The Comparison Tax: A Social Media Detox Protocol

Upward comparison, rather than screen time, is what makes a feed hurt. This page carries the diagnostic that separates the two problems, the seven-day baseline you capture before day one, the account-by-account triage into unfollow, mute and keep, a week-by-week thirty-day protocol, what to do when the urge hits, and re-entry rules that stop comparison returning.

What is a social media comparison detox, and how long should it last?

A social media comparison detox is a fixed-length break from the feeds that rank people against one another, run to interrupt upward social comparison rather than to reduce screen time, and thirty days is the length that gives most people a readable result. The distinction matters because the two problems have different fixes. A screen-time problem responds to timers, grayscale and app limits. A comparison problem survives all of those, because the damage is done in the seconds you spend looking at one person's life, and a fifteen-minute daily cap still delivers plenty of those seconds. Thirty days works as a unit for a practical reason rather than a mystical one. The first week is mostly withdrawal: irritation, phantom reaching, a strange amount of free time you have no plan for. The second week is where mood readings start to mean something. By the third and fourth weeks those readings are stable enough to set against your baseline without withdrawal noise sitting on top of them. A weekend break or a seven-day challenge measures your withdrawal, not your life without the feed. Running longer than thirty days may still be worth doing for its own sake, though the extra weeks tend to tell you less than the first four already did.

Leon Festinger described social comparison as a basic human process in 1954, arguing that people evaluate their own abilities and opinions by measuring them against others whenever no objective standard is available. Thomas Ashby Wills set out downward comparison theory in 1981, separating comparison with someone judged worse off, which he argued can raise how a person feels about their own situation, from the upward direction, comparison with someone judged better off. A feed built to surface the most engaging posts from several hundred accounts loads the upward direction almost exclusively. On the length question there is genuine disagreement. Hunt Allcott, Luca Braghieri, Sarah Eichmeyer and Matthew Gentzkow recruited about 2,700 American Facebook users ahead of the 2018 midterm elections, then paid a randomly chosen subgroup to deactivate their accounts for the four weeks ending just after the vote while a comparable group carried on as usual. The paid group reported small but measurable gains in subjective wellbeing, along with something more useful: they were still using Facebook substantially less once the paid month was over. The month changed the habit itself, which is the more durable result and the one worth aiming at. Others hold that anyone whose mood tracks a feed should leave permanently. Both camps agree on the first move: stop, and measure what changes.

Is thirty days a rule or just a convention?

Thirty days is a convention, chosen because it clears the withdrawal period with room left over to read a stable signal. Nothing about the number is fixed. Twenty-one days will do if your baseline was steady and your withdrawal was mild. Forty-five is reasonable if week two was chaotic for unrelated reasons, such as a deadline or an illness that muddied the readings. Pick a length before you start, and write it down.

Does a comparison detox mean leaving every platform?

No. Leave the platforms where comparison actually happens to you, which for most people is one or two rather than all of them. Someone who feels nothing looking at short video but goes flat after ten minutes of photographs from people they went to school with has a specific problem, and shutting down a messaging app or a hobby forum alongside it adds cost without adding information. Cut what hurts.

What counts as a failed detox?

A detox fails when you cannot tell at the end whether anything changed, which usually means the baseline was never captured or the rules quietly leaked. Checking a partner's phone, reading screenshots sent by friends, or browsing logged-out profiles all keep the comparison stream running while letting you feel you abstained. Returning early is not a failure. Returning early and calling the question answered is.

How does upward social comparison on a feed actually make you feel worse?

Upward social comparison on a feed makes you feel worse because it hands you a rigged sample: the best moments of several hundred lives, edited and timed for effect, stacked into one column and delivered faster than any offline environment could manage. Comparison itself is ordinary. You use it to work out whether your salary is fair, whether your child is developing normally, whether your work is any good. Offline it is episodic and self-correcting, because you see the same people on bad days as well as good ones, and you see enough of their circumstances to discount what you are looking at. A feed strips both corrections. You get the wedding and not the argument in the parking lot afterward, the finished manuscript and not the four years of silence, the apartment in a city you cannot afford with no mention of who paid the deposit. Then it repeats the exercise a few hundred times before lunch. The result is a slow recalibration of what you think normal looks like. Nothing dramatic happens in any single instance, which is exactly why people report feeling worse after scrolling without being able to point at anything that went wrong.

