How do I stop doom scrolling?

Nobody decides to lose two hours. You just never decide to stop, and the feed does not have an end.

How do I stop doom scrolling?

To stop doom scrolling, aim to reduce rather than eliminate, put a stopping cue back into the apps that removed it, and choose in advance what happens in the time you get back. That last part is not a nice extra. It is the part with the best evidence behind it, and it is the part almost everyone skips.

The instinct is to go cold turkey, delete everything, and treat it as a purity test. The research does not support that instinct, which is oddly good news, because the all or nothing version is also the version people abandon by Thursday.

The evidence on deleting versus cutting back

Three findings sit next to each other and tell a coherent story.

Deleting does little on its own. In a systematic review and meta-analysis published in Scientific Reports in 2025, Laura Lemahieu and colleagues pooled 10 studies covering 4,674 people and found social media abstinence had no significant effect on positive affect, negative affect or life satisfaction. Every result was null.

Cutting back does a bit. A meta-analysis of 32 randomised controlled trials covering 5,544 people, published in SSM Mental Health in 2025 by Kaitlyn Burnell and colleagues, found social media restriction improved subjective wellbeing with a Hedges g of 0.17. That is a real effect, and it is a small one. The authors themselves concluded that the field should focus on specific mechanisms rather than broad time restrictions.

Cutting back plus adding something beats both. This is the interesting one. In a randomised trial of 642 people published in the Journal of Public Health in 2022, Julia Brailovskaia and colleagues ran four arms: reduce social media by 30 minutes a day, add 30 minutes a day of physical activity, do both, or do neither. At six month follow up, the combination group was down to 75.6 minutes a day of social media, against 94.4 in the reduction only group and 114.8 in the control. Their depression scores were lower than every other arm, and their physical activity was still elevated at 224 minutes a week versus 158 in the activity only group.

Two weeks of doing both was still visible half a year later. Neither half managed that alone.

Why the feed is hard to leave

There is no stopping cue. Researchers use that term specifically. A glass empties. An episode ends. An infinite feed is engineered so that moment never arrives, and a 2024 study in the International Journal of Communication by Nikhila Natarajan documented teenagers manually reinserting friction to recreate the cue the design removed.

The cleanest demonstration is about autoplay. In a randomised study of 76 Netflix users verified against their own account data, Brennan Schaffner and colleagues found that simply disabling autoplay cut average daily watching by 21 minutes and shortened the average session by nearly 18. One stopping cue, restored, bought back a third of an hour a day. That result is currently a preprint, so hold it lightly, but the direction matches everything else here.

The reward is unpredictable. In a study published in Nature Communications in 2021, Bjorn Lindstrom and colleagues analysed over a million posts from more than four thousand people and found that social media engagement follows the same reward learning mathematics that describes animals foraging for food. In a controlled experiment with 176 participants, artificially varying how many likes a post received changed how quickly people posted again. This is not an analogy borrowed from a book about casinos. It was measured directly.

You do not know how much you are doing. People underestimate their own phone use substantially, and even after being shown their past usage, participants in Allcott's 2022 study still underestimated their future use. Awareness is weaker than it feels.

What to actually do

Move	Why it works	Evidence
Set a target of less, not none	Restriction beats abstinence, which shows no effect at all	Burnell 2025, Lemahieu 2025
Turn off autoplay and infinite scroll where the app allows	Restores a stopping cue the design removed	Schaffner 2024, preprint
Switch the screen to greyscale	Removes the reward from the surface itself	Rahmillah 2023, up to 38 minutes a day
Use an app limit, but expect modest returns	About 22 minutes a day, roughly 16 percent	Allcott 2022
Get the phone out of the bed	Each hour of in bed screen use is associated with 59 percent higher odds of insomnia	Hjetland 2025
Decide what fills the hour before you free it	The only intervention here still working at six months	Brailovskaia 2022

On that fifth row: in a study of 45,202 Norwegian students published in Frontiers in Psychiatry in 2025, Gunnhild Johnsen Hjetland and colleagues found each additional hour of screen use after getting into bed was associated with 59 percent higher odds of insomnia and 24 minutes less sleep. Notably, it did not matter much which app. Social media was no worse than anything else. It is the screen in the bed, not the specific feed.

