The Single Notification That Lifted Workout Adherence by 36% (And What It Means for Your Fitness App)

In 2021, a research team at the University of Illinois ran a controlled field experiment on a fitness app called ActiveTrack. They tested which combination of variables — personality type, message content, and notification timing — produced the highest exercise plan completion rates across 89 adult users over two months.

The message content — whether it was emotionally encouraging or logically structured — had no statistically significant effect on adherence. Neither did the specific wording, tone, or framing.

One variable dominated everything else: whether users received a single notification 30 minutes before a scheduled workout. Without it, mean adherence was 63.4%. With it: 86.1%. A difference of 22.7 percentage points, statistically significant at p=.001.

One notification. Thirty minutes. Nothing else in the message design mattered comparably.

What the study tested

Sun et al. (2021) built ActiveTrack on Apple ResearchKit and recruited participants through gym trainers. The 89 active users were randomly assigned into a 2×3×2 factorial design — two personality types, three message types, and two reminder conditions — and tracked over eight weeks.

The three dimensions:

Personality type was the MBTI thinking/feeling dimension only. Feeling-type participants (driven by emotion and interpersonal factors) showed 89.8% baseline adherence. Thinking-type participants (driven by logic and analysis) showed 71.7%. A gap of roughly 18 percentage points before any intervention was applied.

Message type tested three conditions: emotional messages ("You are doing a fabulous job!"), logical messages (claim plus supporting data), and no message. The hypothesis was that message framing would interact meaningfully with personality type — emotional resonating with feeling-types, logical with thinking-types. The data didn't support it. Neither message type produced statistically significant adherence lift over no message (p=.24).

Repetition was the advance reminder: one notification sent 30 minutes before a scheduled workout session, or no reminder. This was the variable that mattered.

The finding and its specific mechanism

The adherence lift from the advance reminder — 63.4% to 86.1% — is a 36% relative increase in workout completion rates from a single design decision that adds effectively zero engineering complexity and zero user burden.

The mechanism is timing, not content.

At 30 minutes before a planned session, a user has made a cognitive commitment to the time slot but hasn't yet resolved competing options for it. They haven't started something else that will run long. They haven't mentally written off the session. The 30-minute window is the highest-receptivity moment: the decision about whether to train is still genuinely open, and a specific prompt with a concrete anchor — this session, in 30 minutes — can tip it.

After a session is complete, the same prompt has zero relevance. Before a session is on the schedule at all, the prompt lacks the anchor that makes 30 minutes meaningful. The advance reminder works because of when it arrives relative to a specific commitment, not because of what it says.

This explains why generic daily workout reminders are a structurally different intervention from what Sun et al. tested. "Don't forget to work out today" is not an advance reminder — it's an unsolicited nudge with no specific session anchor. The evidence behind 30-minute advance reminders doesn't extend to it.

The personality finding: who the reminder actually helps

The most useful secondary finding in the study is how the adherence lift distributed across personality types.

Thinking-type users — those who showed 71.7% baseline adherence without a reminder — gained 27.3 percentage points with the advance reminder (p<.001). Their adherence jumped to roughly 85%, reaching statistical parity with feeling-type users.

Feeling-type users — those who showed 89.8% baseline adherence — showed a change of -0.83 percentage points. Effectively zero.

The advance reminder has near-zero effect on people who were already completing their planned sessions, and a large, statistically robust effect on the people who weren't. It doesn't push people who are already adhering to do more — it rescues people who plan to train and then don't.

This is the correct distribution for an adherence feature. It helps the people who need help and costs nothing for people who don't. The feature doesn't impose intervention burden on high-adherence users; their behavior is unchanged. It specifically addresses the gap between intention and follow-through in users for whom that gap was real.

What most apps do instead

The fitness app notification ecosystem is built around assumptions that the research doesn't support. This is part of a broader pattern in how fitness apps are designed for the wrong user.

Daily generic reminders. Most apps send daily or near-daily workout prompts on a fixed schedule, independent of whether the user has a session planned, whether they've already trained, or whether they're in any condition to receive the prompt usefully. These aren't advance reminders — they're habituating noise that users learn to ignore, contributing directly to the intervention fatigue documented in the JITAI literature.