The research here is thinner and more contested than headlines suggest. Melissa Hunt and colleagues at the University of Pennsylvania, in work published in 2018, monitored 143 undergraduates for a baseline week and then randomly assigned them either to limit Facebook, Instagram and Snapchat to ten minutes per platform per day for three weeks or to carry on as usual. The limited group reported lower loneliness and fewer depressive symptoms at the end. That result is suggestive rather than settled: 143 students at a single university is a narrow base, three weeks is a short follow-up, and every outcome was self-reported. An earlier survey by Hui-Tzu Grace Chou and Nicholas Edge, published in 2012, reported that people who had used Facebook longer were more likely to agree that others were happier and had better lives, a correlation that cannot establish which way the arrow points. The field's central argument is worth knowing before you read any of it. Jean Twenge has held since her 2017 book iGen that smartphones and social media drove the rise in teenage depression. Amy Orben and Andrew Przybylski reanalyzed three large datasets in 2019, covering more than 350,000 adolescents between them, and found the association explained at most 0.4 percent of the variation in adolescent wellbeing, a smaller effect than the one attached to wearing glasses. Neither camp denies that some individuals are badly affected. They disagree about how many, and about whether an average that small can say anything about you. For a method of judging small psychological studies, including sample size, sham arms and the limits of self-report, see Reiki Benefits: Evidence-Based Research on Pain, Anxiety, Depression & Sleep.

Is all social comparison harmful?

No. Comparison is how people calibrate in the absence of an objective yardstick, and downward comparison, looking at someone worse off, often lifts mood in the short term, though the effect is inconsistent and some people come away feeling worse. Upward comparison can motivate as easily as it deflates, and which one you get depends largely on whether the gap looks closable. A colleague two steps ahead in a career you are also in tends to energize. An unreachable life tends to flatten.

Why do algorithms increase upward comparison?

Ranking systems promote the posts that draw strong reactions, and the posts that draw strong reactions are disproportionately the extremes: the new house, the transformation, the award. An unranked chronological feed of the same accounts would carry far more ordinary content. The algorithm is not selecting upward comparison on purpose. It is selecting engagement, and the top of a person's life is what people engage with.

Does knowing a post is staged stop the effect?

Knowing a post is staged reduces the effect without removing it, because the comparison happens faster than the correction. You register the image, feel the drop, then apply what you know about lighting, selection and paid partnerships several seconds later. The felt response has already landed. This is why media-literacy arguments rarely fix a comparison problem on their own, and why removing the input works better than reasoning about it.

How do you tell whether comparison, rather than screen time, is your real problem?

You tell a comparison problem from a screen-time problem by checking whether your mood tracks the content you saw or the minutes you spent. Log both for a week. If two hours of video, games or group chats leaves you tired but level, while eleven minutes of photographs from three specific people leaves you flat for the rest of the evening, volume is not what is hurting you. Run three checks. First, the account test: can you name the three to five accounts that reliably produce a dip? A volume problem has no such list. Second, the aftertaste test: a volume problem feels like dullness and lost time, while a comparison problem feels like a specific downward tug attached to a person, often with an urge to post something in response. Third, the substitution test: replace twenty minutes of feed with twenty minutes of a book or a walk and see whether the relief is proportionate. Someone with a volume problem feels better roughly in proportion to the time saved. Someone with a comparison problem feels much better after cutting a small amount of one particular kind of looking, and that disproportion is the signature you want.

Most people have some of both, and the mix decides the protocol. A person with mainly a volume problem gets more from timers, notification changes and a phone that sleeps in another room than from an account cull. A person with mainly a comparison problem can hold their screen time constant and still recover, provided the specific inputs are gone. The measurement discipline matters more than the instrument: fix everything else you can, change one thing, and hold that change long enough to read it. Almost nobody manages that on day one. A detox begins every January alongside a new gym membership and a new diet, and by February nobody can say which of the three did anything. Change one thing at a time or learn nothing. One warning about self-diagnosis. If low mood, panic or anxiety is interfering with work, sleep or relationships regardless of what you looked at, that is a threshold for professional assessment rather than a behavior protocol, and Reiki for Anxiety & Depression: Calming Positions, Research & When to Seek Therapy sets out where the line sits.

What does a comparison hangover feel like physically?

People describe a sinking sensation in the chest or the stomach that arrives within seconds of a particular post, often with a brief tightening in the jaw or throat and a sudden interest in checking their own profile. It fades in minutes but leaves a flatness that can last hours. The speed is the tell. Ordinary tiredness from a long session builds gradually. This one drops.

