Attempts to change behaviour through resolve fail predictably, and the research points consistently at context as the more effective lever.
The context dependency
Habits are learned associations between a situation and a response.
Which means the cue is external as much as internal — a location, a time, a preceding action, the presence of particular people.
Studies of people relocating find established habits disrupted, with behaviour becoming deliberate again until new associations form.
That window is when behaviour change succeeds most readily, which is a practical finding.
Friction
Small changes in the effort required have effects out of proportion to their size.
Removing stored payment details, uninstalling an application, moving something out of reach, or keeping something out of the house all work by interrupting automaticity.
Which forces a decision where previously there was none, and decisions frequently go differently from habits.
Defaults
What happens if nothing is decided.
Which determines behaviour to a degree that surprises people, demonstrated repeatedly in field studies across domains.
Setting defaults deliberately — what is in the fridge, what the phone does at night, what happens automatically with income — does more than intention.
Implementation intentions
Specifying when, where and how a behaviour will occur, in an if-then form.
Which has consistent evidence across many domains for improving follow-through.
The mechanism proposed is that it creates a cue-response link directly rather than relying on remembering to decide.
Stacking
Attaching a new behaviour to an established one, so the existing habit provides the cue.
Which uses an existing reliable trigger rather than requiring a new one to form.
It works best where the new behaviour is small enough not to disrupt the existing routine.
The timeline
Research on automaticity development found a median around two months with enormous individual and behavioural variation.
Which means expecting three weeks produces abandonment at the point where the effort still feels substantial.
Missing a day did not measurably affect eventual automaticity in that research, which is reassuring and contradicts streak-based framing.
Social context
Behaviour is influenced by what people around you do, which is documented across many domains.
Which means changing environment includes changing who you spend time with, and this is uncomfortable and effective.
Joining a group doing the thing is one of the more reliable behaviour change strategies for this reason.
Breaking a habit
Harder than building one, since the old association persists and resurfaces under stress and in the original context.
Which means substitution — a different response to the same cue — works better than suppression.
Avoiding the cue entirely, where possible, is more reliable than resisting it.
Identity and self-concept
Some research suggests behaviours framed as part of who someone is are maintained better than those framed as things they do.
Which is a plausible mechanism with evidence that is thinner than the popular accounts imply.
It costs nothing to try, and it does not substitute for the environmental changes that have stronger support.
Tracking and monitoring
Self-monitoring changes behaviour, which is well established across domains.
The effect appears to come from the attention rather than from the record, and it diminishes when monitoring stops.
Which makes it useful during establishment and unnecessary once a behaviour is automatic.
Starting small
Behaviours small enough to complete on a bad day survive better than ambitious ones.
Which is because consistency during the establishment period matters more than magnitude.
Scaling up after the habit is established is easier than establishing an ambitious one from nothing.
Lapses
Research on automaticity found that missing a single day did not measurably affect eventual outcomes.
Which contradicts streak-based framing, and the damage comes from abandonment following a lapse rather than from the lapse.
Motivation and its limits
Motivation fluctuates with mood, sleep and circumstance.
Which means systems depending on it work on good days and fail on the days when the behaviour matters most.
Habits work because they do not require motivation once established, which is the entire reason for building them.
Multiple changes at once
Establishing several habits simultaneously competes for the limited capacity that effortful behaviour requires.
Which is why sequential change generally outperforms simultaneous change.
The exception is where changes reinforce each other, such as sleep and morning activity.
Relapse
Returning to old behaviour under stress, in the original context, or after a break is expected rather than exceptional.
Which means planning for it — identifying likely triggers and deciding in advance what to do — is more useful than assuming it will not happen.
Rewards and their limits
External rewards can undermine intrinsic motivation for behaviours already enjoyed, which is well documented.
Which suggests using them for behaviours nobody enjoys initially and removing them once the habit forms.
Intrinsic satisfaction from the behaviour itself is what sustains it long-term.
Which is why the environment does the work that willpower is credited with.