Last reviewed: Next review due:
What is data journalism?
Data journalism is the practice of using structured data — spreadsheets, databases, APIs — to find, verify, and tell stories that would not emerge from interviews alone. A data journalist asking “how many councils missed their potholes target last year?” can answer that question rigorously across all 317 English councils at once; a reporter relying on press releases can only report what they are told.
You do not need to be a programmer. The majority of published data stories in UK regional journalism are produced with nothing more sophisticated than Excel. Coding becomes useful when datasets are very large, when analysis needs repeating automatically, or when you are building a live interactive. Start where you are.
UK data journalism teams to follow
Guardian Visuals
World-class data visualisation and investigative data journalism. Follow their methodology notes to learn how they verify and present complex datasets.
BBC Shared Data Unit
Produces data analyses and shares them with regional and local newsrooms. An excellent model for collaborative local data journalism.
Bureau Local
Collaborative investigative network with a strong emphasis on community data stories. Open about methodology and data sources.
FT Visual Journalism
Sets the global standard for data visualisation in a subscription news context. Their chart style guides are publicly available.
When data journalism matters most
- 1Holding institutions to account at scale — comparing every NHS trust, every council, every police force simultaneously.
- 2Spotting trends that are invisible to individual sources — a rise in food bank use that no single foodbank manager can see nationally.
- 3Verifying official claims — does the government's stated reduction in knife crime hold up when you apply consistent geography and methodology?
- 4Finding the local angle in national data — which constituency in your patch has the highest waiting list? Which ward has the most planning refusals?
- 5Adding specificity to a qualitative story — a survivor's account plus the data showing how common their experience is.
Red flags in data journalism
- Using whichever average (mean, median, mode) produces the most dramatic number rather than the most accurate one.
- Comparing datasets that use different geographies, time periods, or definitions without adjusting.
- Treating correlation as causation — two things rising together does not mean one causes the other.
- Ignoring uncertainty — a survey of 500 people has a margin of error; reporting it as fact without caveat misleads readers.
- Not checking whether the data is complete — missing rows in a dataset look like low numbers, not missing data.
- Building a story on a dataset you do not fully understand — always read the methodology notes.
The data-journalism workflow
1
Question
Start with an editorial question, not a dataset. "Are councils spending more on consultants than frontline staff?" is a question. Once you have the question, work out what data would answer it.
2
Find the data
Check if the data exists and is published (ONS, data.gov.uk, NHS England, Companies House). If not, file an FOI request. If the data is on a website but not downloadable, consider scraping.
3
Analyse
Clean the data first (see our Data Cleaning guide). Then run your analysis — pivot tables, SQL queries, or Python scripts depending on scale. Document every step.
4
Find the story
What does the data actually say? Is it surprising? Does it contradict official claims? Does it show a trend? The story might not be what you expected — follow the data, not your hypothesis.
5
Check and sense-check
Run your key numbers twice. Ask whether the finding passes the common-sense test. Share with a specialist in the field for a reality check. Check your methodology against the data's own methodology notes.
6
Publish with a methodology note
Publish the key finding, then explain how you reached it in a methodology section or separate note. Include a link to the underlying data where possible. This builds trust and allows others to replicate your work.
Getting started checklist
- I have a clear editorial question I am trying to answer with data.
- I have checked whether the data I need is already published before filing an FOI.
- I have read the methodology notes for any dataset I am using.
- I have cleaned the data and documented what I changed.
- I have run my key calculation at least twice using different methods.
- I have had a specialist or trusted colleague check my findings.
- I have prepared a methodology note explaining how I reached my conclusion.
- I have cross-referenced my finding against related published statistics.
Need data that is not published?
Use our FOI Request Builder to draft a request for unpublished data from any UK public authority. Check our UK data sources guide for datasets that are already available.
Common mistakes
- Starting with the dataset rather than the question — leads to stories that are technically correct but editorially pointless.
- Not understanding how the data was collected — assuming survey data and administrative data are equivalent.
- Using absolute numbers when rates are more appropriate (e.g. crime numbers rather than crime rate per 1,000 population).
- Mixing data from different years without adjusting for changes in methodology or coverage.
- Publishing before checking whether a different, better dataset already exists.
- Not approaching the organisation the data relates to for comment before publication.
Related guides
Spreadsheet Essentials
Pivot tables, VLOOKUP, and conditional formatting with UK examples.
Statistics for Journalists
Averages, percentages, margins of error, and correlation explained plainly.
Data Storytelling
Turning numbers into narrative without misleading your readers.
UK Data Sources
The definitive guide to finding publicly available UK data.
Primary sources
Frequently asked questions
What skills does a data journalist need?
At a minimum: numeracy, spreadsheet fluency, the ability to read official statistics critically, and strong editorial judgement. Beyond that, SQL, Python or R, and basic data visualisation are valuable additions. Many UK data journalists are self-taught — the BBC Shared Data Unit and Bureau Local have trained journalists with no technical background at all.
How do I pitch a data-driven story to an editor?
Lead with the finding, not the method. Your editor does not want to hear that you ran a pivot table — they want to hear that council taxi spending rose 47% in three years while frontline services were cut. Always translate the number into human impact. Attach a simple chart and a methodology note. The methodology note shows your working and pre-empts 'how do you know?' questions.
What are the main ethical risks in data journalism?
Privacy (reporting granular personal data that could identify individuals), misrepresentation (cherry-picking the metric that tells the story you want rather than the most accurate one), harm (publishing data that could endanger a person or group), and uncertainty (presenting findings as definitive when they carry significant margins of error). The key discipline is to report what the data actually says, not what you hoped it would say.
Is Guardian Visuals a good model for UK data journalism?
Yes — Guardian Visuals is widely regarded as one of the best data journalism teams in the world. Their work combines rigorous analysis with clear, accessible visualisation. The BBC Shared Data Unit is particularly good for local data stories, as it shares analysis with regional and local newsrooms across the UK. Bureau Local is a collaborative model that is worth following for local and hyperlocal data journalism.
Related guides
Primary sources
- Office for National Statistics — datasets and methodology— ONS
- data.gov.uk — UK government open data portal— UK Government
- Code of Practice for Statistics— UK Statistics Authority
- BBC Shared Data Unit — open methodology and datasets— BBC
- Bureau Local — collaborative data journalism— The Bureau of Investigative Journalism
- NRS Scotland — Scottish vital statistics and data— National Records of Scotland
- NISRA — Northern Ireland statistics and research agency— NISRA