Does habit tracking work? What 138 studies say, and when to stop

Does habit tracking work? The short answer from the research

The most complete answer comes from Harkin et al., 2016, a meta-analysis published in Psychological Bulletin. The authors pooled 138 randomized studies with 19,951 participants. Each one compared people who were prompted to monitor their progress toward a goal with people who weren’t. Most were about health: weight, diet, physical activity, blood pressure, blood glucose, asthma, depression.

Two numbers matter. The interventions were very good at getting people to check their progress more often (an effect size of d = 1.98, which the authors point out is more than twice a conventional “large” effect). And that extra monitoring helped people reach their goals, with an average effect of d = 0.40, which they describe as small to medium. The detail that makes this more than a correlation is that the change in how often people monitored mediated the effect on goal attainment: the interventions helped to the extent that they got people checking.

Other reviews point the same way. Michie et al., 2009 sorted 122 evaluations of healthy eating and physical activity programs (44,747 people) into 26 behavior change techniques. Self-monitoring explained more of the difference between programs than any other technique. Programs that combined it with at least one other technique from control theory, such as setting a goal or giving feedback, averaged an effect of 0.42, against 0.26 for the rest. For weight loss, Burke et al., 2011 reviewed 22 studies and found a consistent link between self-monitoring and losing weight, while rating the evidence as weak because of how the studies were designed.

So yes, with realistic expectations. A tracker won’t carry a habit by itself, but people who keep an honest record of what they do tend to do more of it.

Why writing it down changes what you do

The mechanism comes from control theory, which Carver and Scheier laid out in 1982 around a discrepancy-reducing feedback loop. You hold a standard (meditate ten minutes a day), you notice where you actually are, and if there’s a gap, you act to close it. Harkin and colleagues put it bluntly: setting a goal only fixes the reference value. The real work of pursuing it is checking your progress against that value and responding to what you see.

The loop tends to break at the comparison. Harkin’s team call this the “ostrich problem”: people avoid information about how they’re doing, especially when it might be bad news. Few of us check a bank balance or count what we eat unless something prompts us. At its simplest, a habit tracker is that prompt.

Recording seems to matter on its own, too. The authors suggest a few reasons: a written record is easier to remember, harder to ignore or explain away, and putting something down in writing may raise your commitment to the goal. Those are their hypotheses. The meta-analysis shows the difference in results, not which explanation is right.

The benefits of habit tracking depend on how you do it

The most practical part of the meta-analysis is the moderators: the features that made monitoring work better or worse.

• Write it down. Studies where people physically recorded what they observed averaged 0.43, against 0.29 when they only noticed it. When goal attainment was measured objectively, the gap was wider: 0.57 versus 0.23.
• Tell someone. When the information was reported to at least one other person, the effect was 0.47. Private monitoring that nobody else saw came in at 0.19. Fully public monitoring, like weighing in at a group class, reached 0.55, but only three studies tested it, so hold that number loosely.
• Track the thing you want to change. Monitoring a behavior had a large effect on that behavior (0.79) and no reliable effect on outcomes; monitoring outcomes moved outcomes (0.62) but not behavior. If your goal is to run three times a week, log runs. If it’s a lower resting heart rate, the number itself is the target.
• Pair it with a goal and quick feedback. Programs that added goal setting, feedback or action planning to monitoring had larger effects, and immediate feedback on behavior beat the delayed kinds.

When tracking backfires

Measurement has a cost most people don’t expect. In six experiments, Jordan Etkin had people color shapes, walk or read, some with a counter showing their output and some without. Measurement made them do more and enjoy it less. In a field study, students who chose to wear a pedometer for the day walked more than students whose pedometer was sealed shut, and rated walking as less enjoyable. In another experiment, people who had seen a page counter were less likely to choose to keep reading afterwards (27% versus 49%).

Etkin’s explanation is that counting output makes something you do for its own sake feel like work, which wears down intrinsic motivation. Two details make this useful. People didn’t see it coming: 88% of participants offered a step count wanted it, and they predicted it would make walking more enjoyable. And a clear goal changed the picture. In a follow-up with 300 readers, measurement reduced enjoyment for people reading without a goal but slightly increased it (a marginal result) for people told to read as many pages as possible.

