Yes/no, quantity, timer, or a 1-to-5 scale

Self-monitoring works because it makes a behaviour visible. It fails when the log is the wrong shape for the behaviour. A yes/no tick is a good instrument for flossing. It is a bad instrument for water, a meditation sit, or how the day felt. You end up with a calendar of green squares that cannot tell you whether the habit is growing, shrinking, or only being performed as a checkbox.

Harkin and colleagues, in a 2016 meta-analysis in Psychological Bulletin, found that monitoring progress toward a goal promotes attainment, and that the effect is stronger when the monitoring is recorded. Michie and colleagues, in a 2009 meta-regression of healthy eating and activity interventions, had already flagged self-monitoring as one of the behaviour-change techniques that actually move outcomes. Neither paper says "use a binary check for everything." The instrument has to capture the dimension you intend to change.

The log is a hypothesis about the behaviour

Carver and Scheier's control-theory account of self-regulation is simple: you compare a current state to a reference value, then act on the gap. If the reference is "did I do it at all," a yes/no is enough. If the reference is "two litres," "twenty minutes," or "a 4 rather than a 2 on sleep quality," a binary log throws the gap away. You cannot steer toward a quantity you refused to write down.

Wendy Wood's work on habits as context-cued automaticity adds a second constraint. The action that becomes automatic is a specific action in a specific setting. "Workout" as a checkbox can mean a five-minute stretch or a hard session. The context will cue whichever version you actually repeated. If you want duration to grow, duration has to be the thing you record. Otherwise the easy version wins, and the tracker still looks successful.

Phillippa Lally's 2010 study of everyday habit formation measured automaticity of a particular behaviour people had defined for themselves (eating fruit with lunch, walking after dinner). The learning curve is for that action, not for a vague category. A log type is how you keep the definition honest after week one, when motivation would rather redefine success downward.

Four instruments, four jobs

Yes/no is for behaviours that either happened or did not: floss, take the pill, put the phone in another room before bed. Partial versions exist, but they are not the point. Quantity is for volume: pages, glasses of water, repetitions, minutes only if you are actually counting a stock rather than sitting a timer. A timer is for elapsed time you want to protect: reading, practice, a walk you start and stop. A 1-to-5 scale is for qualities that have no honest unit: mood, energy, how restorative the sleep felt. Scales are not fake numbers. They are ordinal snapshots. Treat them as snapshots.

Do not attach all four to one habit. That is a dashboard, not a behaviour. Choosing one metric per habit is the companion rule. If you need quantity and a rating, you have two habits, or you have a project.

What a mismatched log does to the week

A binary meditation habit teaches you to sit for thirty seconds and tap done. The tracker cannot see the difference between that and twenty minutes, so your brain stops offering twenty minutes. A quantity habit with an inflated target (ten thousand steps, three litres, fifty pages) teaches all-or-nothing skipping: if you cannot hit the number, you log nothing, which then looks like a zero day. Lally's group found that a missed day did not, on average, wreck automaticity. A tracker that only accepts a perfect quantity still feels like a wreck, which is how people quit.

Timer logs fail in the other direction when you leave them running as theatre. The number goes up. The attention does not. If the timer is the metric, the rule is: the timer runs only while the behaviour runs. Scale logs fail when you use them as a grade. A 2 on mood is information. A week of 2s that you treat as a character assessment is how tracking turns into grading your day.

Charles Duhigg popularised the cue-routine-reward loop for a general audience. The log sits on the reward side more often than people admit: the satisfying tap. That is fine for yes/no. It is a problem when the tap is more rewarding than the session. Match the effort of logging to the behaviour. A quantity should take a second. A timer should start with the cue. A scale should be one tap, not a journal entry, unless journaling is the habit.

BJ Fogg's tiny-habits work is useful here as a starting rule, not as a forever rule. Begin with a version so small that a yes/no is true. Once the cue is stable, switch the instrument if the behaviour has a dimension you care about. Flossing can stay binary. Reading should become pages or minutes once "open the book" is no longer the hard part.

Log type	Use it when	It lies when
Yes/no	The act is binary and small	Duration or dose is the actual goal
Quantity	A count or volume is the point	The target is a stunt number you never hit
Timer	You are protecting a block of time	The clock runs while you do something else
1-to-5 scale	Quality has no honest unit	You average it into a verdict on yourself

Gollwitzer's implementation intentions still apply after you pick a type. "After I put the kettle on, I start the reading timer" is a better sentence than "read more." The type tells you what "done" means. The if-then tells you when it happens. Write both, or the tracker will fill with habits named like wishes.

Habit AI is a useful example of matching the instrument to the habit: yes/no, quantity, a timer, or a 1-to-5 scale, rather than one checkbox for every row. That is a logging choice, not a personality upgrade. The science does not require that product. It requires that the number you store is the number you mean.

If you already have a graveyard of binary habits that never changed, do not add five new ones. Pick one behaviour, change only the log type, and watch two weeks. The question is not whether the calendar looks greener. The question is whether the dimension you care about moved.

Read next

Self-monitoring and why tracking changes the behaviour is the evidence this post sits on. Paper tracker versus an app is the other fork: the instrument can be a notebook if the columns are right. For four log types in one place, see how Habit AI works.