Data journalism is a broad field spanning newsroom analysis, freelance reporting, teaching, coding, visualisation, and investigative research. The State of Data Journalism reports from 2021, 2022, and 2023 show a profession that is relatively young, often self-taught, and increasingly connected to artificial intelligence and open-source intelligence (OSINT). They also show persistent pressure around time, money, skills, and access to reliable data.
Contents
- Who works in data journalism
- Employment, income, and experience
- How data journalists learn and work with code
- What data journalism contributes to reporting
- Data sources, newsrooms, and collaboration
- AI and OSINT in data journalism
- The main barriers and the pandemic effect
Who works in data journalism
The 2023 State of Data Journalism report describes a field with a near-even gender split. Men represented 49% of respondents, women 48%, non-binary or genderqueer respondents 1%, and those who preferred not to answer 1%. The age profile was comparatively young: more than 69% were younger than 44, and the largest age group was 25–34, at 33%.
Respondents worked across a geographically distributed profession. The United States had the largest share by country of work, at 8%, followed by Italy at 7%, Germany at 6%, and the United Kingdom and Russia at 5% each. These figures describe the distribution of respondents by work country, not the size of national data-journalism industries.
Education was a prominent feature of the 2023 sample. A master’s degree was the most common education level, at 46%. Among women respondents, 59% had obtained a master’s degree or PhD. The occupational mix was broader than traditional newsroom employment: 36% worked full time at news companies or organisations, 13% were full-time freelancers, 10% were part-time freelancers, 9% were educators, and 7% were students.
Students were the youngest occupational group, with 77% under 34. Experience was more varied. The largest single group, 30%, had three to five years in data journalism. Respondents with less than three years represented 35%, while 30% had more than five years. Experience also differed by role: 51% of editors had more than six years in data journalism, as did 46% of educators. Women were more likely than men to be newer to the field: 42% had less than three years of experience, compared with 28% of men.
Employment, income, and experience
Income information was provided by 95% of 2023 respondents. Among those reporting income, 52% earned less than $25,000 annually and 8% earned more than $75,000. The report also found that nearly one in four respondents working for organisations with 500 or more employees earned over $50,000. These income figures should be read as respondent-reported annual earnings, not as a salary benchmark for every data journalist.
Employment arrangements were often mixed. Overall, 35% worked for two organisations or more. The share was 56% among full-time freelancers and 38% among part-time freelancers. By contrast, 72% of full-time employees worked for one organisation, as did 49% of part-time employees. Just over one in five respondents worked in companies with at least 500 employees.
The 2023 employment picture can be compared with the 2021 report, although the years measure separate respondent groups:
| Measure | 2021 | 2023 |
|---|---|---|
| Full-time at a news company or organisation | 28% | 36% |
| Educators, teachers, trainers, or faculty | 15% | 9% |
| Students | 8% | 7% |
| Companies with 500+ employees | 22% | Just over 20% |
| Self-taught through online or offline resources | 70% | 23% solely self-taught |
The self-taught figures are not directly interchangeable: the 2021 statistic covers respondents who used online or offline resources, while the 2023 figure identifies respondents who were solely self-taught. The difference is therefore a change in wording and measurement as well as a difference in year.
How data journalists learn and work with code
Online self-directed learning remained important in 2023. The most common learning route was teaching oneself through online resources, at 53%. Just under a quarter learned data journalism through higher education, and 23% were solely self-taught.
Coding was part of the work for 19% of 2023 respondents. Among those who coded, Python was used by 69%, HTML/CSS by 61%, JavaScript by 46%, and R by 39%. Programming experience was substantial within this subgroup: 64% had known how to code for more than six years, while 58% used programming daily.
The earlier State of Data Journalism 2021 blog summary records a related tool picture. Among respondents who programmed, Python was used by 63%, HTML/CSS by 51%, and R by 46%. Across the wider 2021 respondent group, Excel was used by 75%, Google Sheets by 60%, Datawrapper by 37%, Flourish by 32%, and Tableau by 27%. The 2021 and 2023 coding percentages come from different reports and should not be treated as a single continuous trend.
Formal journalism training was reported by 56% in 2023, while 14% had training in machine learning. Interest in additional machine-learning education was higher: 57% wanted to upskill in machine learning. Self-assessments were more modest in some core areas. Twelve percent rated themselves advanced in data analysis and 15% advanced in data visualisation. Among students, 16% rated themselves advanced in journalism.
