Summary

  • An original study mapped the public-facing web, mail and DNS infrastructure of European local authorities. Its base contains 95,498 local administrative units; 69,220 have website domains in the linked records. The map reveals a particularly wide gap between where municipal email servers sit and where their operators are registered.
  • Across countries, the estimated EU/EEA location share for inbound municipal mail exceeds the EU/EEA registered-operator share by an average of 39.7 percentage points. A data centre can give city hall a local-looking address while the service contract, operating company and recovery choices sit elsewhere.

The website is at home. The mailroom took a train.

Imagine the mayor proudly unveiling a new municipal website, hosted nearby, with a reassuring flag in the procurement slides. Then a resident emails the council about a benefit application. That message enters a different infrastructure market altogether. The website and the mailbox may belong to different suppliers, different networks and different jurisdictions. A local homepage makes an excellent ribbon-cutting photograph. It tells the public almost nothing about who can restore the mailroom after an outage.

The original Municipalytics paper gives this difference a measurable shape. Its researchers started from Eurostat's Local Administrative Units, the shared administrative geography used across the European statistical system. Their base spans 95,498 units in the EU's 27 states plus Iceland, Liechtenstein and Norway. They linked 88,948 of those units to Wikidata entries; 69,220 had website domains in that linked layer. The last number matters. A chart of 95,498 places is not a direct measurement of 95,498 working town-hall websites.

The team resolved web addresses, mail exchangers and authoritative name servers, then assigned the visible IP addresses to estimated hosting countries and autonomous-system operators. Its public map lets a reader explore the resulting observations. This is a view of the Internet-facing machinery: where a domain points, where a mail exchanger appears to run, and which organization is associated with the network. That is enough to puncture a slogan. It is not a procurement contract or a court file.

On the paper's unweighted average across countries, 60.1% of municipal websites, 51.5% of inbound mail and 65.7% of authoritative DNS were hosted domestically. The three services take separate routes to the public. Country averages give a small state and a large state equal weight; weighting by municipal population raises the corresponding domestic shares to 76.7%, 66.6% and 76.3%. A minister can choose the friendlier column for a speech. A service owner has to know which column describes the people and systems it is actually responsible for.

The most revealing comparison is within email. For inbound mail, the share of infrastructure estimated to be physically in the EU or EEA exceeded the share attributed to EU/EEA-registered operators by an average 39.7 percentage points across countries. The comparable gap was 12.4 points for websites and 1.9 for authoritative DNS. The researchers also found a familiar pattern across much of northern Europe: many municipal websites remain domestically hosted while mail has consolidated around a large external provider. Their figures put that provider's mail service at 62% to 90% of municipalities in the cited countries and territories.

The town hall can keep its own front door and rent out the letterbox.

There is a practical reason. Shared mail is difficult to run well. Spam defence, continuity, account security and support favour scale. A council buys a service that works on Monday morning. Yet the institution cannot discharge its public duty by pointing to the server's postcode. It needs to know which legal entity operates the service, who can change its settings, what happens to archives during a dispute, how residents are notified when mail fails, and whether an exit plan can be executed before the next emergency. The paper measures the first part of that chain. The remaining links belong in public procurement and continuity records.

The study's limitations make the commissioning question sharper. Domain names came through Wikidata, whose entries can be absent or stale. A local administrative unit is not the same kind of municipality in every country; Portugal and Cyprus have sub-municipal units, while Greece and Ireland require parent-level handling. The measurement is a snapshot. Geolocation estimates can point to a proxy rather than the origin service, and a domain can resolve to multiple operators. An SPF record authorizes senders; it does not show who handled every outgoing message.

These are named blind spots, each with a testable repair: validate the municipal domain list, date the DNS observations and obtain the supplier's service and exit documents.

The distinction between location and authority is the real story. A server inside the EU can be operated by an organization registered elsewhere. A server outside the country can be part of a service whose contract provides robust local continuity and review. The original paper does not adjudicate either contract. It exposes the question that every town hall ought to be able to answer without a research grant: who owns the operational decision when the service goes dark?

The data-centre postcode is a poor emergency contact

Digital sovereignty is often sold as a map pin. A municipal service needs a runbook. The map pin can help locate a dependency, but the runbook names the supplier, the lawful decision maker, the recovery path and the evidence that the arrangement works. Municipalytics has placed the mailroom on a map. The next public record should put a person and a deadline beside its door.