Ask someone three months into a running goal how far they have gone this year and you will get an estimate. Somewhere around four hundred kilometers, maybe. The number feels like information. It is a guess wearing the clothes of a fact, and the gap between the two is where a lot of stalled goals live.
You already know the answer is not to try harder. You have read the difference between habit trackers and milestone trackers and understood that the shape of the goal matters. What sits underneath that difference is something more basic. A goal you cannot read the current state of cannot be steered.
What the meta-analysis measured
In 2016, Benjamin Harkin and a group of colleagues published a review in Psychological Bulletin that pulled together 138 randomised studies with 19,951 participants between them. Every one of those studies compared two groups. One group received an intervention designed to make them check their goal progress more often. The other group did not. Every study then reported what happened to goal attainment.
The pooled effect on goal attainment was d = 0.40, with a 95 percent confidence interval running from 0.32 to 0.48. The interventions also raised how often people monitored, at d = 1.98 on the same scale.
That second number is the interesting one. Because the review then tested whether it explained the first, and it did: changes in the frequency of monitoring mediated the effect of the interventions on goal attainment. The act of checking is not a symptom of motivation that was already there. It is one of the things producing it.
Why a count you hold in your head fails
An unrecorded total lives in memory, and memory runs on gist. You remember that you have been consistent, that the last month went well, that you missed a week in July. What you cannot do is compare this month against a reference value you no longer hold, because the reference value was never written anywhere. So you rebuild it each time you think about the goal, and the rebuild inherits whatever mood you happen to be in.
The awkward case is the one where the estimate feels accurate. A runner who is 340 km into a 500 km year can say “about four hundred” and be wrong by a full month of training. The error is invisible because there is no figure to check it against. Two people with the same goal, one tracking at 340 km and one estimating at “about four hundred,” will make different decisions about October, and only one of them is working from the real position.
The loop that changes behavior
The mechanism behind the effect is older than the meta-analysis. Charles Carver and Michael Scheier described self-regulation as a feedback loop in a 1982 paper in Psychological Bulletin, later expanded into a book on the self-regulation of behavior. The loop has a few parts. You perceive your current state with respect to the goal. You compare that perception against a reference value, which is the state you are trying to bring into existence. The gap between the two drives the output, which is behavior.
Carver and Scheier use a gas gauge to make the point. A driver who reads empty and believes the station is a mile away behaves differently from a driver who reads empty and knows the station is ten miles off. The fuel is identical in both cases. The perception carries the decision.
Then there is the layer above it. Carver and Scheier describe a meta-monitoring loop that tracks the rate at which the gap is closing, which is how you notice that you are solving the problem slower than you planned and how you decide to change the plan. Both layers need an input. A mental tally supplies a noisy one, and a noisy input produces noisy corrections.
What a written record adds
The review tested two conditions that sharpen the effect, and both are worth keeping in hand.
Reporting the progress to another person produced a larger effect than keeping it private. Monitoring in private without telling anyone sat at d = 0.19, reporting the information to at least one other person sat at 0.47, and monitoring in public sat at 0.55.
Recording the information physically produced a larger effect than monitoring without a record, at d = 0.43 against d = 0.29. When goal attainment was measured objectively rather than self-reported, the gap widened to 0.57 against 0.23.
Harkin and colleagues offer their own reading of why the record matters, and they are careful to label it as a set of possible explanations. A log strengthens how the information is stored and how it is retrieved later. It also gives the person something to examine over time, which is how you spot the weeks that went well and the conditions attached to them. The record turns a set of separate moments into a trend you can read.
None of this requires an audience. Why showing up every day is not the same as making progress covers the failure mode where attendance replaces distance, and the monitoring literature draws the same line from the other side. What the loop needs is a comparison against a target, which is the argument the finish line framework makes about sizing. Monitoring without a target gives you a number with nothing to sit against.
The week the record looks bad
Here is the reason most people stop logging, and the review names it too. A record survives. It sits on the page with the week you would rather forget in it. Harkin and colleagues cite work by Carlson, and by Karlsson, Loewenstein and colleagues, on how information about goal progress can reflect badly on the person reading it, and they suggest this is one reason people avoid their own data.