That study is observational, so it shows association rather than proof of cause. The causal evidence points the same way: in a randomised trial of 111 people published in BMC Medicine in 2025, Christoph Pieh and colleagues capped smartphone use at two hours a day for three weeks and found insomnia scores improved by 18 percent, alongside a 27 percent drop in depression scores. Anxiety, for what it is worth, did not move significantly. An honest article says that part too.

The hour is the whole problem

Every finding above converges on one practical point. Removing the phone creates an empty hour, and an empty hour with a phone in the room is not a plan. Brailovskaia's combination arm is the only intervention in this entire body of research that was still working six months later, and the only thing it did differently was put something in the gap.

So name the gap. The first ten minutes awake. The ninety before sleep. The pause after finishing something hard. Then decide what goes there, specifically, in advance, and tell someone what you decided. A Crew makes that a daily habit rather than a one off intention.

Make the hours you got back visible

Screen time dashboards already tell you the number, and the number has not changed your behaviour, because a total at the end of the week is not a decision in the moment. It is a report card for a test you already failed.

Door 24 works on the moment instead, and on the record afterwards. Progress is proof. You capture a Proof, a photo or a voice note or a line of text, timestamped onto a dated timeline. The evening you got back. The book you actually read. The morning you got up without reaching for the phone first. A Daily Quest gives you one deliberate thing to do with the ten minutes you would otherwise have given away without noticing, which is the replacement behaviour the research says matters most.

Your Growth Score is a 42 day rolling average rather than a streak, which matters here more than for almost any other habit, because you will have a bad evening. Everyone does. A bad evening bends the trend. It does not delete the month.

The Freedom Ledger counts hours as well as money, and for this one the hours are the number that lands. Two hours a night, across a month, put somewhere else.

If you want the wider frame, read why abstinence is not enough and how to build sober habits that stick. On measuring change without a streak, see streaks versus rolling averages. For the product side, how Door 24 handles doom scrolling covers it.

You are not weak for losing the evening. You are up against a design with no edge in it. Put an edge back, and put something on the other side of it.

Sources

• Lemahieu L et al., "The effects of social media abstinence on affective well-being and life satisfaction," Scientific Reports, 2025.
• Burnell K et al., "The effects of social media restriction: Meta-analytic evidence from randomized controlled trials," SSM Mental Health, 2025.
• Brailovskaia J et al., "Experimental longitudinal evidence for causal role of social media use and physical activity in COVID-19 burden and mental health," Journal of Public Health, 2022.
• Lindstrom B et al., "A computational reward learning account of social media engagement," Nature Communications, 2021.
• Schaffner B et al., "An Experimental Study of Netflix Use and the Effects of Autoplay on Watching Behaviors," preprint, 2024.
• Natarajan N, "Whether, How, and Why Teens Insert Frictions Into Social Media's Infinite Scroll," International Journal of Communication, 2024.
• Allcott H et al., "Digital Addiction," American Economic Review, 2022.
• Rahmillah FI et al., "Evaluating the Effectiveness of Apps Designed to Reduce Mobile Phone Use," Journal of Medical Internet Research, 2023.
• Hjetland GJ et al., "How and when screens are used: comparing different screen activities and sleep in Norwegian university students," Frontiers in Psychiatry, 2025.
• Pieh C et al., "Smartphone screen time reduction improves mental health: a randomized controlled trial," BMC Medicine, 2025.
• Satici SA et al., "Doomscrolling Scale," Applied Research in Quality of Life, 2023.
• Ofcom, "Online Nation 2025," 2025.