Method
What is measured
For each municipality we query three public DNS records of its mail domain: MX (does the domain receive mail?), SPF and DMARC. We query public DNS resolvers only. No mail is sent and no municipal server is contacted by us.
Sources
Municipality list: the FSO’s official register of municipalities. Websites: Wikidata. The mail domain comes from a checked list: usually the website’s domain; where the municipality demonstrably publishes its address under another domain, that one, with the evidence. If a municipality uses several domains (say, for mail and for the website), the least protected one counts, since each can be used to fake mail in its name. The municipality page shows each domain and which one determines the rating. Boundaries: swisstopo swissBOUNDARIES3D.
Queries go through Google Public DNS and Quad9 at a limited rate, and both answers are compared. Without a record at _dmarc.domain, the organizational domain’s record applies (e.g. ne.ch for commune.ne.ch), with its sp= if set.
Classification
| Reject policy | p=reject, with no pct below 100 and no t=y |
|---|---|
| Quarantine policy | p=quarantine, with no pct below 100 and no t=y |
| Monitoring only | p=none with a report address (rua), or a policy with pct below 100 or t=y |
| No effective DMARC | no DMARC record, an invalid one (e.g. several records), or p=none without a report address: such a record asks for no action and nobody receives reports |
| No data | no known website, the domain does not exist, or a DNS error. Domains without MX are rated by their DMARC record like all others. |
Classifier v4
Limits
For about 550 municipalities the mail domain is confirmed on the official website, including almost all poorly rated municipalities in the weakest cantons; for the rest, the website’s domain is used. In a random sample of 100 municipalities, 3 were wrong (since corrected), so a few dozen assignments may still be off. Corrections are welcome. DKIM is not checked, because the selectors are unknown. The measurement is a snapshot of DNS records and says nothing about how individual messages are actually handled. Every municipality counts the same, whatever its population.
A DMARC policy is a request to receiving servers. Whether a given message is rejected depends on the receiver and on SPF/DKIM alignment, which this snapshot does not check. The rating depends only on the DNS records, not on any tool.
Release
All data, including that for individual municipalities, comes from public DNS records that anyone can query. Each municipality has its own page with its domains and their rating.
Corrections
Mail domain assigned wrongly, or a record changed since? Write to research@kaitiaki.tech or report it in the public issues on GitHub. Emails are not published.
We answer within 5 working days. Confirmed corrections are rescanned and published within 10 working days.
Every correction appears in the correction log below, with date, municipality and what changed, without the reporter’s name.
Correction log
No corrections reported so far.
Frequent questions
- Why is a domain that receives no mail rated?
- A domain without a mailbox can still be used as a forged sender. A DMARC record with
p=rejectis how its owner says “no mail comes from here”. Without it, receiving servers have to guess. - Why does the least protected domain count?
- Each domain of a municipality can be forged on its own. One unprotected website domain is enough for a convincing fake. The municipality’s page shows which domain decided the rating.
- Our IT provider handles our mail. Why are we listed?
- The record is published under the municipality’s domain, whoever operates it. Your provider can change it in a few minutes; the guide on this site describes the steps.
- We have a DMARC record. Why “no effective DMARC”?
- A
p=nonerecord without a report address (rua) asks for no action, and nobody receives reports. Add anruaaddress and the domain counts as “Monitoring only”; withp=quarantineorp=rejectthe policy is enforced.