How to track a habit you're cutting down, not quitting

The rule always gets written somewhere specific. On the walk back from the shop. In the taxi. Standing in the kitchen at half eleven with the light off, phone in hand, deciding that this is the last night of this. The rule is reasonable. It is a good rule. It starts tomorrow.

And it works, for a while. That’s the part nobody mentions. Most of these rules work for a fortnight, some of them for two months. Then something ordinary happens and the rule turns out to have been made of paper. So you write a better one. The next rule is more specific, more clever, better defended against exactly the thing that broke the last one.

That is the loop I want to look at, because there is a real reason it runs, and it is not the reason most people land on.

The rules you have already set yourself

Whatever you are trying to do less of, you have almost certainly built rules for it. They tend to arrive in the same handful of shapes, and the shapes travel across behaviours that otherwise have nothing to do with each other.

There is the number. Two, and never on a weeknight. There is the time gate: nothing before 8:30pm, nothing after 10pm. There is the place rule, which is the one that always sounds most sensible, because it feels structural rather than moral. Only out of the house. Never in the bedroom. Only at the café, never at the desk.

Then the rules get inventive. You make the thing more expensive, so cost does the braking for you. You buy the version you do not much like, on the theory that the taste will stop you. You put the phone in a box in the kitchen and set the box for 10pm. You hand someone else the passcode. You start a timer at the first one and you are allowed a window, after which you stop. You go greyscale, so the whole thing is duller and less worth picking up.

Read a long enough list of these and something becomes obvious. The rules do not get more effective as they go. They get more inventive. Each one is a smarter patch on the specific hole that the last one leaked through, which means each one is designed against the past and not against tomorrow.

The shape	Drinking	Phone	Caffeine
A number	No more than two, and not on a weeknight	Under two hours a day	Two cups, both before noon
A time gate	Nothing before 8:30pm	Nothing after 10pm	Nothing after 2pm
A place rule	Only out of the house, never at home	Not in the bedroom, not at the table	Only at the café, never at the desk
A friction tax	Only the bottle above a certain price	The phone in a box in the kitchen	Only if I grind the beans by hand
A quality dodge	Only the one I do not much like	Greyscale, so it is duller	Half caf
An external lock	Only ever with two specific people	Someone else holds the passcode	The tin stays at work
A timer	Two hours from the first, then stop	An app timer that cuts in at the limit	A timer on the afternoon window

None of it is stupid. Every one of those rules took real thought, and the thinking is being spent at a considerable rate.

Why the bar keeps moving

Two things are going on, and they compound.

The first is that the bar is set against a number you are guessing at. You decide that two is a reasonable limit, or two hours, or two cups, and the decision feels grounded because it is anchored to a sense of what you currently do. But that sense is a memory of a pattern, not a record of one, and memory keeps stories rather than counts. The nights you stopped at one are more memorable than the ones you did not, because stopping at one felt like something and the other nights felt like nothing much. So the baseline you are correcting from is already flattering, and the limit you set against it is closer to where you already believe you are than to where you are. It looks reachable because it was drawn on the wrong map.

The second thing is that the rule and the breach are made by two different people. The rule gets written in the calm part of the day. It is a considered, slightly optimistic piece of policy from someone with nothing much going on. The breach happens somewhere else entirely, in a moment with its own weather, and the person in that moment was not consulted and does not feel bound. This is why the rules read as clever in the abstract and evaporate in contact. They were drafted by the wrong version of you, about a situation they were not in.

The person who sets the limit and the person who breaks it are never in the room at the same time.

So the failure has a location, and the location is not you. You are setting a bar in the dark. You cannot see the ground you are standing on, so you keep adjusting the height, and each adjustment is a guess about a distance you have never measured.

Fixing the problem with the problem

Screen time shows this most clearly, because the apparatus is visible. Nobody builds a physical machine to manage their drinking. Everyone builds one to manage their phone, and it goes in layers.

• The blocker. It works, and then you learn where its edges are.
• The second blocker, for the thing the first one does not cover, running alongside it.
• The lock handed to someone else, which lasts until the evening you decide to ask for the passcode back, or work it out yourself.
• The workaround, then the counter-workaround. You block it properly this time, at a level you cannot casually undo, and then you spend an evening undoing it properly.
• The tracker for the tracker, so you can see whether the blockers are working.

Every one of these is a thing to maintain. And the maintenance has a shape that should be familiar, because it is the same shape as counting through an evening out: one drink, then a glass of water, then wondering whether it has been an hour yet, then doing the arithmetic again. The behaviour costs what it costs. The management of the behaviour costs more, and it runs all day.

Most people arrive here at the honest complaint, which is not that they failed. It is that the whole arrangement is more tiring than the thing it was arranged around. You end up orbiting the problem, tuning the system that was supposed to solve it, thinking about the phone constantly in the service of thinking about the phone less.

