Service · App development

Custom open data app development.

A custom app that lets you visualise and compare open data from government, EU and scientific sources, and make it interactively available to your users. For data journalism, policy monitoring, citizen tools, replicable research and business intelligence on public datasets. Built on transparent pipelines, with code ownership and no vendor lock-in.

PDOK & CBSEurostat & OECDOpenStreetMapINSPIRE-compliant

Open data is everywhere — making it usable is the real work.

Dutch and European governments publish an impressive stream of open datasets: PDOK for geographic data, CBS (Statistics Netherlands) for statistics, Open Spending for government finances, Eurostat for European figures, the EU Open Data Portal for everything the Commission publishes, INSPIRE for spatial information and the OECD for international comparisons. There are also community-driven sources such as OpenStreetMap, Wikidata and Common Crawl. Each dataset is technically accessible on its own — but for a working app you need more than the download link.

The gap isn't in the data, but in what happens between the data and the user. A CBS table with 87 columns and code numbers instead of labels is literally publicly available, but practically unreadable for a council member. A PDOK WMS delivers raster tiles in the RD New projection, which first have to be transformed for a mobile map. Between "this data is public" and "a citizen, journalist or policymaker can do something with it" lies a lot of engineering work that nobody does for you.

We build custom open-data apps for organisations that want to bridge that gap: a data-journalism team, a research institute that wants to distribute replicable research, a policy organisation that wants to show air quality or the energy mix in real time, a municipality that wants to turn its open data into a citizen tool, an NGO that wants to open up government spending. The work is always the same: scraping, structuring, normalising, linking, validating and presenting, in an interface that suits the people using it.

Open data touches several disciplines within our practice. It overlaps with our broader app development, with our experience of geographic apps for spatial open data, with our data engineering platforms as the underlying ingestion layer, with our BI and dashboard tooling for the visual side, and with our work on software for municipalities. A good open-data app draws on elements from all of these areas.

Three common forms.

Which form suits your organisation depends on your audience (journalists, policymakers, citizens, researchers, internal users), the nature of the source data (static, periodic, real-time) and how much interaction you want to offer (reading, comparing, downloading, taking action). In the first conversation we advise which variant will deliver the most.

Compact project · fixed sprint budget

Data journalism and citizen tool app

An interactive mobile or web app that makes open data from CBS, Eurostat or a municipality understandable to a broad audience. Think of a home-search app based on BAG, a local air-quality tool using RIVM data, a comparison tool for Dutch municipalities based on CBS figures, or an interactive, browsable version of a research report. Always with an accessible interface, in your house style, and with a data structure that stays correct after the next dataset update.

CBS & PDOKAccessible (WCAG)Mobile-firstEmbeddable
Mid-sized project · fixed sprint budget

Policy monitoring and real-time dashboard app

For government bodies, environmental agencies, NGOs or stakeholder platforms that want to show current public data: air quality, energy mix, traffic flows, water levels, climate indicators, municipal spending, EU funds. The app continuously pulls data from official APIs (Luchtmeetnet, ENTSO-E, NDW, Open Spending, Eurostat), validates and harmonises it, and presents the results in a dashboard that both administrators and interested citizens can follow. With alerts when values cross thresholds. For the heavier data layer beneath such an app, we often draw on our data engineering platform practice.

Real-time APIsAlertsComparisonsOpen Spending
Larger project · fixed sprint budget

Research and BI platform on open data

For researchers, policy economists, journalists or analysts who want to combine open data with their own data on a structural basis, for replicable research, longitudinal studies, sector comparisons or pre-deal due diligence. A platform with a query layer, visualisation layer and publication flow, built on a dedicated data warehouse into which OECD, Eurostat, CBS, INSPIRE and Common Crawl data arrive via version-controlled pipelines. With code notebooks alongside charts, transparent methodology, and exportable dataset snapshots so that others can reproduce the work. The overlap with our BI tool practice is considerable.

DuckDB & ParquetReproducibleFRBR attributionNotebooks

What an open-data app can do as standard.

