Add DMARC and SPF adoption data for 36,120 large organizations

Collected August 2026. Population is organizations with 5,001 or more employees and resolvable company information, matching stackscan/lets-encrypt.

README.md

@@ -0,0 +1,217 @@
+---
+pretty_name: "DMARC and SPF Adoption Among Large Organizations"
+license: cc-by-4.0
+tags:
+  - cybersecurity
+  - email-security
+  - dmarc
+  - spf
+  - email-authentication
+  - technographics
+  - technology-adoption
+  - company-data
+  - phishing
+  - business
+task_categories:
+  - tabular-classification
+language:
+  - en
+size_categories:
+  - 10K<n<100K
+dataset_info:
+  features:
+    - name: Popularity Rank
+      dtype: int64
+    - name: Domain
+      dtype: string
+    - name: Company Name
+      dtype: string
+    - name: Company Size
+      dtype: string
+    - name: Year Founded
+      dtype: float64
+    - name: Industry
+      dtype: string
+    - name: LinkedIn URL
+      dtype: string
+    - name: Locality
+      dtype: string
+    - name: Country
+      dtype: string
+    - name: TLD
+      dtype: string
+    - name: Language
+      dtype: string
+    - name: Instagram
+      dtype: string
+    - name: Twitter
+      dtype: string
+    - name: YouTube
+      dtype: string
+    - name: TikTok
+      dtype: string
+    - name: SPF
+      dtype: string
+    - name: DMARC
+      dtype: string
+    - name: BIMI
+      dtype: string
+    - name: MTA-STS
+      dtype: string
+    - name: Posture
+      dtype: string
+  splits:
+    - name: train
+      num_examples: 36120
+---
+
+# DMARC and SPF Adoption Among Large Organizations
+
+## Overview
+
+This dataset records which of 36,120 large organizations publish SPF and DMARC records on their primary domain, with firmographic context for each: industry, employee band, country, locality, founding year and LinkedIn profile.
+
+SPF lists the servers allowed to send mail for a domain. DMARC tells receiving servers what to do with mail that fails that check, and where to send reports. A domain with SPF but no DMARC has published the list without publishing an instruction, so receivers are left to decide for themselves.
+
+Collected in August 2026. DNS records change, so treat the data as a snapshot of that month rather than a current state.
+
+Full statistics are published at StackScan for [DMARC](https://www.stackscan.com/technology/dmarc) and [SPF](https://www.stackscan.com/technology/spf-sender-policy-framework).
+
+## SPF and DMARC Coverage
+
+Of the 35,563 large organizations that publish SPF, **45.7% also publish DMARC**. The other 19,315 publish SPF alone.
+
+| Posture | Organizations |
+| ------- | ------------- |
+| SPF only | 19,315 |
+| SPF + DMARC | 16,248 |
+| DMARC without SPF | 557 |
+
+SPF is close to universal among organizations that send mail, so the SPF group serves as a denominator and the percentage above is a rate rather than a raw count.
+
+## Adoption by Country and Industry
+
+DMARC adoption among SPF publishers, by country. Groups with fewer than 150 organizations are omitted.
+
+| Country | Organizations | DMARC |
+| ------- | ------------- | ----- |
+| Sweden | 305 | 59.7% |
+| Argentina | 219 | 56.6% |
+| Denmark | 188 | 56.4% |
+| Belgium | 305 | 53.4% |
+| United Kingdom | 1,733 | 51.9% |
+| Australia | 596 | 51.8% |
+| South Korea | 199 | 34.2% |
+| Vietnam | 268 | 34.3% |
+| China | 1,456 | 30.2% |
+
+By industry, same threshold:
+
+| Industry | Organizations | DMARC |
+| -------- | ------------- | ----- |
+| defense & space | 168 | 56.5% |
+| medical devices | 197 | 55.8% |
+| banking | 650 | 55.5% |
+| airlines/aviation | 252 | 54.0% |
+| insurance | 638 | 53.4% |
+| security and investigations | 318 | 35.5% |
+| professional training & coaching | 150 | 28.7% |
+
+Employee band makes little difference. Organizations with 5,001 to 10,000 employees reach 46.7%, and those with more than 10,000 reach 44.9%.
+
+BIMI and MTA-STS are rare in this population. Across all 36,120 organizations, 17 publish BIMI and 5 publish MTA-STS.
+
+## Dataset Fields
+
+`Email-Authentication.csv`, 36,120 rows, 20 columns. Coverage is the share of rows where the field is populated.
+
+| Column | Type | Coverage | Description |
+| ------ | ---- | -------- | ----------- |
+| `Popularity Rank` | int64 | 95.1% | Website popularity rank, lower being more popular. Rows are stored in ascending rank order |
+| `Domain` | string | 100% | Primary website domain. Unique across the file |
