Avoid it or get used to it? The first two weeks off caffeine are a different project from the rest
Every guide to quitting caffeine gives you one strategy. But avoiding coffee entirely is the right move in week one and the wrong move by month six — and the people who never switch back end up with a permanently coffee-shaped hole in their life. Here's the phase test that tells you which one you're in.
Last week I published a piece about going caffeine-free next to someone who still drinks coffee, and the advice at the centre of it was this: don’t remove the morning ritual, duplicate it. Stand at the same counter at the same hour with your own brew in your hand while your partner makes theirs. Same room, same minute, different mug.
A reader named Alan W., in Perth, wrote in to say that this failed for him. Not partially — it failed twice, and he could tell me exactly why. Standing at the counter beside a real coffee in his first fortnight wasn’t neutral. It was a side-by-side comparison that he lost every single morning. What actually worked was the opposite of what I’d written: he took his drink out to the back step and stayed out of the kitchen entirely for about three weeks. Then he came back in, and the two-mug arrangement worked exactly as described.
I conceded the point in the comments, and then I kept thinking about it, because it isn’t a small correction. It’s a hole in nearly everything I’ve written on this site.
Alan W. was right and I was wrong
Here’s what I got wrong, precisely. I wrote a piece of phase-two advice and published it as day-one advice.
“Duplicate the ritual” assumes a level of steadiness you may simply not have in week one. My own first fortnight — logged day by day at the time — happened to be a stretch where I was too foggy to want much of anything, so I never had Alan’s experience of standing next to a coffee and wanting it badly. I generalised from a quirk of my own timing and dressed it up as a method.
And it isn’t just that piece. Look at the pattern across the whole cluster. The office pot. The 6am airport. The kitchen ten feet from your desk. Brunch with people who all drink coffee. Mornings with young kids. Every one of those gives you one strategy for one situation, and not one of them says this advice has an expiry date on the front and a start date on the back.
Week one and month six are two different projects
The thing Alan surfaced is that quitting caffeine isn’t one project with one correct approach. It’s at least two, and they want opposite instructions.
Phase one is triage. The goal is to get through the days when your head hurts and your attention won’t hold still, and the correct strategy is to reduce the number of moments in which anything is being decided. Avoidance is genuinely right here.
Phase two is durability. The goal is a life that doesn’t require special arrangements, and the correct strategy is more or less the reverse: be around coffee, on purpose, and let the wanting show up and pass without being obeyed.
The reason this matters is that the transition between them is silent. Nothing tells you the first project is over. And the failure mode isn’t people who avoid too little — it’s people who avoid indefinitely, because avoidance keeps working just well enough that there’s never an obvious day to stop.
Why avoidance is the right call early
There’s a study I keep coming back to, and it’s the most useful thing I’ve read on this question. Researchers took data from two studies in which smokers trying to quit carried devices and logged what was happening around them in real time, many times a day, over the first two to four weeks — 61 people in one study, 207 in the other. Because the reports were made in the moment rather than remembered later, you get an unusually honest picture of the circumstances in which a quit actually breaks.
The finding on availability is stark. In the moments when cigarettes were easily to hand, the odds of slipping were enormously higher than when they weren’t — on the order of 37 times higher in the smaller study and about 12 times in the larger one. Being around other people smoking mattered too, but far less. The single biggest situational risk was simply that the thing was there and gettable.
Two caveats, and they’re not small. This is tobacco, not caffeine — nicotine dependence is a considerably more serious business than a coffee habit, and I’m not going to pretend the two are equivalent. And these are observational data, so they describe the circumstances of lapses rather than proving what caused them.
But the shape transfers, and it matches what people tell me constantly: the mornings that go wrong are not the mornings of weak character, they’re the mornings when there was a full pot sitting there. If the biggest lever in the early weeks is availability, then reducing availability is a rational thing to do — not a confession that you can’t handle it. Alan going out to the back step wasn’t a retreat from a fight. It was declining to hold the fight in the worst possible venue.
The cue is doing something measurable — in caffeine, specifically
Most of the useful research here is about alcohol and tobacco, which is a real limitation. But there is a caffeine-specific study, and it’s worth knowing about, because it answers the question of whether the smell of coffee is a genuine event or something you’re dramatising.
Researchers had 65 daily coffee drinkers come in after 18 hours without caffeine, gave them gum that either did or didn’t contain caffeine, and separately told them either accurately or inaccurately what they’d been given. Then they presented coffee-related cues — the sight of coffee, and the smell and sound of it being made — alongside neutral ones.
