Skip to content
Giacomo Ravetta · PoliticsItaliano
Illustration: Deciding with data: open data for the Municipality

Proposal · MethodDifficulty easyReading time: 8 min

Tools the Municipality already has and a free regional platform: it takes a procedure, not a purchase.

Deciding with data, and leaving it open

A municipality produces data every day: budgets, payments, contracts, building permits, enrolments in services, waste collected, road accidents. Today in Saronno that data reaches citizens, when it does, as documents to read. The proposal is to publish it also as data to use, in the formats and under the licences the law already provides for, and to use it to justify decisions.

The idea in brief

  • Today the Municipality publishes documents, often as PDF: they can be read, not reused. The law requires data in open formats, and a PDF is worth one star out of five.
  • Every table that goes on the transparency portal also comes out as CSV under a Creative Commons licence, on the free platform of the Lombardy Region, which passes it on to the national catalogue.
  • A working group led by the Digital Transition Officer takes stock of the data the offices already produce: the cost is a procedure, not a purchase.
  • Every resolution cites the dataset its numbers come from: open data thus becomes a way of deciding, not just an obligation.

The starting point

Documents yes, data no

The transparency portal of the Municipality of Saronno fulfils the obligations of Legislative Decree 33/2013: the documents are there and can be downloaded. But the contents are almost always PDF files, that is, photographs of a table. Thepayment timeliness indicator, a table by definition, is published quarter by quarter as an attachment to download; theannual accountsare the resolution with its accounting annexes. The portal offers an “open format list” as CSV, but it exports the index of documents, not the numbers they contain. To compare two years of spending or add up the grants paid to one sector you have to copy the values by hand. Those who do make mistakes; those who lack the time give up.

That it can be done differently is shown by the same portal: the list ofconsultants and collaborators is published on the page and downloadable as CSV with the real data inside. And the2025-2027 budget has, next to the full 65-page document, two statements also published in OpenDocument format, an open spreadsheet: revenue by title and type, expenditure by mission and macro-aggregate. We compared PDF and spreadsheet figure by figure: the 142 revenue amounts and the 6,006 expenditure amounts match, starting from the 2025 total of €84,715,633. But the PDF of those two statements can be read only by eye, because the font it is printed with carries no mapping to letters and a program extracting its text gets meaningless symbols; and the rest of the budget, some twenty statements including summaries, the overall balance sheet, budget balances and borrowing limits, exists only as PDF. These are partial exceptions, and they should be the rule.

In the national catalogue dati.gov.it and the regional portal dati.lombardia.it, queried in September 2026, no dataset owned by the Municipality of Saronno appears. The data about Saronno found there is what the Region and Istat produce about every municipality. The Municipality itself has not opened anything yet.

This is not a technical detail. An administration that does not keep its own data in a processable form struggles to use it internally too: every report starts from scratch, every comparison between offices goes through a spreadsheet rebuilt by hand. Opening data to the outside is the simplest way of forcing oneself to keep it in order inside.

The yardstick

Why a PDF is not enough

The 2023 AgID guidelines on opening data adopt Tim Berners-Lee’s five-star scale. It is the yardstick with which the law measures how open a piece of data really is:

  1. One star. The data is online under an open licence, but in an unstructured format. That is the PDF.
  2. Two stars. Structured but in a proprietary format, such as an Excel sheet: it takes specific software and often manual work.
  3. Three stars. Structured in a non-proprietary format: CSV for tables, JSON for nested data, GeoJSON for what sits on a map. It is the minimum requirement set by the guidelines.
  4. Four stars. Every row has a stable identifier (URI) and uses semantic web standards.
  5. Five stars. The data is linked to other data: the Istat code of the municipality points to the municipality, the supplier’s tax code to the supplier.

The Digital Administration Code defines an “open format” as one that is public, exhaustively documented and neutral with respect to the tools needed to read it, and “open data” as data available under a licence that lets anyone use it, including commercially, in disaggregated form, suitable for automatic processing and accompanied by metadata. A PDF meets neither of the last two requirements. The guidelines are explicit: publishing at one or two stars only is no longer allowed, unless accompanied by level three.