Can you have both a volume problem and a comparison problem?

Yes, and it is the common case. Long sessions raise the number of comparison exposures, and comparison drives longer sessions, because a dip prompts more looking. Treat them in order. Cut the comparison inputs first, since that change carries the larger emotional return per unit of effort, then decide whether the remaining hours still bother you once the sting has gone out of them.

Does the diagnosis change what you do?

It changes almost everything. A volume problem is solved with friction: app limits, notifications off, the phone charging outside the bedroom. A comparison problem is solved with selection, meaning which accounts reach you at all. Someone who applies volume tools to a comparison problem ends up taking a shorter, sharper dose of the same harm, then concludes that detoxing does not work for them.

What should you measure before day one so you know whether it worked?

Measure seven days of baseline before you change anything, recording four things: a single mood score taken at a fixed hour, the number of times you open a feed, the minutes those opens total, and a one-line note naming whatever produced the worst moment of the day. Take the mood score at the same time every day, on the same 0 to 10 scale, worded the same way. Consistency of the instrument counts for more than its sophistication, and a rating scribbled on paper at nine each evening will outperform three different apps used inconsistently. Add a fifth column if you can manage it: a running list of accounts that produced a dip, written down at the moment it happened rather than reconstructed later. That list becomes your triage sheet in week one. During the baseline week, change nothing. Do not cut back early, do not start a diet, do not begin a new exercise routine, do not start or stop a medication without talking to your prescriber. Every variable you move during the baseline is a variable you cannot rule out afterward, when you are trying to work out what the thirty days actually did.

Self-report is a weak instrument, and pretending otherwise is how people talk themselves into results they did not get. A mood score reflects the weather, an argument at work and how well you slept, all of which move independently of your feed. Two habits make it usable anyway. Keep the scale and the hour fixed so the noise stays roughly constant, and note the obvious confounders in the margin: a deadline week, a bout of illness, seasonal darkness, a menstrual cycle phase if that is a reliable factor for you. Sleep deserves a row of its own, since sleep and mood move together and late-night feed use sits directly upstream of both. Record roughly when you fell asleep. One number, no detail, no plan attached to it yet. Photograph your baseline sheet at the end of week one so you cannot quietly revise it later, because memory will edit any record it can still reach, and it edits in the direction that makes the effort look worthwhile.

How long should the baseline week be?

Seven days, minimum. That span captures a working week and a weekend, which matter separately because feed use and mood both change shape on days off. Fourteen is better if your schedule is irregular, or if week one contained something unusual. Fewer than five will not average out an ordinary bad Tuesday, and one bad Tuesday can swallow the entire signal by itself.

Should you use your phone's screen-time report?

Use it for minutes and opens. Trust it for nothing else. Built-in reports undercount browser use, miss time spent inside a feed embedded in another app, and handle shared devices inconsistently. They remain the least effortful source of a number you would otherwise guess at badly. Copy the weekly figure into your own sheet by hand, so you still have it after the report rolls over and takes its history with it.

What if measuring makes you scroll less already?

Expect it. Being watched changes what is watched, including when the watcher is you, so a baseline week almost always comes in lower than a normal one. That biases the comparison against finding an effect, which is the safe direction to be wrong in: if the numbers still move, the move is real. Note the drop. Read your final figures against a baseline you already know was flattering.

What does a 30-day comparison detox look like week by week?

A thirty-day comparison detox runs in four distinct weeks: removal and triage, the trough, replacement, and re-entry design. Week one is mechanical. Delete the apps from the phone, log out of every browser session, remove the shortcuts, then work through your following list account by account, sorting into unfollow, mute and keep. Do the triage in week one even though you are not looking at the feed, because doing it later means doing it while the pull is back. Week two is the trough and asks for nothing except that you do not quit. Expect irritability, a sense of being out of the loop, and several hours a day you have no plan for. Week three is where the freed time gets a job: a standing morning slot that is not the phone, an evening activity that gets you out of the house, one social contact made directly rather than through a feed. Week four is spent writing the re-entry policy, in advance and on paper, while you are still clear-headed enough to be strict. Keep taking the daily mood score through all four weeks without looking back at the earlier numbers.

