Research · From Default to Author · Journey
Research · Journey × University of New Haven

From Default to Author.

Every nudge system hits the same ceiling. This one didn’t, because students stopped being targets of the architecture and started authoring it. A nine-month SMS pilot, with the same students surveyed in the fall and again in the spring.

The ceiling

A decade of optimization hasn’t moved the number.

Digital nudge systems get 30 to 40% initial engagement, then most are abandoned within thirty days. Campus email does worse, roughly 7%. Better copy, better timing, better design: the number holds.

The field treats this as a content problem. We think it’s a category error. Population nudge theory personalizes by sorting people into bins and serving matched content. The student’s job is to be classified and served.

Students who engaged most didn’t describe being nudged. They described being reminded of who they said they wanted to be, in their own words, at a moment they could act.

The pilot · N = 100 · nine months

What changed across the year.

74%

Weekly engagement

vs. roughly 7% for campus email, 30 to 40% for comparable platforms.

72 → 85

Holistic health score

Composite, 0 to 100, intake to year-end. An 18% rise.

50 → 6

High-risk students

High-risk first-years fell from 50 to 6 in a single year.

2,111

Confirmed actions

Appointments booked, events attended, exercises done, not opens.

The authorship signature

The items that measure authorship moved the most.

The same 19 students answered the same questions in the fall and again in the spring. A halo effect starts high and stays high. These started near zero and climbed to a majority, which is what a relationship built from a student’s own accumulating language should look like.

Fall Spring

“Made it easier to do what I already wanted to do”

14%
71%

“Arrived at the right moments”

0%
57%

“Reflected goals I actually care about”

14%
57%

“Felt personalized”

0%
50%

“I’m scrolling less”

43%
80%

Overall satisfaction also rose from 3.4 to 4.1 out of 7. The item that moved most was about enablement, not intimacy. Students agreed Journey made it easier to act far more than they agreed it understood them.

Awareness · top-2-box, fall to spring

Students noticed more about themselves.

Aware of my mental-health patterns12 → 40%
Aware I need breaks during study sessions38 → 60%
Aware of opportunities to connect with others25 → 53%
Aware of career-prep steps to take13 → 27%

Doomscrolling survey · separate sample · n = 13

Why students can’t put the phone down.

A separate Fall 2025 survey, run through journeynudge.com. The pattern underneath the habit: it isn’t laziness, it’s stress, boredom, and not knowing where to start.

After a scroll, they feel

Guilty / frustrated4
Informed / satisfied4
Numb / zoned out2
Slightly regretful2
Anxious / overwhelmed1

The biggest barrier

Too stressed, need the escape4
It’s how I stay connected3
Don’t know where to start3
Tried before and failed1
Afraid I’ll miss news1

What pulls them in

Boredom8
Habit6
Procrastination5
Anxiety / stress2
Loneliness · FOMO2

Most students picked “about half the time” or “most of the time” when asked how often a quick check turns into 10+ minutes of scrolling.

In their own words

The praise and the criticism, kept side by side.

Verbatim, de-identified. We left the hard feedback in. It’s the part that tells us what to build next.

“If I spent less time on my phone, I’d focus on self-care in the mornings. I’d wake up slowly, stretch, play some music while getting ready, and enjoy breakfast without rushing or checking my phone.”
Doomscrolling survey · what they’d do differently
“When I told Journey I felt lonely cuz I had no friends it just told me to try joining clubs and I’m like 😑 cuz it was insensitive.”
Criticism · fall · “change one thing”
“I would dedicate more time to other things. I’d read more, play the piano more, or dedicate more time to my studies.”
Doomscrolling survey · what they’d do differently
“Increase the response speed to near-immediate, and potentially develop into a mobile app; add more features other than just messages.”
Criticism · fall · “change one thing”
“Encouragement on why I should want to work, a reminder of my goals.”
What they’re held back from
“I’d sit in a quiet area and complete a few hours of work to catch up, and learn more about things that will actually further my education and career.”
Doomscrolling survey · what they’d do differently

The finding that cuts the other way

A system that mirrors your values also mirrors your distortions.

The same mechanism that ends the engagement ceiling, reflecting a student’s own language back to them, opens a new problem. A student who arrives catastrophizing hands the system a distorted draft to mirror. We measured this as sycophancy across 1,422 logged conversations. It was not uniform.

9%
of guidance conversations overall
25%
in relationship guidance, where framing is most one-sided
6%
where claims were grounded in checkable facts

Authorship without an honest broker isn’t a feature with a bug. It’s a design that isn’t finished.

What we won’t oversell

Real movement. Not yet a product students would miss.

17% → 21%

Asked how they’d feel if Journey disappeared, the share answering “very disappointed, I rely on it” rose only from 17% to 21%. The common product-market-fit benchmark is 40%. We’re not there yet. The system earned real behavioral and attitudinal change across a year without yet becoming something most students would miss, and the distance between those two facts is the clearest guide to what to build next.

Paired, not pooled. The longitudinal survey is the same 19 students answered twice, fall and spring, paired by participant, so every change is within-person.
Measured, not derived. Every figure is taken directly from the form export, de-identified, with the participant key held by the lead researcher.
Small substantive base. Eight students answered the deeper items in the fall and fifteen in the spring; the within-person read rests on the eight who answered in depth both waves. Treat the survey as directional, not definitive.
Staff saw more than students reported. A December staff snapshot reported higher figures (57% “significantly less” scrolling, 71% personalized) than students rated themselves. That gap is a finding, not a discrepancy to explain away.

Read the full study, or see it run on your campus.

The complete paper covers the system design, the JITAI framework, the help-seeking pathway, and the honest-broker problem in full.

Back to the research overview