How long does it actually take to form a habit
Most people guess three weeks. Lally's 2010 study found a median near 66 days, with a range from 18 to 254, depending on which behaviour you actually choose.
Read the articleHabit AI Blog
Habit building, streaks, and on-device AI
Research-backed writing on how habits actually form, why streak systems make people quit, and what “on-device AI” means when an app claims it. Written by the team building Habit AI, a private habit tracker for iPhone.
More articles
A 30-day habit experiment you can actually finish
Thirty days is long enough to collect evidence and short enough to finish. One behaviour, one metric, a written if-then, and a rule for the first miss.
ADHD and external structure
This is not a diagnosis and not a treatment. External structure (visible cues, written if-thens, less working memory) is scaffolding for habits, not a cure.
Context-dependent memory: why the gym works and the living room does not
Place retrieves the action. The gym works because it is a cue; the living room is already full of competing cues, so the same workout rarely fires there.
75 Hard versus a sustainable streak
75 Hard sells an all-or-nothing restart culture. Habit evidence favours a miss rule, a smaller dose, and a streak that can survive a flu without a courtroom.
The abstinence violation effect
One slip often becomes a full collapse, not because the first slip was huge, but because the rule-break gets recoded as proof that the whole plan is already dead.
Friction: add it to bad habits, remove it from good ones
Seconds of friction decide the evening. Add steps to the unwanted action, remove steps from the wanted one, and stop treating motivation as the lever.
Self-monitoring and why tracking changes the behaviour
Recording a behaviour changes it. Self-monitoring is a cue and a feedback loop, which is why trackers work until they become a second unpaid evening job.
Temptation bundling
Pair a want with a should so the tempting thing becomes a cue. Bundling helps initiation; automaticity still needs a stable context and daily repetition.
Habit stacking that does not collapse
Stacking after an existing cue works until the chain is too long. Keep one link, write it as if-then, and protect that first action from a later collapse.
Tiny habits and the two-minute rule
Shrink the behaviour until it fits a bad day. Fogg's tiny start, plus Lally's weeks-not-days curve, is how consistency survives a motivational dip at home.
Keystone habits: what actually cascades
Keystone habits are a popular story of cascade. What spreads is usually a shared cue, a small win, or a schedule, not a mystical first domino in the week.
Identity-based habits, without the slogan
Calling yourself a runner does not create the run. Identity follows repeated context-cued action, not the other way around, and tiny starts still matter.
Environment design beats a motivational quote
Quotes fade by Tuesday. Context cues, friction, and implementation intentions change the room, which is what Wood and Neal treat as the real engine of habit.
Implementation intentions: if-then plans that stick
An if-then plan ties a behaviour to a cue. Gollwitzer's method beats vague goals, especially when the action stays tiny and the context remains stable.
Automaticity versus willpower
Willpower is a poor daily strategy. Automaticity, context cues, and implementation intentions close more of the intention-behaviour gap than grit talks.
The habit loop: cue, routine, reward
Cue, routine, reward is a teaching diagram Duhigg popularised. Context, automaticity, and if-then plans are what the laboratory work actually emphasises.
The 21-day habit myth
The three-week rule is folklore, not a trial. Lally's curve, plus the intention-behaviour gap, shows why 21 days still sells and then later disappoints.
Try the on-device habit coach
Habit AI is free to download, works offline, and needs no account to get started.