Each week has a characteristic way of going wrong. Week one fails on half-measures: the app deleted from the phone but still logged in on a laptop, or a personal account closed while a second account stays open. Week two fails on substitution. The freed attention moves to news, shopping or a different feed, and the comparison carries on under another name. Week three fails on premature confidence, since feeling better by day eighteen is the standard moment for deciding the problem is solved and reinstalling. Week four fails on vagueness. A resolution to be more mindful is not a policy, and it will not survive the first dull evening. The morning slot is the one to design first. Whatever fills the ten minutes after waking tends to set the shape of the day, and ten empty minutes beside a charging phone have an obvious way of filling themselves. Pick something that needs your hands: coffee made slowly, a page of writing, floor stretches. The only real requirement is that it be specific enough to begin without deciding anything.

What happens on day one specifically?

Day one is administration rather than willpower. Set aside forty minutes. Take your final baseline reading, uninstall the apps, log out everywhere including the browser and any tablet or work machine, switch off every notification from those platforms, and tell the two or three people who message you there where to reach you instead. Then run the account triage while the dip list is still fresh in your mind.

Why is week two the hardest?

Week two is hardest because the novelty of week one has gone and the benefits have not arrived. The habit loop is intact, the freed hours still feel empty rather than available, and there is no reward yet you can point at. Nobody has published a reliable dropout curve for this, but the second week is where people who abandon the attempt tend to say they abandoned it. Expect it. Knowing the shape of the dip in advance is most of the defense against falling into it.

Can you shorten it to fourteen days?

Fourteen days will change how you feel but will not tell you much, because you finish while still inside the withdrawal period. You measure the ache, not the life underneath it. A two-week break is still worth doing as a trial run if thirty days feels impossible, and it often makes the longer attempt easier later on. Treat its readings as provisional.

Which accounts should you unfollow, mute, or keep, and how do you decide?

Sort every account into unfollow, mute or keep using one criterion: what happens in the ten seconds after you leave its post. Unfollow the accounts that reliably produce a dip and to whom you owe nothing. That covers most strangers, most aspirational accounts, and a surprising number of people you last spoke to years ago. Mute the accounts that produce a dip but carry an obligation: a sibling, a boss, a client, the friend whose feelings a visible unfollow would hurt. Instagram and Facebook both state in their own help pages that they do not tell anyone they have been muted or snoozed, and the equivalent control on other large platforms works the same way, though it is worth checking the current help page for whichever one you use. Muting removes the content and leaves the connection intact, which is what an obligation account needs. Keep the accounts that leave you wanting to do something rather than wanting to be someone, along with the purely functional ones: local news, a hobby group, the account that posts when your gym changes its hours. Go down the list once at roughly three seconds per account and mark each one without deliberating. Speed is doing real work here. The first reaction is the one you have not yet negotiated with, and thirty seconds of reasoning will talk you back into following almost anybody.

Three obstacles stop people finishing the audit. The first is guilt. Unfollowing reads as a verdict on a person, and it is not one: the account is a stream of images that happens to be produced by someone you may love, and removing the stream changes nothing about the relationship except that you now hear their news from them. The second is the belief that you must keep watching a competitor, an ex, or a former colleague for information. Ask what decision that information has ever changed. If the honest answer is none, it is comparison wearing the costume of research. The third is the one nobody volunteers: the private pleasure of watching someone you resent do badly. That is not research either, and it charges more than the others. Expect the audit to take under an hour for a following list of a few hundred, and expect to cut far more than you predicted.

Should you unfollow friends and family?

Mute them rather than unfollowing. The dip is real and the content should go, but a visible unfollow inside a family or a close friendship group creates a second problem you then have to manage, usually at a wedding. Muting removes the posts from your feed without notifying anyone or breaking the connection, and it reverses in a second on the day you want to look.

What about accounts you follow for work?

Move them off your personal feed entirely. Looking at them carefully does not work, and everyone tries it first. A separate logged-out browser tab, an email newsletter, or a dedicated account you open twice a week all deliver the same information without the surrounding stream. The trouble with work-justified following is simple: the professional accounts sit next to the personal ones, and you cannot open one without seeing the other.

How many accounts is a normal cut?

There is no measured figure to give you here, and anyone quoting one is guessing. What people report is that they cut far more than they had predicted, and that the oldest accounts, the ones carrying a decade of drifted-apart acquaintances, lose the most. There is no target worth hitting either. A list of forty that contains four painful accounts needs a smaller cut than a list of eight hundred that feels fine and still leaves you flat every evening.