The functional layers that recur in almost every open-data project, whether you have a journalistic tool, a citizen app or a policy dashboard in mind. In each case we adapt this foundation to your sources, your audience and your existing infrastructure.

  • Multi-source ingestionPipelines to PDOK, CBS StatLine, Open Spending, Eurostat, OECD.Stat, the EU Open Data Portal, INSPIRE, OpenStreetMap, Wikidata, Common Crawl and sector-specific sources such as Luchtmeetnet, ENTSO-E, NDW and KvK Open Data. We set up the right authentication, rate limiting and retry logic for each source. We catch changes to a source API with version checks, so your app doesn't quietly stop working.
  • Processing vector, raster and tabular dataNative support for GeoJSON, GeoPackage, shapefile, GeoTIFF and CSV with coordinates for spatial data; JSON-stat, SDMX, CSV and Parquet for statistical data; RDF and SPARQL for linked open data from Wikidata or the Kennisbasis. Heavy datasets are split up via vector tiles or a DuckDB layer, so the app stays smooth on a phone and on slower office networks.
  • Versioned data snapshotsOpen data changes: a CBS table gets a revision, a PDOK dataset is classified differently, a Eurostat series is withdrawn. We store every ingestion run as an immutable snapshot, with checksum and timestamp, so a report from last quarter is still reproducible. This is especially critical for research, journalism and legal cases, where the claim "these figures were correct at the time" has to hold up.
  • Normalisation and harmonisation layerOpen data rarely arrives in a format suitable for direct presentation. CBS uses its own code tables, Eurostat uses NUTS regions, the OECD uses ISO country codes, and INSPIRE has its own vocabularies. We map all codes to a single internal model, provide labels in Dutch and English, and harmonise time dimensions (years, quarters, weeks) so that data from different sources is genuinely comparable. Mapping properly once saves hundreds of ad-hoc lookups later.
  • Visualisation and interaction layerCharts built with Plotly, Observable Plot, ECharts or D3, depending on your look and feel and interaction needs. Vector maps via Mapbox GL, MapLibre or Leaflet on top of OpenStreetMap or PDOK tiles. Tables with filtering, sorting, group-by, drill-down and export. Users can save subsets, share them via URL or embed them in articles, dashboards or policy documents, without having to download the underlying data.
  • Data attribution and source trailFor open data apps, traceability of the source is not a nice-to-have but a prerequisite, for journalistic integrity, for scientific citability and for the Open Government Licence. Every visualisation and every export states the exact source, the ingestion time, the licence terms and any methodological footnotes. We work with an FRBR-compliant attribution model, so users always know which version of which dataset sits behind a chart.
  • GDPR-compliant handling of open dataOpen data sounds harmless, but it regularly contains personal data or data that becomes identifiable in combination. An open BAG address combined with a municipal permit database can identify individuals; CBS data at small-neighbourhood level sometimes involves statistical confidentiality. We build disclosure control and aggregation thresholds directly into the app, with logging of what data was provided, to whom and at what level of aggregation, in line with the GDPR and the Dutch Open Government Act (Woo).
  • WCAG and EAA accessibilityMandatory for public-sector and citizen-facing apps under the European Accessibility Act (EAA) and the Dutch Digital Accessibility (Government) Decree. We build visualisations with alternative text presentations, a high-contrast mode, screen-reader-friendly charts and keyboard-only navigation. For data journalism embeds, we also test against the accessibility standards of the publisher you publish on.
  • API gateway and public data layerYou will often want to expose a structured API of your own too: for your research network, for other journalists, for partner organisations or for the civic-tech community. We build a gateway with rate limiting, API keys, OpenAPI documentation and a data catalogue, so that others can build on your work the way you built on the government data.
  • Embed and sharing flowsOpen-data apps really come alive when their visualisations end up in articles, policy memos and social feeds. We build embed blocks with a responsive layout, a static fallback for anyone who blocks JavaScript, OpenGraph cards for sharing on LinkedIn and messengers, and version-stable URLs that don't break when you develop the app further. We deliberately don't use embeds for the platform formerly known as Twitter/X.
  • Maintenance contract (optional)Ongoing maintenance, monitoring of source APIs and further development. Open-data apps need maintenance more often than regular apps because the sources keep changing, whether it's a revised CBS table or a new NUTS level. We keep track of what's happening at the source owners and adapt the pipelines before your app breaks.