What data journalism contributes to reporting
Respondents in 2023 most often connected data journalism with stronger reporting. Seventy percent said it enhances the reliability of stories, 61% said it helps contextualise stories, 60% said it facilitates discovering relevant stories, and 59% said it uncovers unique newsworthy events. More specialised uses were less common: 33% identified event detection and monitoring as useful, while 24% identified predictive analysis.
The 2022 challenges report presents similar perceptions. Seventy-three percent said data journalism makes a story more reliable, and 71% said it helps readers make sense of what they are reading. These are perceptions reported by survey respondents, rather than experimental estimates of causal effects.
The 2022 work-practices report also shows what data journalists actually did. The largest share, 58%, used stories where data supported the narrative. Analysing data was reported by 75%, data gathering by 62%, creating data visualisations by 54%, cleaning data by 45%, and coding for work by 23%.
Data sources, newsrooms, and collaboration
Public information was central to data-journalism practice in 2022. Seventy-one percent used public official governmental data, while just about one in five used data obtained through freedom-of-information requests. More than half of respondents working in Mexico said they had used FOI data. Census data was used by nearly three in four data journalists in Canada and by 63% of those in the United States. Pakistan was the least likely country in the report to use public official governmental data, at 53%.
The distribution of work across media organisations was also mixed. Nearly half of data journalists worked at an online-only digital outlet in 2022. Another 39% worked at print or broadcast media outlets that had a digital site, and 29% published their work on social media.
Dedicated data units were not universal. The largest share of respondents, 34%, did not work in a dedicated data unit, while around one in four did. Among respondents in dedicated units, two in three worked in teams of fewer than five people. Collaboration extended beyond internal teams: 26% reported collaborations with other news companies or organisations. Of those collaborations, 43% involved separate content production and 56% were one-time collaborations.
AI and OSINT in data journalism
In 2023, just over one in three respondents had used artificial intelligence as part of their data-journalism work. Among AI users, 45% used it to search for content, 38% for content generation, 37% for text mining, and 8% for predictive modelling. Generative AI tools were used by 62% of AI users.
Understanding the technology was the largest reported AI challenge: 56% of AI users cited limited understanding of the tool or technology. A further 41% said a lack of dedicated time to experiment with AI was a barrier. These figures describe the experiences of AI users or respondents reporting the relevant challenge, so they should not be read as shares of all respondents without that qualification.
OSINT was used by one in five respondents in 2023. Its use varied by newsroom setting: 23% of respondents in international newsrooms used OSINT, compared with 12% in local newsrooms. Among OSINT users, nearly 60% used it to verify sources, images, or videos, and 57% used it to navigate maps, satellites, and location-based information. Government data and reports were used by 72%, while around six in 10 used news media and internet media.
The main barriers and the pandemic effect
Access to quality data was the top hurdle in 2023, at 65%. Financial resources were a learning barrier for 53%, time pressure was a challenge for 48%, and lack of adequate knowledge in data analysis was a challenge for 47%. The 2022 report showed a similar ordering: access to quality data was the major hurdle at 57%, lack of financial resources affected 50%, and time pressure affected 49%. Lack of consumer interest was cited by 18% in 2022.
The 2021 pandemic report records how COVID-19 affected participation and work. A quarter of respondents became involved in data journalism because of the pandemic, and 56% worked on a project covering it. Forty-six percent said the pandemic strengthened data journalism as a field, while 43% said it increased audience data literacy. Improved access to data was reported by 28%.
Work arrangements changed as well. Forty percent switched to working from home and 32% moved to a hybrid work policy. Time pressure was unaffected for 30% but increased for 36%, while 44% saw their workload increase. The period also coincided with reported skill development: 40% improved data-analysis skills, 39% improved journalism skills, and 36% improved data-visualisation skills because of the pandemic.
The 2021 employment and demographic reports add context to that period. Sixty-three percent had less than five years of data-journalism experience, 59% identified as men, 39% as women, 1% as non-binary or genderqueer, and 1% preferred not to say. Fifty-six percent were aged 25–44. Just under 60% of income respondents earned between $1 and $49,999 annually. Together with the 2022 and 2023 figures, these results depict a field whose practical value is widely recognised while its access to time, funding, training, and usable data remains uneven.