+| `Company Name` | string | 100% | Organization name, lowercased |
+| `Company Size` | string | 100% | Employee band. Two values: `5001-10000` and `10001+` |
+| `Year Founded` | float64 | 71.2% | Founding year |
+| `Industry` | string | 99.1% | Industry label, lowercased |
+| `LinkedIn URL` | string | 100% | Organization LinkedIn profile path |
+| `Locality` | string | 79.7% | City or region, where available |
+| `Country` | string | 100% | Organization or website country |
+| `TLD` | string | 100% | Top-level domain |
+| `Language` | string | sparse | Detected site language |
+| `Instagram` | string | sparse | Instagram handle |
+| `Twitter` | string | sparse | Twitter handle |
+| `YouTube` | string | sparse | YouTube channel |
+| `TikTok` | string | sparse | TikTok handle |
+| `SPF` | string | 100% | `yes` or `no` |
+| `DMARC` | string | 100% | `yes` or `no` |
+| `BIMI` | string | 100% | `yes` or `no` |
+| `MTA-STS` | string | 100% | `yes` or `no` |
+| `Posture` | string | 100% | `SPF only`, `SPF + DMARC` or `DMARC without SPF` |
+
+The language and social columns are populated for fewer than 1% of rows.
+
+## Methodology
+
+Domains were checked for published SPF, DMARC, BIMI and MTA-STS records in August 2026. An organization is included when at least one of SPF or DMARC was found on its primary domain.
+
+The population is large organizations, meaning 5,001 or more employees, for which company information could be resolved.
+
+## Important Notes
+
+* `DMARC` means a DMARC record was found. It does not mean the policy is enforced. A policy of `p=none` requests monitoring only and is common, and the policy value is not recorded here, so `DMARC = yes` should not be read as protection.
+* DKIM is not included. Detecting a DKIM record requires knowing the DNS selector in advance, so any count would understate adoption by orders of magnitude.
+* The 557 organizations with DMARC but no SPF are unusual and often misconfigured. DMARC relies on SPF or DKIM alignment, so a DMARC record with neither does little.
+* Records are read from the primary domain only. Large organizations send from many domains and subdomains, and those are not covered.
+* Rates are calculated against organizations publishing SPF, not against all large organizations.
+* Results represent observations made in August 2026.
+
+## Use Cases
+
+* Email security and anti-phishing research
+* Benchmarking an organization or sector against its peers
+* Public sector and regulatory policy analysis
+* Journalism on email security preparedness
+* Technographic and market research
+* Data science and machine learning
+
+## Citation
+
+If you use this dataset, please credit StackScan and link to the source page.
+
+Plain text:
+
+StackScan. (2026). *DMARC and SPF Adoption Among Large Organizations* (data collected August 2026) [Data set]. StackScan. https://www.stackscan.com/technology/dmarc
+
+BibTeX:
+
+```bibtex
+@misc{stackscan_email_authentication_2026,
+  author       = {{StackScan}},
+  title        = {DMARC and SPF Adoption Among Large Organizations},
+  year         = {2026},
+  month        = {aug},
+  publisher    = {StackScan},
+  howpublished = {\url{https://www.stackscan.com/technology/dmarc}},
+  note         = {Data collected August 2026. Dataset: \url{https://huggingface.co/datasets/stackscan/email-authentication}}
+}
+```
+
+## Related
+
+[stackscan/lets-encrypt](https://huggingface.co/datasets/stackscan/lets-encrypt) uses the same population definition and collection period, so the two can be joined on `Domain`.
+
+## Access to More Data
+
+The files in this repository are CC BY 4.0 and free for anyone to use, including commercially. No request or permission is needed.
+
+StackScan also provides free access to larger and custom datasets for data journalists, academic researchers, non-profits and other organizations working on non-commercial projects. Ask through [stackscan.com](https://www.stackscan.com/technology/dmarc).
+
+## License
+
+Released under [Creative Commons Attribution 4.0 International (CC BY 4.0)](https://creativecommons.org/licenses/by/4.0/). You are free to share and adapt the data, including commercially, provided you give appropriate credit.
+
+Attribution must name StackScan and link to the source page:
+
+> Data from [StackScan](https://www.stackscan.com/technology/dmarc), licensed under CC BY 4.0.
+
+For academic or published work, use the citation above.