There Is No Right Number of Minutes
What the screen-time evidence actually supports
If you go looking for the scientifically correct amount of daily social media, you will find numbers. You will find several, they disagree, and the best-designed studies suggest the question is partly wrong. Here is the actual state of the evidence, because we had to read it to set a default, and you may as well get the summary.
Where "1–2 hours" comes from
The largest study in the area (Przybylski & Weinstein, 2017; 120,115 fifteen-year-olds) found an inverted U: wellbeing rises with moderate use, peaks, then declines. The weekday peak for smartphone use landed around 1 hour 57 minutes. Notably, zero use tracked with lower wellbeing than moderate use. A US replication in ~35,000 children found the same shape: 1–2 hours a day looked slightly better than either abstinence or heavy use.
The caveat is size. The same lab later showed that technology use explains at most 0.4% of the variance in adolescent wellbeing — statistically about the same as eating potatoes (Orben & Przybylski, 2019; >300,000 respondents).
Where "30 minutes" comes from
One experiment: 143 undergraduates limited to 10 minutes per platform per day for three weeks showed reduced loneliness and depression (Hunt et al., 2018). Real result, small sample, and the control group also improved on anxiety, which suggests self-monitoring was doing some of the work. A later trial that held students to 10 minutes a day for most of an academic year found no significant effect at all.
What the 2025 meta-analyses say
Three meta-analyses of randomized trials landed in 2025, pooling roughly 17,000 participants between them. Reduction interventions improve wellbeing by d ≈ 0.2 — small, positive, consistent.
The finding that matters more: dose-response mostly fails to appear. Longer interventions didn't reliably beat shorter ones, and total abstinence did not beat partial reduction. In one meta-analysis, limiting use produced a significant effect (g = 0.33) while quitting entirely did not (g = 0.15, not significant). The largest single experiment ever run — 35,000 people paid to deactivate Facebook or Instagram for six weeks (Allcott et al., 2024) — moved a wellbeing index by 0.06 standard deviations. That was full removal of the platform.
If more restriction doesn't buy more benefit, there is no optimal dose to find.
What predicts harm better than duration
Two things outperform the clock in the literature:
Timing. Use in the hour before sleep is a much sharper predictor of poor outcomes than total daily time, and disturbed sleep carries much of the association between social media and depression. This is the strongest single finding in the field.
Pattern. Compulsive checking and distress-on-withdrawal track mental health far better than hours do. How often you reach for the phone tells you more than how long you stay.
Even the May 2026 Surgeon General's warning on screen use, which treats the issue as a genuine public health concern, declines to name a daily hour limit. That refusal is informative.
How HabitApp uses this
The default earning rate in the app works out to about 60 minutes a day — half the evidence-supported ceiling, double the tightest defensible floor. You can set it stricter; the app will not recommend going below 30, because nothing in the literature supports it.
There is no total-lockout mode, because abstinence doesn't outperform limiting. There is an optional pre-sleep window, off by default and always overridable, because the sleep finding is strong enough to act on and not strong enough to impose. And the trends view plots your opens per day alongside your minutes, because check-frequency is the better measure and nobody else charts it.
A strong finding earns a strong recommendation. It never earns a lock you didn't ask for.
Sources: Przybylski & Weinstein (2017), Psychological Science. Przybylski, Orben & Weinstein (2020), JAACAP. Orben & Przybylski (2019), Nature Human Behaviour. Hunt et al. (2018), Journal of Social and Clinical Psychology. Liu et al. (2025), Behavioral Sciences, 20 RCTs. May, Malouff & Meynadier (2025), 10 RCTs. Allcott et al. (2024), PNAS. US Surgeon General's Warning on the Harms of Screen Use, May 2026.