Coffee cues increased craving regardless of what people believed they’d been dosed with. More striking, the visual coffee cue measurably raised average and maximum heart rate compared with the neutral cue. Not “participants reported feeling stirred up” — their heart rate went up, at the sight of a drink they hadn’t touched.
The reason I care about the heart-rate result is that it changes what we’re asking of people. “Just push through the smell” sounds like a request for a bit of resolve. What the data suggest is that you’re being asked to sit with a small physiological response while it runs its course — which deserves decent conditions, and in week one that may mean not being in that room at all. It’s also why something warm in your hand during the withdrawal stretch helps more than it should: you’re not replacing a chemical, you’re occupying a hand and a slot.
Why avoidance is the wrong answer forever
Now the other half, from the same smoking study — and this is the part almost nothing written for a general audience mentions.
Alongside measuring what made a lapse more likely in the moment, the researchers counted something else: how many times each person had already been in a situation where cigarettes were available and had not smoked. Every one of those is, in learning-theory terms, a practice repetition — the cue showed up, the behaviour didn’t follow.
Those accumulated repetitions predicted a substantially lower chance of lapsing later on. Roughly half the odds, in both studies.
So the same data set says both things at once, and this is the crux of the whole article: the presence of the cue is what puts you at risk right now, and successfully surviving the presence of the cue is what makes you safer later. The authors’ own conclusion is that cessation programmes should include both — stimulus control and practice resisting in real environments. Not one or the other. Both, sequenced.
There’s a second reason avoidance alone doesn’t finish the job, and it comes from laboratory work on extinction. The learning that suppresses a behaviour doesn’t seem to erase the original association; it lays down a new one that is tied to the context in which you learned it. Return to the original context and the old behaviour comes back — the renewal effect. Researchers in that tradition note plainly that this is why people can extinguish their responses to drug cues in a clinic and relapse when they get home.
This is animal and clinical research on substances far stronger than caffeine, and I want to be clear I’m borrowing a frame rather than reporting a finding about coffee. But the frame explains something people describe to me all the time: they were completely fine for eight months, and then the first morning back in an office ambushed them. They hadn’t learned “I don’t drink coffee.” They’d learned “I don’t drink coffee in my kitchen.” The office was untrained.
One boundary before I go further, because it matters more than the mechanism does. None of this is a template for quitting alcohol or nicotine, and it isn’t meant to be. Deliberately walking up to a cue is a reasonable thing to suggest for coffee precisely because the worst realistic outcome is a cup of coffee and a poor night’s sleep. That asymmetry is doing most of the work here. If the substance in question can hurt you, this is a conversation for a clinician, not a website about chicory.
The honest ceiling: deliberately using cue exposure as a treatment has a mixed record. A meta-analysis in alcohol use disorders found effects that were small to moderate at best, strongest on how long people went before relapsing, and rated the underlying evidence low quality. So this is not a magic procedure — the narrower claim is what holds up: in real life, over the first weeks, repeatedly not doing the thing while the thing is available appears to help.
The failure mode nobody counts: the coffee-shaped hole
Here’s the version of “success” I think we undercount, because from the outside it looks like a win. Someone quits caffeine. Two years later they haven’t touched it. They also won’t go into a café, won’t take the early flight because there’s nothing for them at the airport, decline the client breakfast, and have not made coffee for a guest in their own home since 2024. Ask them and they’ll tell you they’ve beaten it.
I don’t think they have, quite. They’ve built a life with a room removed. And it’s fragile in a specific way: because nothing was ever extinguished in the presence of the cue, the machinery is still fully loaded, and it will fire the first time circumstances put them in front of a pot with no exit. Which is roughly what people describe when they drink coffee again after a long gap and are shocked by how strongly it lands.
The goal was never to spend your life avoiding a bean. It was to stop needing it.
The phase test: four questions instead of a rule
So I’m not going to give you a number. I’m going to give you a test, and you can run it whenever you like.
1. Is the cue a summons or is it information? This is the main one. When you smell coffee, does it arrive as an instruction — go and get one — or does it arrive as a fact about the room, the way you’d notice someone’s toast? Information means phase two is available. An instruction means stay out a while longer.
2. Can you be there for ninety seconds without narrating it? The aroma peaks early in the brew. If you can stand in the kitchen through that window while thinking about something else, you’re most of the way there. If you’re delivering an internal monologue about how well you’re coping, you’re still working.