That is the friction the monitoring effect runs into. The same durability that makes a log useful, the fact that it holds what a memory would have softened, is what makes a bad week expensive to look at. A tracker that reads absence as evidence of failure makes that cost worse, which is the loop why finishing beats frequency untangles from the motivation side.
The two moderators above describe an average across 138 studies, and applying them means answering concrete questions. Who you report to, if anyone. What a record should hold to be worth keeping. What your tracker shows you on the month the total barely moved. Those questions are the second half of this piece.
Deciding who reads the record
The first moderator is the one people get wrong, because “report your progress” reads like “post your progress.” Harkin and colleagues measured the effect of telling another person what the record says. The studies where monitoring stayed private landed at d = 0.19. The studies where the record went to at least one other person landed at 0.47, and public monitoring at 0.55. The moderation runs on disclosure. Applause plays no part in it, and the paper describes the mechanism as being seen by someone who knows what the number should be doing.
One person clears most of the distance. The gap between a private record and a record one person reads is larger than the gap between one reader and an audience. A training partner who asks how the year is going, a spouse who knows the target, a friend collecting the same weekly total in a message: each supplies the same thing, a second party holding a reference value alongside yours.
The reader does not need to hold you to anything. Telling someone the total and moving on is the version with the strongest evidence behind it. Arrangements where the other person assigns consequences add a layer the review did not test, and the pressure those arrangements produce is a separate problem.
What counts as a physical record
The second moderator is more forgiving than it sounds. Recording the information physically beat monitoring without a record, at d = 0.43 against 0.29. The split widened when goal attainment was measured objectively rather than self-reported, to 0.57 against 0.23.
Anything that survives the week and can be read back qualifies. A paper tally counts, and so does a dot grid on the wall, a spreadsheet column, a weekly sentence in a notes app, or the running total inside a tracker. Format does not drive the effect. Durability and legibility do. A record you can consult in ten seconds gives the loop something to compare against, and Harkin and colleagues suggest that is why it works: it strengthens how the information is stored and how it is retrieved later, and it gives the person a set of readings to examine over time.
The physical part matters more than the private part. A number you rebuild from memory each time you think about the goal behaves like an estimate. A number you wrote down behaves like a measurement.
Record output, not attendance
Two people open the same app for the same six weeks and produce records that mean opposite things. One writes “ran today” twelve times. The other writes four distances that total 34 km. The first record counts the interval between entries. The second counts distance covered.
The writing-sessions deep dive draws this distinction for drafting, and it applies to every countable goal. “Wrote six days this week” holds no distance. “1,400 words this week” does. The review’s objective-attainment subgroup is the reason to care. When goal attainment was checked against a hard measure instead of a self-report, the physical-record advantage grew from 0.14 of an effect size to 0.34. Attendance records leave room for a flattering reading. Output records close it.
Where a streak counter breaks the loop
A streak counter records the spacing between entries, which is the one quantity the monitoring loop has no use for.
Carver and Scheier’s loop compares a current state against a reference value. With a streak, the reference value is yesterday, and it moves with you, so the gap between the two stays fixed at one day. Nothing in the loop has a reason to act. With a total, the reference value is the target, and the gap shrinks by a measurable amount every time you log something. A runner holding 340 km of a 500 km year can see 160 km of work remaining, decide that fits in eleven weeks at 15 km each, and get on with it. The same runner with a streak counter sees a zero after one missed Tuesday and learns nothing about the year.
Why showing up every day is not the same as making progress covers the design side of this. The monitoring literature supplies the mechanism behind it. A tracker that clears the total on a missed day removes the reference value the loop runs on.
Four configurations with arithmetic
The moderators describe averages across 138 studies. Applied to a specific goal, they become a few decisions about interval and instrument.
A 500 km running year, read weekly. Five hundred kilometers across 52 weeks works out to 9.6 km a week, a distance most training plans cover in two or three sessions. Read the total on Sunday, compare it against the quarter marker at 125, 250, or 375, and adjust the next week instead of the year. A month that lands at 31 km against a 38 km expectation is visible inside four weeks, which is early enough to change something.