When an open-data app is the right choice.

Not every project involving public data needs its own app. Sometimes a notebook, a Google Sheet or a Tableau dashboard is enough. Here are six signs that custom open-data software is genuinely worth it.

Audience

Your users aren't data experts

Your audience is a journalist, a councillor, a resident or a generally interested reader, not an analyst with R, Python or Power BI in the background. A raw CBS table or a Eurostat API is unusable to them; they need an interface that presents the data in everyday language and in terms of everyday questions.

Recurring

The data changes periodically

You don't publish a report once; you want the insights to update automatically with every new dataset release. A static infographic would have to be recreated every month or year, whereas an app pulls the latest version and recalculates.

Combination

Multiple sources need to be combined

Your insight doesn't come from a single dataset but from putting CBS, Eurostat, OECD and your own data side by side. Each source on its own shows only half the story. An app harmonises the sources and lets users filter on the common denominator instead of working out joins and lookups themselves.

Reproducible

Your methodology must be demonstrable

For science, journalism with legal consequences or policy documents that must be defensible before the Dutch House of Representatives or the European Parliament, every figure has to be reproducible: which source, which selection, which calculation, which point in time. An app with versioned snapshots and a transparent pipeline is then preferable to a notebook or a spreadsheet.

Real-time

Need continuously up-to-date information

You need continuously up-to-date information

Distribution

Air quality, energy mix, traffic conditions, water levels or public transport delays: if users want to know the value right now, a report or a manually updated dashboard can't deliver that. A live app queries the source APIs at the moment the user looks and sends alerts when thresholds are crossed.

You want fellow researchers, journalists or citizens to be able to work with your analysis: not just see the final picture, but also work with subsets, filters, downloads and API calls. A custom app with an embed and API layer is then a more natural fit than a PDF report or a Tableau Public page.

Not sure about a large project yet?

Test your idea first: a working prototype in 1 day

With OneDayBuild, we make your idea tangible in a single day for €1,150, so you know whether further development is worth the investment. Decide to go ahead with the full build? Then we deduct the cost in full.

View OneDayBuild →

Reach

1

The insight needs to be widely shared

A conversation to pin down the intended use: who will look at the app, with what questions, and which decision follows. We inventory the open-data sources you already use or would like to use — PDOK, CBS, Eurostat, OECD, Open Spending, INSPIRE, OSM — and check licence terms, refresh frequency and the stability of the source API. At the end you have an overview of the sources that are worth it and the sources that are too unstable for production.

2

Introduction and source inventory

We map out the ingestion pipeline: for each source, a fetch frequency, a normalisation step and a target model. We design an internal data schema that holds together across sources, for example a shared region table in which CBS municipalities, Eurostat NUTS and INSPIRE codes sit side by side. Before we start, we test the schema with your data team and two to three end users using interactive wireframes.

3

Building in sprints

Every sprint, we deliver a working version on a staging environment. We start with the ingestion and normalisation layer plus one primary visualisation, such as a map or a comparison table, so you can test with real data as early as possible. We then build out the remaining views, embeds, alerts and API layer. You test along, end users test along, and we keep adjusting as we go. We review the scope at every sprint so that further development lands in the right place.

4

Validation and accessibility audit

Before go-live, we validate the figures in the app against the original source. This is an important step, because with open data an error in normalisation or in a join can carry through into publications for a long time. We carry out a WCAG and EAA audit for public and citizen-facing apps, a GDPR and disclosure-control review for data that comes close to personal data, and a security review of the API layer and the embed mechanism.

