Blog

7 data storytelling examples worth learning from

Seven published pieces built on data, each assessed on its source and access, evidence, narrative and presentation, and what makes each one strong or weak.

By Andre Guelmann · Updated · 14 min read

Contents

The seven examples on this page are finished, published pieces built on data, and each one is live at an address you can open and inspect. Finished work shows how other teams name their sources, build an argument around numbers and present charts or interactive tools. It also shows where they leave gaps. Each piece is assessed on four things (source and access, evidence, narrative and presentation), with no assumptions about traffic, links or business results.

Which examples are on this page, and why

Every example here was published for an outside audience, and each one links to the original so you can check it yourself. A chart, an infographic, a calculator or an interactive map can stand on its own as a finished piece, so format never decides whether an example belongs here.

Chart makeovers, before-and-after redesigns and step-by-step walkthroughs are teaching exercises, so they are left out. Collections such as Storytelling with Data’s examples and makeovers gather those exercises for practice. This page looks only at pieces published for real readers.

What counts as an example here

On this page

Pieces published for real readers, in any format: an article, a chart, a calculator or an interactive map.

Not on this page

  • Chart makeovers.
  • Before-and-after redesigns.
  • Step-by-step walkthroughs.

The examples below sit in three groups, in this order:

  1. Examples from companies and nonprofits
  2. Interactive and visual pieces that stand on their own
  3. Newsroom examples, identified as such

Every note covers evidence, narrative and presentation. None describes outcomes such as page views, shares or sales, because being published does not show how a piece performed.

The seven examples at a glance
  • CSIS Hidden Reach Policy research center, 2025: a scrolling feature on China’s dual-use shipyards.
  • BetterUp Coaching company, 2021: an insights report on inclusive leadership.
  • Albemarle Street Partners Financial-adviser coalition, 2022: a white paper built on a survey of 49 advisers.
  • Periscopic Data visualization studio, 2013 data: an interactive on US gun deaths.
  • The Guardian, Bussed out Newsroom, 2017: an investigation built on public records from 16 cities and counties.
  • The Guardian, The killing times Newsroom, 2019 (updated 2025): an interactive map of frontier massacres.
  • RNZ Newsroom, 2022: an investigation of New Zealand’s threatened species.

How each example is assessed: evidence, narrative, presentation

To judge a published data piece consistently, run four checks in order:

  1. Source and access: Does the piece open without a paywall, who published it, and does it name where its data came from?
  2. Evidence: Does it state the dataset, the time period and the counts, and does it explain its limits?
  3. Narrative: Does it make one clear main point, and does it set the context before showing the numbers?
  4. Presentation: Does each chart carry one clear point, and are the labels, scales and colors easy to follow?

The four panels below apply those checks to Bussed out: How America moves its homeless (The Guardian, 20 December 2017), checked against the original.

Examples from companies and nonprofits

These three come from a policy research center, a coaching company and a financial-adviser coalition. They publish findings from internal data, a survey of their own and compiled records:

  • Murky Waters: Navigating the Risks of China’s Dual-Use Shipyards (CSIS Hidden Reach, published 25 March 2025): A scrolling feature by Matthew P. Funaiole, Brian Hart and Aidan Powers-Riggs. It opens without a paywall and links to the full report, Ship Wars. On its data, CSIS says: “Hidden Reach analyzed over 4,500 shipbuilding records to develop a first-of-its-kind dataset of 307 active Chinese shipyards. Leveraging commercial satellite imagery and other open-source material, the analysis categorizes each shipyard into one of four tiers according to its integration into China’s naval shipbuilding ecosystem and broader defense industrial base.”