3. Has the withdrawal actually cleared? Do not run this experiment while your head hurts. The withdrawal timeline has a shape — it tends to peak in the first couple of days and ease over roughly a week, though it varies a lot — and testing yourself during the worst of it tells you nothing except that you were uncomfortable. Withdrawal is a recognised clinical entity, not a moral test, and there’s no credit for taking it standing up.
4. Would slipping right now feel like losing a streak? If a lapse would wreck a number you’re emotionally invested in, the stakes are too high for a first attempt. Run the test when the cost of it going badly is “that was interesting” rather than “I’ve ruined it.”
Three yeses and a low-stakes fourth: go back in.
And notice what the test replaces. Alan needed about three weeks; I effectively needed none, for reasons that were pure luck. Both are ordinary, because the spread on this sort of thing is enormous — the most-cited study of everyday habit formation found the time to reach 95% of each person’s own plateau ranged from 18 days to 254, median around 66, and that was for something like drinking a glass of water after breakfast. “Two weeks” is a shape, not a schedule. If you’re at week five and the smell still shouts at you, you are not behind — you’re inside a normal range that nobody advertises. What people describe at the one-month mark fits that: the dramatic part is over, and what’s left is unglamorous consolidation.
How to actually re-enter
The renewal research has one clear practical implication — extinction doesn’t generalise well across settings — so do this one context at a time, easiest first, and don’t assume that clearing one clears the rest.
- Your own kitchen, while someone else brews. Home turf, exit available, no audience.
- A café, ordering something you actually want. Not a grudging herbal tea. Go in wanting your order.
- The office kitchen at pot o’clock, once the café is boring.
- The 6am airport, which is the hardest one, and worth planning rather than improvising.
Two mechanics that make a real difference. First, have your drink already in your hand when the cue arrives — the difference between standing there holding something warm and standing there empty-handed is out of all proportion to how small it sounds. Second, go in on purpose. An exposure you chose is a repetition. The same event happening to you by accident is just a bad morning.
This is also where the format of your drink should change, which nobody mentions because most guides never separate the phases at all. Phase one wants the lowest-effort thing going: an instant option you can make with a kettle in another room, since the aim is to spend as little time as possible at the source of the smell. Phase two wants the opposite — something that takes a few minutes and fills the slot coffee used to, so you’re in that kitchen doing a real thing rather than loitering. Loose roasted chicory or dandelion root is the cheapest and most hands-on way in. Among the packaged herbal roasts, Dandy Blend and Pero dissolve fastest, while ground blends like Teeccino go into a drip basket or press — which is the point for re-entry, because the ritual you’re rebuilding is making something, in that room, at that hour. Caveats worth stating: the ground blends cost more per cup than loose chicory, most contain barley or rye and so aren’t gluten-free, and the chicory inulin in nearly all of them unsettles some guts at first.
What I’d change about everything else I’ve written here
Rather than quietly patching the record: the practical advice in the getting-started pieces here is phase-two advice. Duplicate the ritual, stand your ground at the office pot, order confidently at the airport, redirect the gesture rather than refuse it — all of it assumes the cue has stopped being a summons.
If you’re in your first fortnight, read those as a description of where you’re heading, not instructions for tomorrow. Tomorrow’s job is smaller: get through the day, keep the coffee out of arm’s reach, and get the mornings themselves working — light, water, movement, and a look at why the afternoon crash was always partly the caffeine’s doing. Then come back in. That part isn’t optional, and I’d never said so before.
The bottom line
Avoidance and exposure aren’t rival philosophies. They’re the first and second halves of the same project, and almost every guide — mine included, until a reader in Perth pushed back — collapses them into one instruction and picks the wrong half for whoever happens to be reading.
So get out of the room early if you need to; the evidence that availability is the dominant situational risk is about as clear as this kind of evidence gets. Then, once the smell has stopped issuing orders, go back in deliberately, one setting at a time, with something good already in your hand. The repetitions where the cue shows up and you carry on regardless are the ones that make this permanent.
Leave if you have to. Just put a date on coming back.
Sources & further reading
- Does extinction of responses to cigarette cues occur during smoking cessation? — O'Connell, Shiffman & DeCarlo, Addiction (2011)
- The impacts of caffeine administration, expectancies, and related stimuli on coffee craving, withdrawal, and self-administration — Shephard & Barrett, Journal of Psychopharmacology (2022)
- Context-induced relapse after extinction versus punishment: similarities and differences — Marchant et al., Psychopharmacology (2019)
- Cue exposure therapy for the treatment of alcohol use disorders: a meta-analytic review — Mellentin et al., Clinical Psychology Review (2017)
- How are habits formed: modelling habit formation in the real world — Lally et al., European Journal of Social Psychology (2010)
Reader conversation (6)
We read every response. Selected reader notes below.
