Interactive storytelling studio · Baltimore

New York Times-level storytelling, without the newsroom.

We take the research and reports you already publish and give them the hook, the sequence, and the graphics a Times desk would, then show you exactly who stayed.

GCERF · Annual report 2025Full story (opens in new tab)
What we start from

We do not start from blank. We start from what you already published.

Reports
Annual and impact reports, community benefit filings, economic scorecards, grant reports.
Evidence
Research studies, efficacy data, randomized trials, white papers, case studies.
Announcements
Press releases, product and program launches, campaign moments, funding news.
Recurring publications
Blog posts, donor updates, newsletters, member communications.
The format matters less than three things. A real audience you can reach, one action you want from them, and something in the material worth reading.

Story format changes outcomes. Even with the same content.

The win is not decoration. It is clarity, sequence, and a format people actually move through.

2–4×
Reading time
In a sample of 55 articles, the 11 interactive stories produced more than double the reading time of related standard articles. Some reached three to four times.
74.5%
retention
Long-form scroll structure held 74.5% of readers on average, against 68.3% for the same content as a slideshow.
3×
more helpful
Readers rated the scroll-based version significantly more helpful, and less confusing, than the equivalent static format.
Published research on interactive and scroll-based formats. Sources on request.
Stories built for
Why this form is rare · Take a guess
What is the most-read piece The New York Times has ever published?
Select one→

The form proved itself, then stayed rare.

A dialect quiz built from 350,000 survey responses outdrew everything else the Times ran that year, in eleven days. Then the paper's own innovation report said work like it could not be repeated: every piece needed a desk assembled from scratch.

Not because it does not work. Because almost nobody can afford to do it twice.

The desk is what we replace.
The cost and the per-piece assembly. Never the judgment.
Descriptive comparison only. Autumn is not affiliated with The New York Times.
The desk

Four disciplines, one piece.

A newsroom assembles this desk for every story, which is why the work stayed rare. We do all four, and the system makes it repeatable. We replace the cost and the assembly, never the judgment.

Editor
Finds the hook
The number worth opening on, and the question the reader wants resolved.
Writer
Builds the sequence
Evidence in the order a reader needs it, so each scene earns the next.
Graphics
Makes it visible
Findings you can see, beside the claim they support. Not left in a table.
Developer
Makes it hold
On a phone, on weak wifi, with every scene instrumented.
How we make it stick

A reader who has taken a position remembers the answer.

We put a key message in the form of a question, let the reader guess, and only then reveal it. Pre-testing is one of the best-documented ways to increase retention: the guess primes the memory, and the answer lands in the gap it opened. It works even when the guess is wrong.

Paired with a clear narrative and the picture beside the point it proves, it is how an interactive story leaves the reader with the number, not just the impression.

1 · Ask
The message becomes a question the reader wants resolved.
2 · Commit
The reader takes a position. The guess is measured.
3 · Reveal
Held back until the commit, then shown with the evidence.
See it in Maryland Owns the Preakness → (opens in new tab)
From the story · What it carries
Before the number: how many jobs does the Preakness support across Maryland?
About 8,000●
About 18,000●
More than 28,000●
28,000+
jobs
The Preakness supports more than 28,000 jobs, strengthens tourism and hospitality, and brings Maryland to audiences worldwide.
each dot represents 500 jobs
1 · Ask
Recreated from the live story for Greater Baltimore Committee
How it works

Four stages. One audience, one action, one story at a time.

There is no new content to create. You bring the research or the report. We find the number worth opening on, order the evidence the way a reader needs it, give the findings visual form, and build it to hold on a phone.

01
Direction
One audience, one desired next action, the source material, the distribution plan, and how we will measure it.
02
Production
Find the hook, restructure the evidence in the order the audience needs it, then data presentation, interaction design, responsive build, and QA.
03
Distribution
Adapt the story and its visual system for the channels that will actually carry it.
04
Launch and learning
Scene-level measurement, channel links, the readout, and what to change next time.
What we build to

ARC. Attention, readership, conversion.

Three things have to happen in order. Each is a design decision, and each is measured. The ask is available the whole way through; the story is built so more readers are still there to take it.

A
Attention
Get them to open it.
A hook worth stopping for, and a visual that travels with it: a social card, a short animation, a first screen that does work.
measured: how many opened it
R
Readership
Keep them reading.
Every scene pairs one idea with the visual that carries it. The narrative is written to a stated objective, to teach or to persuade, using what is known about how people learn and decide.
measured: how many kept reading, scene by scene
C
Conversion
Get them to do the thing.
One action, present throughout as a persistent call to action and pressed at the moments the evidence has earned it. More readers still there means more who can act.
measured: how many did the thing
ARC is what we build to, and what comes back. Every engagement produces an ARC read.
Tracking · What comes back

Every story comes back with an ARC read.

Not pageviews. Scene-level tracking of true engagement: how many opened it, how many kept reading, how many did the thing, and by whom. Numbers below are illustrative; we will walk through a real one on a call.

A
Opened it
1,240
Readers who reached the first screen, from every channel.
R
Kept reading
610
Still there at the midpoint. Half of everyone who opened it.
C
Did the thing
86
Clicked the ask. One in five of those who reached it.
Then, the four pages behind the numbers
01 · Section-by-section reach
Where the argument lost people.
Cover
100
The number
91
Who was studied
78
What changed
66
Your guess
60
The reveal
57
By district
45
Implementation
34
Cost per student
26
The ask
fewer than one in five
19
Method
15
Close
skipped ahead
22
ILLUSTRATIVE DATA · Readers who reached each scene, %
02 · Where readers lingered
Ten scenes. Three held the room.
6s
9s
14s
31s
27s
12s
22s
8s
17s
5s
Scene 4 · Your guess
Readers stopped to commit. Longest dwell in the story.
Scene 5 · The reveal
They stayed to read why their guess was wrong.
Scene 7 · By district
The chart they explored. Two in three tapped a district.
ILLUSTRATIVE DATA · Median seconds per scene · ● interaction present
03 · Completion by audience
Completion, by who was reading.
District leaders
41%
Partners
37%
Board
29%
Staff
24%
Press
12%
General public
8%
Each audience gets its own link, so each audience gets its own number. The readers with a stake finished.
ILLUSTRATIVE DATA · Reached the final scene, by audience link
04 · Channel and device
Reach up, reading down.
Arrived
Finished
LinkedIn
940
4%
Launch email
620
12%
Partner sites
210
18%
Direct link
150
21%
68%phone
27%desktop
5%
The channel that brought the most people brought the fewest readers. Not whether the work travelled. Whether it landed.
ILLUSTRATIVE DATA · Arrivals and completion by channel · share of readers by device
Then, the next edition.
Specific recommendations: what to move, what to cut, where the ask goes. The first story sets the baseline; everything after is measured against it.
A first read · two business days

Send us something you published. We will read it and tell you where it loses people.

Send it to dawn@onautumn.com
Front matter
How much sits before the substance.
The ask
Where it falls, relative to where readers realistically still are.
Density
Where it spikes.
The buried number
The one worth opening on, and one rewritten headline as a taste.
A diagnosis of what is happening to it. The scene plan is the product.
New York Times-level storytelling, without the newsroom.