The other cost is effort, and it’s why so many people quit. Surveying people who had stopped using tracking tools, Epstein et al., 2016 found that the most common reason among former activity trackers was the work of collecting the data, given by 46% of them. Some felt guilty afterwards, more often those who had started tracking to change a habit. Attig and Franke, 2020, who surveyed 159 former wearable users, note that many people abandon trackers after a few months, and that quitting for good was linked mainly to lost motivation.

What to track (and what to leave alone)

• Track behaviors you control, worded so a day is clearly done or not done: “walk 20 minutes”, not “be more active”. That’s the matching effect from Harkin’s data: logging the behavior moves the behavior.
• Keep the list short. Every habit you log is a small daily task, and the effort of collecting was the top reason people dropped their tools. We’d start with three to five habits you actually care about rather than fifteen you skim.
• Use the simplest measure that answers your question. A yes or no is enough for flossing. Count or measure only when the amount is the point: pages, glasses of water, kilometers.
• Think twice before putting numbers on something you love. If you read, walk or play music for the pleasure of it, a counter can drain some of that pleasure. If you do want more of it, a looser measure (“did I read today?”) is less likely to turn it into a quota than a page count.
• Log it right away. Immediate feedback beat delayed feedback in Harkin’s analysis, and a log filled in from memory at the end of the week is mostly guesswork.

How to read your habit data without turning it into a chore

Logging is the cheap part. The benefit comes from the comparison step, and that’s where people either skip it or overdo it.

• Look weekly, not hourly. A short review once a week (what got done, what slipped, and why) is enough to close the loop. Refreshing your stats several times a day is closer to the step-counter trap.
• Read rates, not single days. Five out of seven is a good week. One empty square says very little, and in Lally et al., 2010 missing a single day didn’t measurably slow habit formation (more on missed days). A run of empty Mondays, on the other hand, tells you something you can fix.
• Compare with your own target. In Harkin’s data, comparing with where you were before worked about as well as comparing with where you want to be (0.43 versus 0.41), so your own history is a fair yardstick.
• Change the plan, not your opinion of yourself. When a habit keeps failing, the fix is usually to make it smaller, move it to a different time of day, or tie it to a clearer cue with an if-then plan.
• Consider telling one person. Reported monitoring did far better than private monitoring. That can be a weekly message to a friend; it doesn’t need to be a public feed.

When to stop tracking a habit

Tracking is scaffolding. In Lally’s study, automaticity rose along a curve that flattened out, and reaching the plateau took anywhere from 18 to 254 days (how long habit formation takes). Once a habit runs on its own, logging it every day adds work and very little information.

Signs it may be time: you do the habit before you remember to log it; the data hasn’t told you anything new in weeks; logging feels like the task instead of the habit; or it’s something you do for fun and the numbers have started to spoil it.

Stopping isn’t failing. In Epstein’s survey, some people quit because they had “learned enough”, and some kept applying what they had learned after they stopped. The authors conclude that abandoning a tracker “is thus not always indicative of failure”. A sensible approach is to stop logging, keep the habit, and look in now and then. If it starts to slip, track it again for a few weeks.

What the research doesn’t tell you

• Almost all of it is about health goals with a clear target. Harkin’s team flag the lack of data on behaviors unrelated to health.
• The effects shrink under scrutiny. Adjusted for possible publication bias, Harkin’s estimate for goal attainment fell to 0.19 (the authors still read the true effect as small to medium). Effects were also smaller in studies where participants and researchers were blind to conditions.
• Control groups were usually assumed to do no monitoring at all, which can widen the contrast, and the moderators overlap: studies that asked people to record may also have differed in other ways.
• Burke’s weight-loss review relied mostly on samples of white women and on self-report.
• Etkin’s experiments were short, from minutes to a single day, and she notes that how long the effect lasts is an open question.
• None of these studies tested a habit app as such, or measured whether tracking makes a behavior automatic. They measured whether people reached their goals.