Well, this is a strange thing to open on a Tuesday morning. Thank you for taking it seriously rather than filing it under “one bloke’s mileage may vary”. I’d add one correction to my own comment now that I see it laid out properly: I didn’t actually plan the three weeks. I fled the kitchen on about day four in a mild panic, and the reason I came back in at all was that my wife asked whether I was going to keep having breakfast on the step through winter. So the “put a date on it” instruction is better than what I did — I got the sequence right by accident and only because someone prodded me. If I’d been left alone I suspect I’d still be out there.
That detail is worth more than the original comment, and it’s slightly alarming: the thing that got you back inside was a passing remark from someone else. Which means the people who live alone, or whose household never comments, are the ones most likely to drift into the permanent version without noticing. I’d been thinking of the date as a piece of tidiness. You’ve made me think it’s the actual safeguard.
I am the coffee-shaped hole and I did not enjoy reading about myself. Six years off caffeine, entirely successful by any measure you’d put in a magazine, and I have not been inside a coffee shop since 2020 — I wait outside while my daughter gets hers, and we both pretend this is normal. I’d always framed it as a preference. Reading “a life with a room removed” was unpleasant in the way that accurate things are. I don’t know that I’ll do anything about it, but I’ve stopped telling myself it’s a preference.
I’m sorry it landed that hard, and I’m glad you said it, because I wrote that section slightly worried I was inventing a problem. On the “don’t know that I’ll do anything” part: I’d gently say there’s no obligation. Six years is six years and you’re not in danger. But if you ever did want the room back, the version I’d suggest is absurdly small — go in with your daughter, order a peppermint tea, stay four minutes, leave. Not to prove anything. Just so the building stops being a place you can’t go.
Careful with this one, I think. I work in addiction services, not with caffeine but with alcohol, and “go and expose yourself to the cue on purpose” is advice that needs its guardrails stated more loudly than you’ve stated them. The extinction findings you’re citing come from structured contexts with support around the person. Told to someone with a genuinely disordered relationship to a substance, “deliberately go and stand near it” can be actively harmful. For caffeine I think your framing is fine and probably useful. I’d want a clearer sentence saying it doesn’t generalise upward.
That’s a correction I’ll take without argument, and you’re right that I leaned on the tobacco and alcohol literature for the mechanism while writing for people whose stakes are much lower. To say it plainly here since it isn’t plain enough above: nothing in this piece is a model for quitting alcohol, nicotine or anything else, and the reason exposure is a reasonable thing to suggest for coffee is precisely that the worst realistic outcome is a cup of coffee and a poor night’s sleep. That asymmetry is doing a lot of work and I should have named it.
Day nine here, and the phase test has done something useful that I want to name: it told me to stop testing myself. I’d been going into the office kitchen every single morning to prove I could handle it, treating each one as a small exam, and then feeling wrecked by ten o’clock and unable to explain why. Question three is the one that got me — I’m still getting headaches in the afternoons, so by your framing I’ve been running the experiment in exactly the conditions that guarantee it tells me nothing. Staying out of there this week. It feels like permission rather than defeat, which is a new feeling.
The context point explains something that has bothered me for a year. I quit in early 2025, was completely fine at home within a month, and then in February took a job with a proper old-fashioned office where there’s a pot going from seven in the morning. I lasted eleven days. I’d assumed I’d been kidding myself the whole time and the quit was never real. “The office was untrained” is a much better description of what happened, and considerably kinder. Second attempt going better, and this time I’m treating the office as its own project rather than as a test of the previous one.
Eleven days in a room with a permanent pot is not evidence that the previous year was fake — it’s close to the most demanding version of the situation there is, and you walked into it with no practice in that setting. Worth adding for anyone in the same position: a constantly refreshed communal pot is genuinely different from a machine somebody runs once at eight. There’s no quiet period, so there’s no moment where the cue isn’t present, and the availability finding says that’s the hard part.
Small practical thing that might help others, on the “have your drink already in your hand” point. I could never get the timing right — my partner’s coffee was always made before I’d even filled the kettle, so I’d be standing there empty-handed at exactly the wrong moment. Fixed it by making mine the night before and reheating, which sounds joyless and is honestly fine. The chicory doesn’t suffer. What matters is that I’m holding a hot mug when the grinder starts rather than waiting three minutes for one, and those three minutes were where all the trouble lived.
Have something to add? Email us and we may include it in a future update.