What do you do when the urge to check hits during the first ten days?

When the urge to check hits during the first ten days, do not argue with it: timestamp it, delay ten minutes, and spend those minutes on a physical action chosen in advance. Urges behave like waves rather than like rising pressure. They build, peak and fall away on their own if nothing feeds them, and clinical rules of thumb usually put that arc at under half an hour, though the figure is a teaching device rather than a measured constant. The only skill required is outlasting one without making a decision. Write the delay action down before day one, so it is not a choice you have to make while craving. Three options work for most people: step outside and walk to a fixed landmark, do something with your hands that needs both of them, or make direct contact with one person by voice or message. Keep the list to three items, because a longer one becomes another thing to browse. Some people use a short physical routine such as EFT tapping for the first sharp spike of envy. The evidence behind it is thin and disputed, and it has not been tested to a clinical standard for this use, so treat it as a way to occupy two minutes rather than as a treatment. EFT Tapping: The Points, the Script and What the Evidence Shows sets out what has and has not been shown. Do not treat the delay as willpower spent. You are not refusing the feed forever, only declining it at eleven forty on a Tuesday, and that is a far smaller thing to win.

Urges attach to contexts far more than to feelings, which is why willpower is the wrong lever. Spend the first three days writing down where each urge happens rather than trying to resist it, and a short list emerges: the queue, the bathroom, the ninety seconds before a meeting starts, the moment the credits roll, the first minute after waking. Once you have that list you can redesign the contexts instead of fighting them. Put a book where the phone used to sit. Leave the phone in a bag rather than a pocket during the specific hour that produces most of the reaching. Set the screen to grayscale so an idle unlock pays less. Sean Parker, Facebook's first president, told an Axios audience in November 2017 that the early team had asked itself how to consume as much of a user's time and conscious attention as possible, and had answered with a social validation feedback loop. Take the confession for what it is worth. It still describes a designed system, and a designed system yields to redesign long before it yields to resolve.

How long does a typical urge last?

An unfed urge rises and falls rather than climbing without limit. Clinical teaching materials commonly describe a craving peaking and easing inside roughly fifteen to thirty minutes when it is not acted on, but those numbers are rules of thumb rather than measurements, and yours may run shorter or longer. A ten-minute delay works less because it outlasts the whole wave than because it converts a prohibition, which invites rebellion, into a postponement, which does not. If ten minutes passes and the pull is undiminished, set another ten. The wave image carries a name in the relapse-prevention literature, urge surfing, from G. Alan Marlatt and Judith Gordon's 1985 work on relapse. Set an actual timer, since a craving makes four minutes feel like twenty.

What if the urge is really loneliness?

Answer it directly rather than through a feed. A good many checking urges turn out to be requests for contact wearing another shape, and a feed answers them with the appearance of contact while leaving the need where it was. Whether that describes yours is something only your own log can tell you. Send one voice message, call someone during a walk, or arrange something for the weekend. If loneliness is the constant rather than the trigger, the detox is a smaller part of the problem than the isolation is.

Is it cheating to check on someone else's phone?

Yes, and it is the most common way a detox quietly ends. Looking over a partner's shoulder, asking a friend to show you a post, or reading forwarded screenshots keeps the comparison input running while letting you record a clean day. Decide in advance how you will answer when someone offers to show you something, because you will be offered, usually inside the first week.

How do you return to social media without the comparison creeping back?

Return one platform at a time, on a written policy you wrote before reinstalling anything, and keep scoring your mood daily for at least two weeks after each return. The policy needs four rules. First, one platform per week, so that any change in your numbers can be attributed to something. Second, fixed windows rather than ambient access: two named slots a day, each with an end time, rather than a permanent open door. Third, the browser rather than the app wherever the site allows it, because a browser is slower, uglier and far easier to close. Fourth, no posting for the first two weeks back, since posting starts the loop of checking who responded and pulls you into the same comparison from the other side. Reinstall the first platform on a day you can watch yourself, not on a Friday night. Then set the two weeks after re-entry against your thirty-day numbers and your original baseline. If the score returns to baseline within a fortnight, either the policy is too loose or that platform is not compatible with you.