The PDF stays. Nobody proposes removing signed documents: they serve those who must read the act and have legal value. The proposal is that, next to the PDF, the same table also exists as a CSV file. They are two different uses of the same data.

The proposal

Seven steps, in order

  1. Take stock

    Every office lists the data it already produces and which system holds it: accounting, taxes, registry, building, social services, schools, environment, local police. No new data is collected: what exists is written down. The AgID three-year plan asks exactly this as a first step.

  2. Publish tables as CSV too

    Everything that goes on the transparency portal as a table also comes out in an open format, with metadata in the national DCAT-AP_IT profile: who produced it, when, how often it is updated, under which licence. The systems the Municipality already uses export CSV: the cost is a procedure, not a purchase.

  3. One open licence, the same for everything

    The AgID guidelines recommend the Creative Commons Attribution 4.0 licence, with CC0 as an alternative. The Digital Administration Code adds that data published without an explicit licence is deemed open anyway: one might as well say so, and remove any doubt for those who want to reuse it.

  4. Lean on the Lombardy Region

    There is no need to build a portal. The Lombardy Region offers municipalities its own open data platform, free of charge, and has defined a basket of fifty standard datasets in ten categories, already modelled, that a municipality can publish as they are. From there the data flows automatically into the national catalogue dati.gov.it, which harvests the catalogues of public bodies through a link communicated once.

  5. A working group with someone in charge

    The three-year plan asks the Digital Transition Officer, a mandatory role in every municipality, to coordinate a group with one contact per area. The plan also sets measurable targets: a municipality publishing fewer than ten datasets must reach at least ten. Saronno starts from zero.

  6. Every resolution cites its data

    This is the step that turns open data into a method of government. The recitals of a resolution about traffic, waste, enrolments or spending name the dataset the numbers come from, with its update date. Readers can check; later decision-makers can compare. If the data does not exist, the resolution says so, and that becomes the next dataset to build.

  7. Start with high-value data

    European Regulation 2023/138 identifies six families of data that take precedence because their reuse is worth more: geospatial, earth observation and environment, meteorological, statistics, companies, mobility. For a municipality this means, concretely: maps and the land-use plan, environmental sensors, mobility and parking data, commerce.

From the point of view of a resident

What you do with it

This very site is an example of what happens when data is open. The solar roofs proposal was built with public, reusable data: solar radiation from the European Joint Research Centre, building footprints from OpenStreetMap, municipal boundaries from Istat, the list of municipal buildings taken from the transparency portal. The last was the only one in PDF, and the only one that had to be copied by hand.

With the Municipality’s data in open formats, a local newspaper can follow maintenance spending street by street; a neighbourhood committee can check how many nursery applications are left without a place; a school can use waste data for a civic education project; a business can build a service without asking permission. And the town council debates numbers everyone saw before the session.

What it takes

Less than it seems

The hard part is neither technical nor financial: it is already written into the law, and the Municipality is bound to apply it.

  • The Digital Administration Code(art. 52) establishes that data published without a licence is deemed open and that opening data counts in the appraisal of managers.
  • The transparency decree (Legislative Decree 33/2013, art. 7) requires what must be published by law to be in an open, reusable format, with the sole obligation of citing the source.
  • Legislative Decree 36/2006, rewritten in 2021 to transpose the European open data directive, regulates reuse and refers to the AgID guidelines for formats and licences.
  • The platform is the regional one, with no licence costs. The work is an export procedure for each system and a few days of training.
  • A policy act of the Council or the Executive adopting these rules, setting the publication calendar and requiring resolutions to cite their data.

Honesty

Where it can go wrong

  • Personal data. Opening data does not mean opening everything. Tables must be aggregated or anonymised before publication and data about people stays out: the transparency decree already says so, and the five-star scale does not change it.
  • Maintenance. A dataset stuck three years ago is worse than no dataset, because it misleads. Cities with rich portals let them decay when the person looking after them left. That is why step five, the working group with someone in charge, is not optional.
  • Publishing is not deciding. Open data does not produce better decisions by itself. It produces better discussions, if someone uses it. Step six, resolutions that cite their data, is what makes the difference between compliance and a method.
  • Quality at the source. Exporting a badly kept table to CSV exposes its errors. That is a real initial cost, and also the reason it is worth doing.

Useful links