Post-miss guilt nudges. "You haven't trained in a few days." "Your streak is about to expire." These notifications arrive after a behavioral gap, not before a planned session. They're timed to the moment of lowest receptivity — when the user has already moved on — and they apply controlled motivation pressure (guilt, streak anxiety) rather than a specific advance reminder anchored to a concrete future commitment. From the JITAI framework, these are interventions deployed at vulnerability states without the timing or framing that makes interventions effective.

Re-engagement campaigns. Apps that detect absent users and send high-frequency re-engagement sequences are compounding the fatigue problem. Nahum-Shani et al. (2017) documented the pattern: median app logins fell from 8 in month one to 1 in month two. By month six, 64% of users in a tracked study had effectively stopped engaging. Sending more notifications to declining users is the intervention most likely to confirm the decision to disengage.

High message volume. Volume is inversely related to receptivity over time. Every unnecessary notification is a withdrawal from the attention budget. When the user finally receives a notification that is well-timed and relevant, the pool of attention available to it has been depleted by weeks of low-value prompts.

The JITAI principle: provide nothing

Nahum-Shani et al. (2017) introduced the just-in-time adaptive intervention (JITAI) framework specifically to address the question of when health behavior support should be delivered — and when it should not be.

The framework identifies two distinct user states: vulnerability (heightened risk of a negative outcome, like skipping sessions) and opportunity (heightened receptivity to a positive change, like a strong training week). These require different interventions and, critically, different thresholds for intervention at all.

The principle that most directly applies to notification design: "provide nothing" is a valid intervention option. When a user is adhering well, the correct response from the app is silence. A notification at that moment is not neutral — it signals that the user requires external prompting, which is precisely the wrong message to send to someone building autonomous motivation. It also adds to intervention burden, accelerating the fatigue that eventually drives disengagement.

From this framework, the advance reminder works not just because of its timing but because of its specificity: it intervenes at exactly the decision moment, on exactly the relevant behavior, and then provides nothing else. It's a targeted tool, not a broad-spectrum notification campaign.

The fitness apps that send five notifications per day are running the opposite strategy: low-specificity, high-frequency interventions applied uniformly regardless of user state. The JITAI research predicts the outcome — declining engagement, habituation, and eventually dropout — with reasonable accuracy.

What the research actually supports

Synthesizing Sun et al. and Nahum-Shani, the notification design that the evidence backs looks like this:

Advance timing is the only variable that matters. Thirty minutes before a scheduled session. Not morning, not evening, not triggered by absence. Triggered by a specific scheduled session.

Message content is secondary to timing. Sun et al. tested emotional vs. logical framing and found neither produced significant lift. Write something encouraging, keep it brief, don't overthink the copy.

Low frequency, high specificity. One reminder per planned session. Nothing when the user is not vulnerable. Nothing when the user has already completed the session. Silence as a default, not as a gap between high-frequency nudges. This is what sustained adherence actually requires from a support system.

User-controlled scheduling. The advance reminder works when it's anchored to a session the user planned. A system that lets users schedule sessions — and then reminds them 30 minutes before those sessions — is delivering the intervention the evidence supports. A system that prescribes when users should train and then reminds them is mixing intervention types.

Avoid post-miss intervention as a primary re-engagement strategy. It's timed wrong, it applies the wrong motivational mechanism, and it adds burden at a moment of low receptivity.

The irony in the fitness notification landscape is that the apps sending the most notifications are doing the most damage to their own engagement metrics, while the notification strategy the research supports — one well-timed advance reminder per planned session, silence otherwise — is both simpler and more effective.

For the research on why the type of motivation these notification patterns build matters as much as the adherence lift itself, see why fitness motivation isn't your problem.

Fortiv uses a 30-minute advance reminder for scheduled sessions — anchored to sessions you plan, silent when you're already on track. Join the waitlist to be among the first to try it on iOS.

---

For the broader framework on why mHealth apps lose most of their users by month six — and what intervention design principles address it — the Nahum-Shani 2017 paper is the primary reference. For the predictable dropout arc that the timing window in this post is designed to interrupt, see Why You Quit Your Fitness App in Month 4.