5

Rollout and maintenance

A phased rollout to your audience or stakeholders, with monitoring of source APIs and data pipelines. We alert you when a CBS table changes, a Eurostat series is replaced or a PDOK layer gets a new version. Further development runs through the maintenance contract. Many open-data apps grow organically when a new source becomes interesting or a new audience appears, and maintenance keeps that smooth without you having to start a new project each time.

Fabian van Dijk Business developer · Appfront

Frequently asked questions.

The questions clients ask when they are considering an open-data app.

Which open-data sources can you connect?
In practice, we work most with PDOK for geographic data, CBS StatLine and the CBS Open Data Portal for statistics, Open Spending for government expenditure, Eurostat and OECD.Stat for European and international comparisons, the EU Open Data Portal for everything the Commission publishes, INSPIRE for standardised spatial data, OpenStreetMap for map content and geocoding, Wikidata for linked open data, and sector-specific sources such as Luchtmeetnet, ENTSO-E, NDW, KvK Open Data, BAG and BRO. For research projects, also Common Crawl and the Internet Archive. If you have a specific source in mind that is not listed here, almost any open API with a licence can be integrated, and we will discuss that in the first conversation.
How stable are open-data APIs, really?
Honestly, it varies. PDOK, CBS Open Data and Eurostat are production-grade stable, with versioning and clear change logs. The EU Open Data Portal and some municipal open-data portals are less predictable: datasets can be moved, renamed or quietly withdrawn. That is why we always build an ingestion layer with validation, version pinning and our own snapshot archive. If a source API goes down temporarily or makes a major change, your app keeps running on the last valid snapshot while we adapt the pipeline.
What about personal data in open data?
Open data is not automatically free of personal data. An open BAG address combined with an open WOZ value and an open permits register can identify a specific household. CBS data at small-neighbourhood level can lead to individuals through statistical triangulation. For every project we run a GDPR pre-assessment: which combinations of sources could lead to re-identification, which aggregation thresholds are appropriate, and which disclosure controls we apply. For projects with serious privacy implications, we carry out a formal DPIA together with your data protection officer or privacy officer. For government data, the Open Government Act (Woo) and the re-use regime under the Open Data Directive are also relevant.
What is INSPIRE, and do we need to comply with it?
INSPIRE is a European directive that requires public-sector organisations to publish certain spatial datasets in a standardised format with standardised metadata. If you want to offer data yourself in addition to what you consume — for example, a citizen tool that enriches government data and exposes it again as an open service — INSPIRE may apply to you. We provide metadata in ISO 19115 format, GML or GeoJSON export, and an integration with the Nationaal GeoRegister (the Dutch national geo-register). INSPIRE usually doesn't apply to private and research parties, but interoperability with INSPIRE-compliant datasets is often still desirable. For the overlap with geospatial questions, see our page on geographic apps.
Which licence terms apply to open data?
Most Dutch open data falls under CC0, CC-BY-4.0 or the Open Government Licence. CBS data is generally CC-BY with an attribution requirement, PDOK varies per dataset, and Eurostat has its own open-data licence that is comparable in practice. OpenStreetMap data falls under the ODbL, which carries share-alike obligations for derived datasets. We build attribution and licence notices directly into the app, with the correct source credit for each visualisation and, where the licence requires it, a downloadable data export with the accompanying licence text. That saves a lot of work later in legal reviews.
How do you make sure our app is accessibility-compliant?
We build in WCAG 2.2 AA from the design stage, not as an afterthought. For public-sector and citizen-facing apps this is mandatory under the Dutch Temporary Decree on Digital Accessibility of Government (Tijdelijk besluit digitale toegankelijkheid overheid) and, since 2025, under the European Accessibility Act (EAA), which also covers many private providers. Concrete measures include: alternative text presentations of visualisations (a map also gets a structured data-table mode), high contrast and dark mode, keyboard-only navigation, screen-reader-friendly charts via ARIA attributes, avoiding colour-only encoding, and captions for any video content. In the final sprint we run an audit with an automated tool plus a manual check with a screen reader.
Do you host the app, or do we do that ourselves?
