Sonanta
No hour of the day reviews your spending better than any other, and Sunday night is among the worst options available. What the evidence does support is narrower: a plan with a time holds, and a plan without one doesn't. In an experiment with 3,272 employees, asking them to write down the date they would get vaccinated raised vaccination by 1.5 points, a difference that was not statistically significant. Asking for a date and a time raised it by 4.2 points, and that one was.
A plan without a time isn't a plan
Katherine Milkman, John Beshears, James Choi, David Laibson and Brigitte Madrian ran the experiment at a large Midwestern firm. Every eligible employee got a mailing listing the times and locations of the free vaccination clinics. Some were also asked to write down the day they planned to go. Others, the day and the time.
The control group vaccinated at 33.1%. The group asked for a date reached 34.6%, a difference the authors report as not significant. The group asked for a date and a time reached 37.3%.
Everyone got the same information. The only thing that changed was how specific each group was asked to be.
It's worth sitting with how cheap this was. Nobody got money, an extra nudge, or a second letter. The whole difference between 33.1% and 37.3% was one more line on a piece of paper, asking for an hour instead of a day.
Plans beat wanting to
The family of interventions that line belongs to is called implementation intentions: deciding the when, where and how in advance. If X happens, then I do Y. Peter Gollwitzer and Paschal Sheeran pooled 94 independent tests covering more than 8,000 participants and found a medium-to-large effect on goal attainment, d = 0.65.
That number lands differently next to the one we cited when we wrote about why expense tracking gets abandoned: wanting to do something moves behaviour by roughly d = 0.36. Deciding when moves it almost twice as much. Motivation is the weak lever. The plan is the strong one.
Why doesn't Sunday night work?
Sunday night is precisely the middle group in the experiment: a day with no hour attached. "Sunday" doesn't say six or eleven, before dinner or after, or what you'd have to be doing to remember. It's the version of the plan that produced no significant effect.
It also has a problem of its own. Sunday is the furthest point in the week from nearly everything you bought. By the time you sit down, Tuesday's purchase is five days old and Monday's is six.
The week, seen from Sunday night
Each mark is a purchase. The intensity stands for how much detail survives: the amount holds, the merchant and the payment method go first.
Reviewing is reading, not capture, and it can only read what got saved. If Monday went down as "$430, no idea where", Sunday won't recover anything. It will confirm that you don't know.
The point: the problem with Sunday isn't Sunday. It's that "Sunday" is a day with no hour, and that it arrives too late for most of what you wanted to understand.
What the evidence doesn't say
This is where it's worth stopping, because the advice you'd expect by now — hook the review onto something you already do — has less backing than it usually claims. Lea Keller and her team ran a randomized controlled trial on exactly that question. Some participants planned their behaviour against a routine cue ("after lunch") and others against a clock cue ("at two o'clock"), with daily questionnaires over 84 days. There was no difference between the groups. Not in automaticity, not in plan enactment.
The same study leaves another figure that sets expectations. Of the 135 participants analysed, only 27 formed the habit measurably, and among those the median time to peak automaticity was 59 days, with cases at 4 and cases at 335.
There is some evidence favouring routines as anchors, but it's weaker: in an exploratory study on dental floss, people who flossed after brushing formed stronger habits than those who flossed before, and were still doing it eight months later. That's an observation, not a trial designed to compare cue types.
So the honest version is this. That the cue be a routine rather than a clock time is not established. That there be a concrete cue at all is.
What to do with that
What the evidence backs fits in three sentences. Pick a moment with an hour on it, not a day. Write it down, because writing it is what got measured, not thinking it. And expect around eight weeks before it stops costing effort, not one.
Then there's the part no vaccination study can tell you, which is that the review has to be short enough to survive its own appointment. A thirty-minute plan on Sunday competes with the whole of Sunday. A two-minute one competes with nothing.
And there's a step before all of it: the review can't repair what was never recorded. If the week arrives empty on Sunday, moving the hour changes nothing. That's a different problem, and it comes first.
Sunday doesn't fail for being Sunday. It fails for being a whole day.
Questions
- When is the best time to review your spending?
- The evidence doesn't point to one hour of the day over another, but it does separate a plan with a time from a plan without one. In the Milkman experiment, writing down only a date raised behaviour 1.5 points without statistical significance; writing date and time raised it 4.2 points with it. Pick a specific hour, not a day.
- Does anchoring it to something I already do help, like after dinner?
- It isn't established that it does. In an 84-day randomized controlled trial, Keller and her team compared planning against a routine cue with planning against a clock cue and found no difference between conditions. What the rest of the evidence does support is that the cue be concrete.
- How long until reviewing your spending becomes automatic?
- Longer than the enthusiasm lasts. In Keller's study, only 27 of 135 analysed participants formed the habit measurably, and among them the median time to peak automaticity was 59 days, ranging from 4 to 335. Planning for two months is more realistic than planning for a week.
- Why doesn't reviewing spending on Sunday night work?
- Two reasons. "Sunday" is a day with no hour, which is exactly the version of the plan that produced no significant effect in the vaccination experiment. And it arrives late: by Sunday night, Monday's purchase is six days old, and what survives from a six-day-old purchase is usually the amount and little else.
Sources
The vaccination rates and sample size come from Katherine L. Milkman, John Beshears, James J. Choi, David Laibson and Brigitte C. Madrian, "Using Implementation Intentions Prompts to Enhance Influenza Vaccination Rates", PNAS 108(26), 2011, pp. 10415–10420; the working paper with the experimental detail is NBER Working Paper 17183. The implementation intentions meta-analysis is Peter M. Gollwitzer and Paschal Sheeran, "Implementation Intentions and Goal Achievement: A Meta-analysis of Effects and Processes", Advances in Experimental Social Psychology 38, 2006, pp. 69–119. The routine-cue versus clock-cue comparison is Lea Keller and colleagues, "Habit formation following routine-based versus time-based cue planning: A randomized controlled trial", British Journal of Health Psychology, 2021. The flossing study is Gaby Judah, Benjamin Gardner and Robert Aunger, "Forming a flossing habit: An exploratory study of the psychological determinants of habit formation", British Journal of Health Psychology 18, 2013, pp. 338–353.