The lights are still off. All that has changed is the quality of the equipment you adjust the bar with.

Before you set a bar, find the number

There is a move here that almost nobody makes, because it does not look like progress. You stop trying to change the behaviour, and you find out what it actually is.

That is an awareness habit: tracking with no target attached, no bar, no number to move. You record the days. You do not try to make the record look better. The point is not to change anything yet, and if that feels like giving up, the last several attempts at changing it did not change it either. Awareness is a cycle type rather than a preliminary, and it exists because your sense of a behaviour is unreliable in one direction, which quietly wrecks every limit you set on top of it.

The awkward part is that it has to run long. The first weeks of any awareness cycle are the least honest ones, because attention alone moves behaviour, and because you know you are watching. Seven days of tracking gets you seven days of good behaviour and a number you will not believe by March. 30 days is the usual shape of a cycle and it is often still short for this. Three months gets you far enough past the observation that you forget to perform for it, and what settles out then is closer to the truth than anything the first fortnight showed. What tracking without a target actually does is replace an estimate with a record, and the record needs time to stop flinching.

A few years ago I tracked my drinking this way. Not to cut down, just to see, because a beer at the end of a hard day had become routine without me noticing the routine part. After a couple of months I looked properly: I had been drinking on more than 60% of days. I would have guessed about half that. The number did not tell me what to do. It told me what was true, and every limit I had set before then had been set against the version I had guessed. The fuller account is on the awareness habits page.

That is what turning the lights on does. It does not tell you where to put the bar. It tells you where you are standing.

What the number decides

Once you have a real record, the thing you were avoiding becomes possible, which is deciding. And there are three honest answers, not one.

Sometimes the record is much worse than the estimate and the cost is one you can now name, and the answer is to stop rather than to reduce. Nothing about that is an escalation. The bar you kept setting had been doing the work of a decision you had not made, and the record is what made the decision available.

Sometimes the record is roughly what you thought and the cost is small, and the answer is to leave it alone. People expect an awareness cycle to indict them, and it often does the opposite. Some things you have been quietly managing for years turn out not to be a problem. Finding that out is a real result and it gives you the rest of your attention back.

And sometimes the record is higher than you thought and less is genuinely the target, which is the common case for caffeine, for the phone, for spending, and for plenty else besides. Less is not a compromise position in these. It is the correct outcome, and there is nothing about it that needs to escalate into stopping.

What the record shows	How it reads	What follows
Far more than you estimated, at a cost you can now name	The gap itself is the finding, and it keeps being the finding on re-reading	A break cycle with the bar at zero, for one cycle, and then a decision
About what you estimated, at a cost that is genuinely small	The record is dull, and the dullness holds when you look again	Nothing. Leave it. Spend the attention elsewhere
More than you estimated, and less is the actual point	The pattern is legible and clusters somewhere specific	A break cycle with the bar above zero, which is the rest of this post

Which of the three you are in is the decision the moving bar had been standing in for. The record settles it, and nothing else can.

Setting a bar you can actually read

Doing less is not its own kind of cycle. It is a break cycle with the bar set somewhere other than zero, and everything else about it works the way a break cycle works.

The bar still needs to be answerable at the end of the day, which is the part that most reduction tracking gets wrong. Under two hours, no more than four units, two cups maximum: each of these asks you to keep a running total in your head all day, decide in the moment whether the next one crosses the line, and then produce an honest verdict at bedtime about a count you were never really keeping. That is not a limit. It is a job, and it is a job with no clean answer at the end of it, which means the cycle ends with nothing crisp to look at.

The way out is the same shape binary habits describe for the building side of practice: one behaviour, tied to one trigger, answered yes or no. The target changes, the design does not. Instead of policing a quantity across a whole day, you tie the bar to the specific moment where the decision actually gets made, and then the day has an answer.

Domain	The count you were policing	The question you answer at the end of the day
Drinking	No more than four units in a day	Did it stay to drinks with dinner only, yes or no
Phone	Under two hours of screen time	Did the phone stay out of the bedroom, yes or no
Caffeine	No more than two cups	Did I stop at the cup after lunch, yes or no

A limit you keep score against all day is not a limit. It is a job.

The right-hand column is harder to write and much easier to live with, and writing it is most of the work of the setup. It also has to be a moment you can name in advance, because the whole point is that the moment does the deciding rather than you doing it fresh each time under conditions you have already lost. If the record from your awareness cycle showed the pattern clustering somewhere specific, which it usually does, that cluster is where the bar goes. That is a height you can see, set against ground you have now measured.

The note is where the size lives

Here is the part of this that changed my mind about the whole exercise.

