The habit clock

Sixty-Six Days To Automatic, Sixty Days To Decide

The research figure for a once-daily habit is 66 days on average, not 21, and the money-back window is 60. The routine becomes automatic just after the window shuts, which is a useful thing to know on day one rather than on day 59.

By GLPro Team Last updated October 29, 2026 9 min read The habit clock · Reviewed by Dr. Edward La Cava, MD, FACE

Sixty-six days, and where the number comes from

There is a figure for how long a daily action takes to stop requiring a decision, and it is not the one people quote. Gardner, Lally and Wardle, writing in the British Journal of General Practice in 2012, report that a once-daily action repeated after a fixed cue reached its automaticity plateau after an average of 66 days, with wide variation between individuals. Their practical advice to clinicians is to tell people to expect around ten weeks.

Sixty-six days is an average of a range — not a threshold anyone crosses on schedule. Some people arrive far sooner and some take months. What the figure is good for is calibration: if a daily capsule still feels like a task at week six, that is the ordinary case rather than a failure.

• The cue does the work. The action is repeated after something that already happens, not at a time.
• The plateau is where effort stops, not where a benefit starts.
• It is behaviour research, with nothing to say about blood sugar or about any capsule. The routine page keeps the two apart as well.

Why twenty-one days is the wrong figure

The twenty-one day rule is the most repeated number in this whole subject and the paper above calls it a myth in as many words. It traces back to a surgeon’s observation about patients adjusting to an altered appearance, not to a study of daily habits — and it has been carried forward for decades because it is short enough to sell.

It matters here for one practical reason. Anyone who believes three weeks is the mark will read week four as a failure, and week four is exactly where a daily capsule gets abandoned. Expecting ten weeks turns the same fortnight into the ordinary middle of a process.

The fortnight where the two clocks overlap

Put the two numbers on the same calendar and the shape of the problem is obvious. The habit clock plateaus around day 66 on average. The money-back window runs 60 days from the date of purchase. The window therefore closes shortly before the routine stops taking effort, and a return has to be started inside it.

Day 66 is an average with wide variation, from behaviour research that has nothing to do with blood sugar. Day 60 is a purchase term.

That is not an argument against the guarantee, which is longer than most in this category. It is an argument for starting the count on paper, because the two clocks run at once and only one of them sends a reminder.

How reliably a daily thing is actually taken

The second uncomfortable number is about whether the capsule gets taken at all. Krass, Schieback and Dhippayom reviewed 27 studies of adherence to type 2 diabetes medication and found it ranged from 38.5 percent to 93.1 percent, with only six of the 27 studies reporting adherence of 80 percent or better.

Those are prescription medicines, taken by people with a diagnosis and a clinician watching — and the figures still spread that wide. A capsule bought online, with no prescription and no appointment attached to it, is not obviously going to do better. That is the honest reason a sixty-day trial is mostly a test of a routine: on the days it is not taken, there is nothing to test.

• ✓A daily capsule is only a daily capsule on the days it is taken.
• ✓Count the blanks rather than estimating them. Estimates flatter.
• ✓A bottle that finishes early or late tells you something a memory will not.
• ✓Order before the last bottle runs out, because a gap waiting for a delivery is a gap in the record. Shipping and tracking has the timings this desk publishes.

Week three, and what usually breaks

Week one runs on novelty and week two on the memory of it. Week three is where a daily capsule quietly stops, and the reason is rarely a considered decision. It is a week with something in it: travel, a deadline, illness, a weekend away from the shelf the bottle lives on.

The habit research is explicit that the occasional lapse did not seriously impair the process, and the timeline page says the same about the sixty days.

What this desk’s own record says about day counts

This desk holds eight reviewers in detail, and the only thing that record is genuinely good for is the day counts attached to each line. They were written on days 29, 32, 41, 45, 46, 47, 49 and 59 of those buyers’ own sixty days.

Five of the eight, all five-star, carry the same summary line: the daily habit stuck and they reordered before the last bottle ran out. Two four-star lines say it was fair at a month and that they had wanted more in week one. One three-star buyer wanted a faster change, called order support on day 47 and stopped at two bottles, which left thirteen days of the window and is the part of that story worth copying. These are individual experiences supplied by buyers, not typical results and not evidence. The reviews page prints the whole record and the arithmetic that does not reconcile in it.

Building a routine that survives an ordinary month

Everything above points at the same short list, and none of it depends on what is in the capsule.

1. One cue, chosen on day one, that happens seven days a week.
2. The bottle stored where the cue happens, not where it looks tidy.
3. A written start date, because every other date is counted from it.
4. A visible record, with blanks marked rather than smoothed over.
5. A decision date in the calendar, set on day one for day 45. The refund policy explains why the decision needs runway rather than a deadline.

Why the decision belongs on day 45

Day 59 is a bad day to decide anything. A return has to be started, bottles have to be posted, and a call has to be made to order support first. Day 45 leaves a fortnight for all of it and loses almost nothing in information, because the habit clock has not plateaued by day 60 either.

By day 45 you know the two things a sixty-day window can actually tell you: whether the routine held, and whether anything you named in advance changed. The rest belongs to a longer clock than the guarantee runs on. The professional re-assessment interval is about three months, which is well past the window in either direction.

The conversation that comes before all of it

One group settles this before day one rather than at day 45. NIDDK records low blood glucose as a side effect of insulin and of other medicines that help the pancreas release insulin, and sulfonylureas and meglitinides are named among them. If you take any of those, the person who prescribed them is the person to raise a new capsule with — and nothing about a habit or a window changes that.

NCCIH advises telling your health care providers about every supplement and medicine you take, because some interactions are serious. With no ingredient list published, this desk cannot check one for you and says so rather than implying there is nothing to check. Who this is for sets out the exclusions in full.

The short answer

A once-daily action takes about 66 days on average to stop requiring a decision, and the guarantee runs 60. Twenty-one days is a myth. Missing a day is not a reset, and daily medicine is taken less reliably than almost anyone assumes, which is why a sixty-day trial is mostly a test of whether the routine held.

So pick one cue, keep the bottle where that cue happens, write down the start date, mark the blanks, and put the decision in the calendar for day 45. The pricing page sets out which pack covers the window with capsules still in hand.