    • Evidence: It cites China’s national shipbuilding industry association for the $43 billion export figure for 2024. Its charts credit “Authors’ research and S&P Global Sea-web”, including the chart of yearly production in gross tonnage beside the text’s figure for China’s share of the world total (5% in 2000, over 53% in 2024). A text table defines the four tiers, and the piece says that “sparse public information makes it difficult to calculate how revenues are distributed across the tiers”. The gaps: the text doesn’t say where the 4,500 records came from, and the page shows no single shipyard’s evidence for its tier.
    • Narrative and presentation: Context comes first: US yards built five large merchant vessels in 2024, while one Chinese state shipbuilder delivered over 250. Then one main point: foreign orders and technology flow into yards that also build China’s navy. The layout scrolls, with satellite imagery. The strongest visual is a dot chart (“Each = 1 shipyard”) shown first by count, where private Tier 4 yards are 74% of all shipyards in China, then resized by average annual tonnage to show that state-owned yards produce more.
  • BetterUp Insights Report, Fall 2021: A new kind of inclusive leadership for a new world of work (BetterUp, Fall 2021): A company report that opens without a registration form; a “Request a demo” button sits at the top. On its data, BetterUp says: “Since the start of the pandemic, we’ve tracked over 10,000 BetterUp Members engaged in our 1:1 personalized coaching. These members represent a cross-section of the workforce and provide a window into the unique differences, challenges, and needs of populations impacted by the pandemic.”

    • Evidence: The page has no methods section and no notes. Headline figures such as “3.4X more job satisfaction”, “2.7X higher organizational commitment”, “1.6X more likely to have low belonging”, a “17% boost” and “28%” for parents come without respondent counts, time periods or definitions. The 28% chart does show its comparison (low versus high support, scored 1 to 5), but not counts, period or definitions. Two broader workplace figures have no named source. And the page says members represent a cross-section of the workforce without showing how that was checked.
    • Narrative and presentation: Opens with the workplace changes of 2020 as context, builds to one main point (managers need inclusive leadership) and sets out four insights for managers, with big-number callouts, a carousel and shareable charts.
  • A Manifesto to Better Support Independent Financial Advisers (Albemarle Street Partners for the Coalition of the Independent, 2022): A coalition white paper that opens directly and is marked “For Professional Investors Only” and “CAPITAL AT RISK”. Albemarle sells investment services to these advisers, and Chapter 5 promotes Albemarle as an investment partner. On its data, Chapter 2 says “A recent survey of 49 advisers by Albemarle”, and the final chapter says “Our research was conducted between January and February of 2022 through online email survey.”

    • Evidence: The respondent count and the fieldwork dates sit in different chapters, and percentages appear without per-question counts. Several shares (16.6%, 87.5%, 6.2% and 14.5%) fit 48 answers rather than 49 (8 of 48 is 16.67%), and 40% fits neither, so several questions had different bases that the paper never states. Its charts show no counts either. It also cites PIMFA survey figures and FCA rules.
    • Narrative and presentation: Uses consolidation since the Retail Distribution Review as context and argues that independent firms should work together. It runs to six chapters with a contents list and videos, labels figures emphatically (“A staggering 59.1%”, “A huge 73.4%”) and prints survey questions as chart titles in the final chapter.

Interactive and visual pieces that stand on their own

Interactive pieces let readers filter, hover or play through a timeline, and that changes how you judge them. Check what data drives the visual, whether the source is named, which assumptions the reader can see or change, and whether the interaction makes the main point clearer or hides limits behind controls.

  • U.S. Gun Deaths (Periscopic, 2013 data): A public-service interactive that opens in the browser. A “Sources and Methods” link at the foot of the screen opens an overlay that offers data from 1985 to 2018 to download.
    • Evidence: Inside the overlay, Periscopic explains why it used a non-government source for 2013: “Our 2010 data was the latest such information released by the federal government for several years and we resorted to a less refined source to obtain current data for 2013. Anonymous Twitter user, @GunDeaths, continuously gathers and tweets gun-related deaths in the U.S.” It adds that “this data is incomplete. Some of the data may be missing or inaccurate, and many killings are not even reported in an online news source”, and says that suicides are included. Projected “stolen years” are estimated from World Health Organization cause-of-death data, weighted to the age distribution of US deaths, with no adjustment for life-expectancy differences between demographic groups. The 2013 note doesn’t say which year of WHO data it used.
    • Narrative and presentation: One main point: the years of life lost. Filters cover sex, age group, region and season. Each death is an arc that starts at age zero, orange for the years lived and gray for the projected years lost. On the main view, running counters show exact totals of people killed and stolen years as the months play. The warnings that the data is incomplete and includes suicides appear only after a click on “Sources and Methods”.

