Data journalism is the practice of building stories from data: collecting or finding a dataset, analyzing it, and presenting what it shows through clear writing and charts. Newsrooms developed the craft, and content marketers borrow it to create pieces that earn coverage and links. It differs from original research mainly in emphasis. Research is about producing the data; data journalism is about finding and telling the story in it, which can include public datasets.
The workflow
Start with a question your audience cares about, not with a dataset looking for a use. Then find data that can answer it, clean it, analyze it and test whether the finding holds up when you change assumptions.
- Question: what would your audience or a reporter want to know?
- Sourcing: public datasets, APIs, your own product data or surveys
- Cleaning: duplicates, missing values, inconsistent categories
- Analysis: simple comparisons often make the clearest stories
- Visualization: one chart per point, labeled so it stands alone
- Publication: method notes, data download where possible
Accuracy and honesty
Mistakes in a data story can travel far once it is picked up. Have someone other than the analyst check the numbers. Distinguish correlation from causation in the text. Show uncertainty where it exists. If you publish a dataset, Google supports Dataset structured data, which can help it appear in Google Dataset Search.
Example
Mapping public data for a niche audience
Hypothetical company Copperlark sells permitting software to builders. Its team pulls public construction permit records from several city open-data portals, measures how long approvals take in each city, and publishes an interactive map with the method and a downloadable CSV. Local reporters covering housing use the findings, linking to the page.
Common mistakes
Cherry-picking
Choosing the time range or segments that produce the most dramatic finding damages credibility once someone checks.
Charts that need the article
Charts get shared without their context. Each one needs a title, labeled axes, source and date.
Questions
Not always. Many strong data stories use simple counts and comparisons that a spreadsheet can handle. Complex statistical claims do need someone qualified to check them.