Either works. For public-sector and research organisations we often host in a European cloud (Google Cloud Frankfurt, AWS Frankfurt, or a local provider). If you have your own cloud tenant, we deploy to your GCP, AWS or Azure environment. The codebase is the same; only the deployment target differs. You are not locked into our hosting, and handing over hosting is part of the final delivery.
Which tech stack do you usually use?
For the app: Flutter or React Native for cross-platform mobile, Next.js or Astro for web. For data: PostgreSQL with PostGIS, DuckDB on Parquet, and optionally Snowflake or BigQuery for heavy research workloads. For pipelines: Python with dbt and Airflow, or Dagster for more complex DAGs. For visualisations: Plotly, Observable Plot, ECharts or D3, and Mapbox GL or MapLibre for maps. We choose what fits your engineering capacity and source data, not what we happen to be used to.
Can you also build an internal BI tool on open data?
Yes, it is actually one of the most requested variants. Policy teams, communications departments, sales teams in the public sector and consultancies often want an internal tool that helps them quickly consult government data for quotes, policy advice or market analyses. We build it as a web app with SSO via your Azure AD or Google Workspace, a query builder on top of an internal data warehouse, and export functions to PowerPoint or Word. This type of project draws on our experience with BI and dashboard tooling and can also remain purely internal, without a public-facing layer.
How do you handle data quality and outliers?
Open data always contains exceptions: a Eurostat series in which zero means "no data" rather than "zero", a PDOK layer with duplicate geometries after a dataset migration, a CBS table in which a revision appears only in the footnotes. We build automated validation checks: range, type, completeness and outlier detection. Findings are logged and flagged in the app, so end users can recognise a suspected outlier instead of presenting it uncritically. For research, we report issues back to the source owner.
How do you combine open data with AI or language models?
More often than before, clients want users to be able to query an open-data app in natural language, for example "show me Dutch municipalities where the housing stock is growing faster than average but the population is shrinking". We build this as an AI layer on top of a structured query layer: the language model translates the question into a validated query, the query runs on the open-data backend, and the result comes with an AI-generated explanation that the user can verify. For our approach and its responsible use, see our page on AI development and our broader work on AI apps.
Do you work with our own developers or data analysts?
Yes, in open-data projects that is more the rule than the exception. Many organisations have a data team that handles the analysis side, while we build the app layer, ingestion pipelines and visualisation layer. We work in the same Git repository, align with your data schemas, and do pair programming where it speeds up knowledge transfer. By the end of the project, your own team can maintain the codebase entirely independently, with no lock-in to us as a supplier, because in open-data projects transparency and repeatability are where all the value lies.
What if we already have a data engineering platform?
Good news: then the step to an open-data app is shorter. We connect to your existing data platform (Snowflake, Databricks, BigQuery, your own Postgres cluster) and use the data already processed there as a starting point. We then don't need to rebuild the ingestion and transformation layer, or only partly; the focus shifts to the app and visualisation side. If you don't have a data platform yet, we often combine the project with the approach from our data engineering platform practice.
Do you also build open-data apps for municipalities or government organisations?
Yes, it is actually an important part of our open-data practice. We work with municipalities, environmental services, provinces and implementing organisations that want to open up their data to citizens, council members or their own organisation. This calls for extra attention to INSPIRE compliance, audit logs, administrative-law traceability and integration with DigiD or eHerkenning for resident functionality. For the overlap with municipal software challenges, see our page on software for municipalities.

Talk to us about your open data app.

A half-hour introductory call, no obligation. We listen to the story you want to get out of the data, look at the sources you already use or would like to use, and give you direction you can act on — even if the advice is to get further first with a notebook, a QGIS publication or a Tableau dashboard before custom development pays off. For related topics, see our broader app development, our page on geographic apps, our data engineering platform practice, our BI and dashboard tool approach and our experience with software for municipalities.

Edit content