Newsroom examples, identified as such

Newsroom data journalism is made under editorial standards and reporting timelines that differ from company content, so these pieces are labeled separately. They are worth studying because they are public and often name their sources.

  • Bussed out: How America moves its homeless (The Guardian, published 20 December 2017): A newsroom investigation built on public records from 16 cities and counties. It pairs a national map with individual accounts and has a full methodology note (assessed in full above).
  • The killing times: a massacre map of Australia’s frontier wars (The Guardian, published 4 March 2019, updated 23 February 2025): An interactive map that opens behind a content warning. Its data is “reproduced in collaboration with the University of Newcastle Colonial Frontier Massacres Project team.”
    • Evidence: Covers mass killings from 1788 to 1928. The text says each marker is a fatal conflict in which more than six people died, but the death-count filter’s lowest band is “6 to 10”, so it isn’t clear whether a killing of exactly six is counted. The limits are stated plainly: verifying sites takes time, the information is what could be carefully researched so far, and the project is complete, with no new sites to be added. The data collection methods sit in an About tab on the map.
    • Narrative and presentation: The piece names its own main point: “a sustained and systematic process of conflict and expansion”. It is a map with a date filter, plus filters for death count, the group with the highest fatalities and the attacker group, and a search by location or postcode.
  • Aotearoa’s vanishing species: Creatures great and small (RNZ, 26 November 2022): Part one of a two-part investigation by Farah Hancock. It opens without a paywall and credits New Zealand Threat Classification System content licensed by the Department of Conservation (CC BY 4.0).
    • Evidence: Shares are in the text and counts are in the charts: 1,352 assessed endemic species, 644 insects, 62 reptiles and frogs, and 57 water species, broken down by threat level. It explains the source’s limits: the system is a rolling summary, with groups of species reviewed on a five-yearly cycle rather than a single snapshot, and the piece calls it “our best view … although knowledge gaps mean it’s a blurry view.” The gap is dating: the piece itself shows no date (RNZ’s own page for it gives 26 November 2022), and nothing says when the classification data was taken.
    • Narrative and presentation: Opens on a scene, a scientist searching burnt tussock for geckos, then moves to national totals, group-by-group cases and gaps in legal protection. The charts and tables are made with Flourish, and every image in the composite species graphics is credited.

What these examples cannot show you

Reading a published piece tells you how it was sourced, structured and presented. It can’t tell you whether the piece earned search rankings, backlinks, media pickup or sales, unless the publisher releases verified performance data. Use these pieces as examples of how evidence, narrative and presentation work together, and of where missing dates, unstated bases or small samples limit what a reader can trust. They are not proof that copying a format will get you the same results.

What a published piece can and can’t tell you

Can tell you

How it was sourced, structured and presented, and where missing dates, unstated bases or small samples limit trust.

Can’t tell you

  • Whether it earned rankings, backlinks, media pickup or sales.
  • Whether copying its format would get you the same results.

From examples to your own data

A dataset becomes publishable content only once a finding has been computed, checked and given an honest angle. To see what records your company holds and what you can release, start with proprietary data for content. Then turning data into content walks through finding, checking and writing up a result. Our hub on data-driven content marketing shows how all the guides fit together.


If you have data and don’t know what to do with it, we find a question it can answer that your customers would care about. Book a demo