Finding the Lede: What Actually Turns Data Into a Story
why a spreadsheet full of correct numbers isn't a story yet, and what it actually takes to find the one sentence a dataset is trying to tell you.
Part 1
Part 2
Leading With the Headline, Not the Methodology
why burying the point under three slides of data sourcing and scope notes loses the reader before the story even starts, and what to put first instead.
Part 3
Writing for the Reader in Front of You
the same set of facts gets a different opening paragraph depending on which desk it runs on, and why a data story needs the same discipline.
Part 4
The Inverted Pyramid: Ordering a Data Story So It Survives Being Cut
why newsroom copy is written to survive being trimmed from the bottom, and how the same ordering principle protects a data report from being skimmed.
Part 5
The Nut Graph: Setting the Scene Before the Number
why a headline number without context is just a number, and how the paragraph that follows it is what actually makes a data story matter to anyone.
Part 6
The Turn: When the Data Isn't What Anyone Expected
the narrative device that makes a story stick isn't the number itself, it's the gap between what everyone expected and what actually happened.
Part 7
Show the Wire Photo, Then Write the Caption
why leading with the evidence and explaining it second earns more trust than announcing a conclusion and backfilling proof, and when to use each order.
Part 8
One Story, One Chart: Avoiding Narrative Overload
why cramming five findings into one dense chart produces zero memorable ones, and the newsroom discipline of running one story at a time.
Part 9
The Anchor Stat That Won't Let Go
why a single well-chosen number, repeated deliberately, outlasts an entire report in a reader's memory, and how to choose one worth anchoring to.
Part 10
The Editorial Desk: A Review Process for Data Narratives
no story runs without an editor reading it first, and no data narrative should ship without an equivalent review checking the same things a copy desk checks.
Part 11
Fact-Checking a Data Story Before It Ships
an editorial review checks whether the story is well told; fact-checking is the separate, narrower discipline of verifying every claim in it is actually true.
Part 12
The Newsroom Style Guide, Applied to Data Narrative
a wire service can run hundreds of reporters' work under one voice because of a shared style guide, and a data team needs the same shared discipline.
Part 13
Running a Correction: Fixing a Data Story After It Ships
a newsroom doesn't quietly edit a published error and hope no one notices; it runs a correction, visibly, and that discipline applies just as much to data.
Part 14
The Wire Service Drafts the First Story
AI-assisted first-draft narrative generation from raw data is a genuine acceleration of the hardest, slowest part of the job, not a replacement for the reporter.
Part 15
The Copy Desk That Checks the AI's Work
an AI copy desk automatically traces a generated narrative's claims back to the underlying data, catching a class of error a human reviewer alone would miss at scale.
Part 16
Same Story, Different Front Pages
AI can now generate genuinely personalized narrative framing for different audiences from one underlying dataset, at a speed that makes consistency checking essential.
Part 17
A Wire Service for Every Desk in the Building
generating data narratives at real organizational scale, across hundreds of dashboards, requires the same discipline this series covers, just systematized.
Part 18
The Confident Byline That's Wrong
the real risk of AI-generated narrative isn't obvious nonsense, it's a fluent, confident claim that sounds exactly as authoritative whether or not it's actually right.
Part 19
The Newsroom That Never Closes
real-time narrative generation for continuously updating data borrows the discipline of a live election-night desk, where the story itself keeps changing underneath the writer.
Part 20
Every Desk, One Newsroom
the lede and the nut graph, the editorial desk and the correction, the wire service drafting fast and the copy desk checking its work, every article's lesson reassembled one last time.