INSIGHTS
Thinking
Notes on behavioral science, design thinking, and the kinds of product decisions they ought to inform.
RECENT WRITING
Research
More Personalization Didn't Help: A 2026 Pilot of an Adaptive Stress App
A May 2026 pilot of a just-in-time adaptive stress app split 46 employees into high- and low-personalization groups. The personalization made no measurable difference, sensor data collapsed to under 60%, and one physiological marker moved the wrong way. The lessons for adaptive health products are specific.
May 30, 2026
Research
Anchoring Beats Reminders: What a Four-Arm Trial Says About Building Habits
A four-arm RCT of 161 college students tested reminders, financial incentives, and anchoring for building physical activity habits. The winning combination pinned the new behavior to an existing routine and paid out only when the timing matched. The design lessons are specific.
May 30, 2026
Behavior Change
Why 25% of health apps get deleted after one session
A 52-study scoping review maps four ways product teams measure engagement, and why the most common one misses the actual problem.
May 23, 2026
Behavior Change Techniques
Half of digital health behavior change programs have no theory behind them
An overview of 23 systematic reviews found that half of digital health interventions don't use an explicit behavioral framework, and it shows in the outcomes.
May 23, 2026
Research
What a Gamification RCT Actually Tells Product Teams
A recent JMIR RCT isolated gamification as the only variable in a mHealth study. Here is what that design actually proves, what it doesn't, and what it means for product teams building behavior-change apps.
May 16, 2026
Research
BCT Umbrella Review: Social Support Wins, and More Features Don't Help
An umbrella review of 20 systematic reviews covering 224,135 participants found social support was the most consistently effective behavior change technique across health domains. It also found that adding more BCTs to a product doesn't make it more effective.
May 16, 2026
Research
Push Notifications Are Losing Their Push: What a Micro-Randomized Trial Found
A micro-randomized trial of 566 health app users found that standard push notifications lose 11% of their effect when a user received one the day before. Novel messages don't habituate the same way. The implications for notification design are specific and actionable.
May 15, 2026
Research
Social Norms Messaging in Health Apps: What 85,759 Participants Actually Tell Us
A 2025 Nature Human Behaviour meta-analysis of 89 RCTs found that social norms messaging has a near-zero effect on health behavior once publication bias is corrected. Here's what that means for teams building social features into health products.
May 12, 2026
research
WhatsApp groups as a behavior change mechanism, what 81 studies actually found
A 2026 scoping review and realist synthesis maps how group chats drive (or don't drive) behavior change in digital health. The mechanisms are not what most product teams think.
May 6, 2026
research
What the Coach factorial trial actually tells us about which BCTs do work
A 5,419-person factorial RCT of six digital behavior-change components, what landed, what backfired, and what it means for product teams shipping behavior change software.
May 5, 2026
research
Persuasive design didn't predict efficacy in 92 mental-health-app RCTs, and that should bother us
A meta-analysis of 92 RCTs and 16,728 participants found no association between persuasive design principles and either engagement or clinical outcomes. What that means if you ship behavior-change software.
May 2, 2026
research
Older adults stayed engaged with a medication app for a median of 595 days, and almost everything we assume about that demographic is wrong
A real-world analysis of 250 adults aged 65+ on the Perx Health app found 95% medication adherence and retention measured in years, not weeks. What that means for the way we design for older users.
May 2, 2026
RESEARCH
What a 5,419-person trial just told us about which BCT actually moves behavior
A factorial RCT published April 2 tested six behavior change components on alcohol, diet, physical activity, and smoking. The winner is boring, and most product teams aren't shipping it.
April 18, 2026
FRAMEWORKS
How to actually build BCT 1.4 (action planning) into a digital product
Action planning is one of the most-used behavior change techniques in health apps. Most implementations are broken. Here's what the research says, and what a working version actually looks like.
April 18, 2026
FRAMEWORKS
COM-B in practice: a walkthrough with a real product
Most product teams treat COM-B as a training slide. Here's what it actually looks like when you apply it to one behavior, one product, one user.
April 15, 2026
BEHAVIOR CHANGE
Why people stop engaging in week two
Most health-tech products have a cliff around day 10. Here's what I've seen cause it, and why usability scores won't predict it.
April 14, 2026
INTRO
Welcome to Insights
What this blog is for, why I'm writing it, and the kinds of posts you'll see here. Short version: applied behavioral science for product teams who want rigor.
April 14, 2026