Using open data
In practice, using open data usually comes down to five steps: first determine your question, then choose a suitable source, check the license, retrieve the data, and only then start analyzing. This sequence prevents you from investing time in a dataset that appears to be public but is not up-to-date, incomplete, or difficult to reuse.
Not every public dataset is automatically usable. It is precisely data quality, structure, and currency that determine whether you can build reliable research, analysis, or communication upon it. In this article, you will read how to practically use open data for reuse in analyses, research, and clear information provision. Do you want to indicate reliability and origin? Then also read more about visualizing data quality.
When a dataset is truly usable
Public information is only truly open data when reuse is possible, not just access. A table available online that lacks clear permissions, explanation, or a machine-readable format is often of limited use. To work with a dataset, you need context: what do the fields mean, how was the data collected, and what period does it cover? Currency also matters, especially if you want to compare trends, policy, or performance. Once this foundation is in place, you can use open data for analysis, research, websites, apps, infographics or dashboards, and communication.
Work from a question, not from a dataset
Start with a concrete question, target group, or decision. Do you want to explain a development, substantiate a target group segment, or present figures clearly using data visualization? Then also read data visualization: explanation and examples. That will help you better understand which data is truly needed.
Determine the variables, period, and level of detail in advance. Consider geographical scale, definitions, units of measurement, and whether you need monthly, quarterly, or annual figures.
Next, create a shortlist of datasets. Do not choose based on availability alone, but on relevance, completeness, consistency, and practical feasibility.
Finally, assess whether open data is sufficient. Sometimes open data forms a strong basis, but you need to supplement it with internal figures, interviews, surveys, or other research to fill gaps.
Where do you find open data and how do you assess the quality?
First search government portals, thematic catalogues, and public data sources of knowledge institutions, regulators, or research organizations.
Always read the metadata: look at origin, update frequency, geographic coverage, definitions, measurement method used, and file format.
Pay attention not only to the data itself, but also to findability and documentation. A good dataset without field descriptions or version information often takes an unnecessary amount of time.
Where possible, compare multiple sources or similar organizations so that you detect differences in coverage, definitions, or currency early on.
File, feed or API: choose the access method that suits you
| Access form | Handy when | Plus point | What to look out for |
|---|---|---|---|
| Download file | If you need a snapshot or a complete bulk set for your own analysis. | You work locally with complete files, which is often useful for archiving and reproducibility. | Check the version, cutoff date, and whether updates need to be retrieved manually or periodically. |
| Feed | If you want to follow recurring updates without having to search again and again. | Suitable for monitoring and automatic refreshing in a fixed structure. | Not every feed offers historical data or full bulk access. |
| API | If you want to retrieve filtered, targeted, or near real-time data. | You retrieve exactly the fields or records you need, handy for apps and dashboards. | Pay attention to documentation, limits, and whether a full export remains available alongside individual requests. |
| PDF or separate web page | Only as a reading source, not as an ideal form of reuse. | Human-readable, but weak for automatic processing. | Preferably choose machine-readable formats such as CSV, JSON, or XML to avoid error-prone retyping or extraction. |
Make complex information clear, understandable, and convincing
This is how you check licenses, rights, and restrictions
An open license determines what you may do with a dataset: copy, modify, combine, and republish. Therefore, always check the conditions in the metadata, source references, or documentation accompanying the dataset. This usually also states whether source attribution is mandatory and whether additional conditions apply to distribution. Reuse does not mean that you acquire the intellectual property rights to the original data; you receive permission to work with it within the scope of the license. Additionally, privacy, confidentiality, or third-party rights may impose limits. Thus, even with open data sources, it remains important to verify whether the content can truly be reused freely and responsibly.
From raw data to insight: cleaning, combining, and visualizing
Raw open data rarely provides immediate insight. Ensure column names are consistent, align units, and check for missing values, duplicate records, and striking outliers before drawing conclusions. Combine datasets only if definitions, time periods, and geographic levels truly align; otherwise, you are comparing seemingly identical figures with different meanings. Document your operations so that analyses remain verifiable and reproducible. If the outcome needs to be shared more broadly, translate it into clear charts, a data visualization, interactive infographic , or dashboard visualization. Especially for complex topics, data visualizations support faster understanding and better decision-making for executive decision-making.
Frequently asked questions about open data and data sources
These are short, practical answers to questions that often arise as soon as you start working with open datasets.
What are 3 types of data?
The three main types are structured, semi-structured, and unstructured data. Structured data is stored in fixed fields and tables, such as CSV or relational databases, and is the easiest to analyze. Semi-structured data does have structure, but it is less rigid, such as JSON or XML. Unstructured data consists of free text, images, audio, or video. Open datasets usually fall into the first two categories. This is advantageous because the data type determines which tools, checks, and operations you need.
How do you collect data for research?
Always begin with the research question and then determine which data are needed to answer that question. Open data is often a good secondary source for context, trends, and comparisons. Next, check whether the available dataset offers sufficient coverage, currency, and detail. Only if gaps remain should you supplement with surveys, interviews, observations, or your own measurements. Always record the source, date, method, and known limitations. This ensures your research remains verifiable and allows you to justify how conclusions were reached later.
What is the best way to store data?
The best approach is to always keep an unaltered source version alongside your working file. This allows you to return to the original later and ensures your analysis remains verifiable. Preferably choose open, machine-readable formats such as CSV, JSON, or XML, supplemented with clear version names and dates. Also store metadata, license information, and edit notes in the same folder or project structure. This prevents the loss of important context and makes reuse, collaboration, and reproducibility much easier.
Ready to take your A to AA?
Whether it concerns an ambition, product, process, technology, or idea: we make visible what makes it special and translate it into imagery that is clear, convincing, and impossible to ignore.