The failure mode after a successful detox is drift rather than relapse. Nothing dramatic happens. The window creeps outward by five minutes at a time, a few accounts get followed back after an interaction, the app returns to the home screen because opening the browser was annoying, and a couple of months later the numbers sit where they started with no single decision you can point at. Two habits catch drift early. Re-audit the following list once a quarter using the same three-second rule, and keep taking the mood score one day a week indefinitely, since a single weekly number is cheap and shows a trend across months. Some people find that one platform never becomes safe again, and removing it permanently while keeping the others is a reasonable settlement rather than a defeat. If the numbers never move across the full thirty days, the feed was probably not the driver, and it is worth asking a clinician what is.

Which platform should you bring back first?

Bring back the one that scored lowest on your dip list, usually a messaging or interest-based platform rather than a photo feed. Restoring the least harmful one first gives you a clean read on whether re-entry itself unsettles you, before you test the platform you actually struggled with. Leave the worst offender until last. You may well find you never want it back, and that is a result rather than a lapse.

How do you stop the follow list refilling?

Set a standing rule rather than deciding case by case. A workable one: you may follow a new account only after a week has passed since you first wanted to, and only if you can say what it gives you. Pair that with the quarterly audit at three seconds per account. Refilling happens through dozens of small unexamined decisions, so the countermeasure has to be automatic.

What if the numbers do not improve at all?

Take that as useful information rather than as a failed experiment. A flat result across thirty clean days means your feed was not the main driver of how you feel, which narrows the search to sleep, work, isolation, physical health or a mood condition that deserves assessment. Check that the baseline was honest and the rules held, then look elsewhere rather than repeating the detox harder.

Frequently Asked Questions

Why do I feel worse after scrolling Instagram even when nothing bad has happened?

Because the harm arrives in small, unremarkable pieces rather than in one identifiable event. Each post shifts your sense of what an ordinary life contains by a fraction, and after a few hundred of them your own circumstances read as worse than they did twenty minutes earlier, with nothing specific to attribute the change to. The absence of a bad moment is characteristic of comparison rather than evidence against it. Keeping a written list of which accounts precede the drop usually makes the pattern visible inside a week.

Is it better to delete the apps or just log out of them?

Do both. Either move alone leaves an easy path back. Deleting removes the icon, the notifications and the muscle memory of the home screen position, which is where most unplanned opens come from: a thumb arrives at the right corner of the screen before you have decided anything. Logging out of the browser closes the loophole people reach for around day three. Keep the password in a manager rather than in your head, so getting back in takes deliberate effort instead of four seconds of autofill.

Should I tell people I am taking a social media break?

Tell the handful of people who actually reach you through those platforms. Skip the announcement post. A short direct message with an alternative contact method prevents the practical problem of missed invitations, which is the most common reason people reinstall during week one. A public farewell does something worse. It creates an audience for your abstinence, and a private experiment becomes a performance you cannot leave without explaining yourself.

Will I lose followers or hurt my business if I stop posting for a month?

A personal account will shed a small number of followers and recover them, which costs nothing that matters. A business account is a different calculation, and the honest answer depends on whether the platform is a real acquisition channel or a habit wearing a business costume. Check where your last twenty customers actually came from before deciding. If a platform genuinely drives revenue, keep posting through a scheduler and hand the checking to someone else, or use a separate browser profile with the feed itself out of reach.

Can I still use social media for work during a detox?

Yes, provided you keep a hard separation between using a platform and browsing one. Post through a scheduling tool, answer direct messages in a single dedicated window each day, and reach the content you need through search or a direct link rather than through the feed. Use a different browser profile or a different device for work access so the personal context never appears. No work task should require you to scroll a ranked feed, and most of them do not.

Is comparison anxiety the same thing as social media addiction?

No. Comparison anxiety concerns what you see and how it makes you rank yourself. Compulsive use concerns how often you check and how hard stopping is. The two travel together and reinforce each other, but they respond to different interventions: selection fixes comparison, friction fixes compulsion. Neither term is a formal diagnosis. Anxiety that persists and interferes with work, sleep or relationships regardless of your phone use is a reason to see a clinician rather than to redesign a habit.

What if I feel more anxious after quitting social media rather than less?

Some increase in anxiety during the first week is ordinary, and it usually reflects withdrawal plus the loss of a distraction that was covering something else. For many people it eases as the second week passes, though there is no fixed timetable and nothing guarantees it will. If it does not settle, or if it worsens, take that seriously rather than pushing through. A feed can be an avoidance mechanism, and removing it can leave grief, loneliness or an untreated anxiety condition uncovered. That is a point to talk to a professional.