A 52-book reading goal, read monthly. Fifty-two books across twelve months works out to one a week, and the honest check is the monthly count against four. A month at two books does not fail the goal. It prices the rest of the year at 4.5 books a month, and that figure is either acceptable or it is a signal to resize the target to forty. Both readings require the monthly number to exist.
A 100-session strength goal, read fortnightly. Two sessions a week finishes the count inside a year with room for a quiet month. Log the session and the total, and skip the daily question. A fortnight at zero sessions shows up as a flat stretch in the total, and the loop answers a flat stretch by raising the next fortnight to four sessions.
A savings target, read against the statement. Money goals come with a record built in, which makes them the clearest place to watch the loop work. A $6,000 year needs $500 a month. The statement performs the measurement, the deposit is the output, and the difference between the balance and the target is that same 160 km gap in a different unit.
Three objections
Recording your own progress sounds like keeping yourself under watch. The gas gauge answers this one. The driver who reads empty and believes the station is a mile away behaves differently from the driver who reads empty and knows it is ten miles off, and the fuel is identical in both cars. A record is an instrument panel for a goal. The decision it informs is the point.
Then there is the week the record looks bad. The review cites work by Carlson, and by Karlsson, Loewenstein and colleagues, on how information about goal progress can reflect badly on the person reading it, and Harkin and colleagues suggest this is one reason people stop consulting their own data. The record’s durability is the same property that makes a bad week expensive to look at. Two design choices lower the cost, and both sit in the tracker rather than in the person: count output rather than attendance, and use a total that never subtracts. All-or-nothing thinking turns a light week into a verdict. A cumulative total has no mechanism for producing one.
Then the version where monitoring becomes the work. This is the failure mode where tracking starts to feel like a second job, and it arrives when the instrument demands entries on days the goal did not move. Monitoring frequency drives the effect, and frequency runs against how often the number changes. Reading one total every Sunday sits at the low end of the workload scale. A daily question with three possible answers sits at the high end and buys nothing.
FAQ
How often should I check the record? Match the interval to the rate the number moves. A distance total moves every session, so a weekly read catches it. A book count and a savings balance move monthly. The mediation analysis says frequent monitoring produces larger effects, and the practical version of frequent is twice as often as the number changes.
Does the person I report to need to care about the goal? Knowledge is the requirement and caring helps. The moderation runs on someone holding the same reference value you hold, which is why a partner who knows the target works better than a colleague who hears a number with no context.
What counts as a physical record? Anything outside your head that survives the week. Paper, a note, a widget, a spreadsheet, the total inside a tracker app. The condition is that you can read it back later without rebuilding it from memory.
What if the goal cannot be counted? Quality goals resist a number, and sessions are the usual workaround, which is the approach the finish line framework takes for sizing. Thirty practice sessions against a target of one finished recording turns a qualitative goal into a countable one without claiming that session count equals quality.
What happens to the effect on a bad week? The effect survives it. The record holds what memory would have softened, and the reading it gives you is a position. A quiet fortnight inside a 500 km year is 15 km behind, which is arithmetic.
Do I need to report to anyone for this to work? No. Private monitoring sat at d = 0.19, which is a real effect on its own. Telling one person roughly doubles it, and that is the cheapest upgrade available if you want it.
What the loop needs
The monitoring effect describes how a feedback loop operates: a readable current state, a reference value, and a gap between them that the record keeps visible. Three settings decide how much of the effect a goal collects. What you write down, which is output rather than attendance. How often you read it, which is roughly twice as often as the number changes. Who else sees it, where one person clears most of the distance.
Every one of those is a setting. None of them is a trait.
Notch is free to download. The full app unlocks with a one-time $9.99 purchase. Every goal carries a target number and a finish line, the total moves forward, and the dot grid shows the year as it went. Download Notch on the App Store
Try Notch
Every notch counts.
A milestone tracker without streaks, guilt, or subscriptions.
Download on App StoreFree · $9.99 Pro · No subscription