer base is not a random sample.
  • Data is from 13 to 19 November 2026; configurations change (11 domains changed in five days).

Recommendations

Runs entirely in your browser via public DNS-over-HTTPS (Cloudflare, Google). Nothing is sent to us. Passive lookup only: no mail is sent and nothing is probed. The result describes the domain's public DNS configuration at the moment of the check; it is not a security audit of the company.

Agencies and developers:

  1. List every service that sends email with the domain: mailbox, CRM, portals, newsletter, invoicing.
  2. Publish SPF with those services, ending in -all or ~all, under 10 DNS lookups.
  3. Turn on DKIM in each service.
  4. Publish DMARC with p=none and rua, read the reports for two to four weeks, move to p=quarantine (a low pct first, then up to 100) and then to p=reject.
  5. Give consultants addresses on the company domain and remove free webmail addresses from the website.
  6. Tell clients in writing, at first contact, that bank details never change by email.

Networks: the network domain is one setting that protects or exposes every office. A p=reject on the main domain, with subdomains covered by sp, protects the whole network at once. Offices sending from their own domains should follow the same path.

Builders: for classe 3-4 companies using only free webmail, the first step is an own domain with SPF, DKIM and DMARC from day one. Those that already have a domain follow the same path as agencies.

Hosting and website providers: the agency website vendor’s result shows that a default protects thousands of clients who never opened a DNS panel. Other providers can do the same.

Buyers and tenants: confirm any IBAN change by phone, on a number you already had; use your bank’s payee confirmation; read the full sender address; check the AMI licence in the IMPIC register.

Credits

A study initiated by flatin.pt, which asked how well home buyers are protected against fake emails. Version for the public and the press: flatin.pt. Technical analysis: elgarde.eu. Press contact: info@flatin.pt.

Contact

hello@elgarde.eu

Tell us what worries you. We reply within one business day, in English or Portuguese.