Somebody kept a journal of every rule they set themselves about drinking across ten years. Roughly 30 of them, listed out, each one failed. It reads as comedy and it is meant to. But near the end of that list there is one entry the author marks as having genuinely helped, and it is not a limit at all. The rule was: if you go past one, write down what led to it. They say the journal is what let them finally see the pattern.

That is habit context, arrived at independently, on a list of 30 failed bars, and identified as the only thing on the list that worked.

It is doing something no tick can do. One drink with a friend and most of a bottle alone are the same broken square on any grid, and they are completely different facts about the cycle. A tracker can only tell you that a day went over. The note is the only place the size of it survives, which is exactly why a partial-completion slider does not solve this either: a half-filled bar tells you the day was partly bad, and never tells you what happened.

One line, on the days it broke. That’s the whole practice.

1. 01

   Step 1: The date and what actually happened

   Not a judgement, a quantity and a shape. Three, at the pub, or most of the evening on the phone, or a second cup at four. The specific version, because the specific version is the only one that can be compared to another day later.

2. 02

   Step 2: What was going on around it

   Time of day, what came just before, who was there, how the day had gone. A hard afternoon, a late finish, a friend staying on for one more. This is the column that turns eight separate bad days into one repeated cause.

3. 03

   Step 3: Whether it was a decision

   The most useful thing you can record and the easiest to skip. Chose it and enjoyed it is a different day from did not really decide, just took the glass, and the two want different responses at the end of the cycle even though the square looks identical.

When the bar moves

At the end of the cycle, the record you read is not just the count of clean days. A break cycle is read on two things a building cycle does not need: severity, which is how big each slip was, and trend, which is whether it is less than it was. Both are read off the note column rather than off any readout, and both are covered properly in the cycle check-in, which is where the reading itself belongs.

The point for setting up is simply that a cycle which went from most days to a handful of days, with the slips getting smaller, is a cycle that worked. Read against a perfect zero it looks like failure, and it is the most common way people abandon a reduction that was going fine.

Then the bar moves, and it moves in one place only. Not on a Tuesday when it is inconvenient, and not on a Saturday when it is going well. It moves at the check-in, with the cycle’s data in front of you, as one of the four decisions. Change is the decision that moves it: down if the bar turned out to be soft, up if seven clean days was never realistic and a floor would serve you better. A bar that moves during a cycle is back to being a negotiation, and the negotiation is the thing you were trying to get out of.

Which leaves the version of this most people actually need, and it is the most permissive answer in the method. You are allowed not to know yet. Not knowing is not the state before the work starts. It is a cycle type, with a shape and a length and an end, and running one is a legitimate thing to be doing for three months while you decide nothing at all. Deciding to stop tracking something afterwards is also fine, and is a decision rather than a failure. The only move that reliably goes nowhere is the one most of us default to, which is setting one more bar in the dark.

Common questions

Is it better to cut down or quit?

That is the question an awareness cycle answers, and it is not answerable in advance. Track the behaviour with no target for long enough to get past your own best behaviour, then read the record. Sometimes it shows far more than you estimated at a cost you can now name, and stopping for a cycle is the honest response. Sometimes it shows roughly what you thought at a cost that is small, and the answer is to leave it alone entirely. Sometimes it shows more than you thought and less is genuinely the right target, which is the usual case for caffeine, the phone and spending. All three are real outcomes, and setting a limit before you know which one you are in is what keeps the limit moving.

How do you track a habit you are trying to reduce rather than stop?

As a break cycle with the bar set above zero. The tick stays binary: one behaviour, tied to one trigger, answered yes or no at the end of the day. Rather than policing a running count all day, tie the bar to the specific moment where the decision gets made, so the day produces a clean answer. Did it stay to drinks with dinner only. Did the phone stay out of the bedroom. Did I stop at the cup after lunch. Then add a one-line note in habit context on any day it broke, because the tick can only say that a day went over and the note is the only place the size of it survives.

How long should you track before deciding?

Longer than feels necessary. The first weeks of an awareness cycle are the least honest, because attention alone shifts behaviour and because you know you are watching. Seven days gets you seven days of good behaviour. 30 days is the usual shape of a cycle and is often still short for this particular question. Around three months gets you far enough past the observation that you stop performing for it, and the pattern that settles out in the middle is much closer to the truth than anything the first fortnight showed.

What happens if you go over your limit?

Record it accurately, write the note, and carry on with the cycle. A day over the bar is data, not a verdict, and it is the day that carries the most information in the whole cycle. Write what happened, what was going on around it, and whether it was a deliberate choice or something you did not really decide. Do not move the bar. The bar moves at the check-in, with the whole cycle’s data in front of you, as a Change decision. Moving it mid-cycle turns the bar back into a negotiation, which is the thing the cycle exists to end.