What is data journalism?
Data journalism is reporting where the data is the source of the story, not decoration added afterward. A data journalist obtains a dataset, checks it, analyzes it, and reports what it shows, often publishing charts, maps or the data itself alongside the article [1][3].
A classic example: a reporter obtains a city's restaurant inspection records, finds that one neighborhood's restaurants are inspected half as often as the rest, and interviews officials about why. The spreadsheet found the story, and conventional reporting confirmed and explained it.
The Global Investigative Journalism Network describes data journalism as using datasets to uncover hidden information and to find patterns, systemic issues and outliers in the public interest [3].
How does data journalism work?
Data journalism works through a repeatable sequence: get the data, understand how it was collected, verify it, analyze it, confirm the findings with traditional reporting, and decide how to publish. GIJN sets out the same stages, from obtaining data through documentation and analysis to supplementary reporting [3].
The tools range from spreadsheets to programming languages such as Python, R and SQL, plus specialist tools for pulling tables out of PDFs [3]. The tools matter less than the habit behind them: treating every dataset as a source that can be wrong, incomplete or biased by how it was gathered.
Where did data journalism come from?
Data journalism grew out of computer-assisted reporting in the 1950s and 1960s and Philip Meyer's precision journalism in the 1970s, and took its current name as the web made publishing data and interactive graphics routine [1][3]. Data has been part of news for far longer: the Data Journalism Handbook points to the Guardian publishing a table of Manchester school data in 1821 [1].
1952: CBS uses a computer on election night
CBS used a computer to predict the result of the 1952 US presidential election, which the Data Journalism Handbook counts as the first organized use of computer-assisted reporting [1].
1967: Philip Meyer surveys Detroit
After the 1967 Detroit riots, Philip Meyer used survey research analyzed on a mainframe computer to show that people who had attended college were as likely to have rioted as high school dropouts, overturning a common assumption [2].
1973: Precision Journalism is published
Meyer's book Precision Journalism: A Reporter's Introduction to Social Science Methods was published by Indiana University Press in 1973 and argued that reporters should borrow the methods of social science [2].
1989: NICAR is established
The National Institute for Computer-Assisted Reporting was established in 1989 with support from the Missouri School of Journalism, giving data reporting a training base in US newsrooms [3].
2006 to 2010: data on the web
In 2006, Adrian Holovaty argued that news sites should publish structured, machine-readable data alongside articles [1]. Large leaked datasets handled by the Guardian and The New York Times in the WikiLeaks era, and early international data journalism conferences around 2010, brought the term "data journalism" into wide use [1].
What is the difference between data journalism and computer-assisted reporting?
Computer-assisted reporting (CAR) is the older term for using computers to analyze records for a story, while data journalism also covers publishing the data, interactive graphics and visual storytelling for readers. CAR was mostly a back-office research method. Data journalism moved the data to the front of the page and, at its best, lets readers check the work themselves [1].
What is the difference between data journalism and data-led digital PR?
Data journalism is done by a newsroom to find out what is true, while data-led digital PR is done by a brand to earn coverage with data it has created or analyzed. The two meet when a journalist uses a brand's data, and that meeting is where most data-led campaigns succeed or fail.
| Data journalism | Data-led digital PR | |
|---|---|---|
| Who does it | Journalists and newsroom data teams | Brands and agencies |
| Goal | Find and explain the story | Earn coverage, mentions and links |
| Typical data | Public records, leaks, official statistics | Commissioned polls, own product data, public data reworked |
| Built to answer | An open question | Usually a chosen angle |
Formats on the PR side include survey-led campaigns, data study link building and statistics pages.
What do data journalists check before using brand data?
Data journalists check who produced the data, who paid for it, how it was collected, how big and representative the sample is, how questions were worded, and whether the headline claim matches the full results. Guidance for journalists from IMPRESS and the Market Research Society lists exactly these checks for surveys and polls and asks reporters to request the data tables and technical notes [4].
A brand pitching data to data-literate reporters should be ready to provide:
- The raw or tabulated data, not just the headline percentages.
- A clear method, including dates, sources, sample size and any exclusions.
- The limits, stated before the journalist finds them.
- Correlation stated as correlation. If the finding is an association, say so. Search Insider's own studies, such as the year-in-title study and its AI Overviews follow-up, spell out why a raw association between a page feature and a result can be misleading until confounders such as page age and rank are accounted for.
Why does data journalism matter for SEO and link building?
Data journalism matters for SEO because original data is one of the few things journalists will cite and link to without being asked twice. A newsroom that reports a statistic usually needs to name its source, and the source page is a natural link target.
The bar is credibility, not volume. Fractl's own campaign data suggests that topic and promotion move coverage far more than survey sample size does [5], and an analysis that a data journalist finds sloppy is unlikely to be cited at all. Data pages that publish full methods and downloadable tables are the linkable assets that keep earning citations after the first wave of coverage.
What are the limits of data journalism?
The limits of data journalism are the limits of its data: records can be incomplete, collected for other purposes, or biased by who was counted. A clean chart can make a weak dataset look authoritative. The honest practice is to report uncertainty, explain what the data cannot show, and back the numbers with interviews and documents.
How should a brand work with data journalists?
A brand should work with data journalists by offering data that answers a question their readers already have, publishing the method in full, and accepting that a journalist may reach a different conclusion. Treat the data as a source, not a press release: supply it early, answer method questions quickly, and never adjust the analysis to protect a headline.