If everything is individual, what works for everyone?
This is the obvious problem with everything else I write. If the conditions that sharpen one person flatten another, and an average taken across people cannot tell you which change would help you, then what is anyone supposed to actually do on a Tuesday?
Here is the short answer, before the evidence for it. A handful of things do hold for almost anyone, and most of them are not instructions about what to do. They are instructions about how to find out.
So here is the other half. Some things do seem to hold up across almost everyone. Fewer than the internet claims, and stranger than you would expect, because most of them are not instructions about what to do. They are instructions about how to find out.
The ones about finding out
Evidence You are a poor judge of how much you vary. Ask people how consistent they are, then measure them five times a day for a week and calculate it. The two answers correlate at between .14 and .30, which is close to no relationship at all (Anvari et al., 2025). Your memory of your own pattern is not the pattern. This is the reason to write anything down, and it is the closest thing to a universal on this page.
Hypothesis Change one thing at a time. This one is not a research finding, just logic. Change your sleep, your coffee and your schedule in the same week and something may well improve, and you will have no way of knowing which of the three did it, or whether it was the fourth thing you did not notice changing. One at a time is slower and it is the only version that answers anything.
Hypothesis Compare like with like. A Monday against a Monday. A first day off against a first day off. Most apparent patterns in someone’s own notes turn out to be the shape of their week rather than the effect of anything they did.
Hypothesis One measurement is not a pattern. You had a good day after a long walk. That is one observation, and the number of true explanations for one observation is roughly infinite. It becomes interesting on the fourth or fifth repeat, and only if you were writing the misses down too.
Hypothesis Do not score yourself out of ten. Where you sit on a scale limits how much you can appear to move on it. Someone answering near the middle has room to swing in both directions; someone answering near an end does not. So a person who rates themselves high will look steadier than a person who rates themselves in the middle, before anything real has happened. Record facts, not scores. Whether you went outside. Whether you ate. How long you slept.
Hypothesis Do not read your theory into your own questions. If you build your record around what you already suspect about yourself, a month of it will hand your suspicion back to you looking like a finding.
The ones about the work itself
Three things survived a verification pass, and each one is narrower than its popular version.
Suggestion Time away from a problem does something. A 2018 study titled “Incubation, not sleep, aids problem-solving” found that a period away from a problem improved how many people solved it, and that spending that period asleep added nothing on top (Brodt et al., 2018). If that holds, the useful instruction is not sleep on it. It is stop working on it, and sleep is simply the most reliable way most people get several hours of not working on something.
Evidence If you want more ideas out of a meeting, have people write alone first. Across 34 tests, people brainstorming together listed fewer non-duplicate ideas than the same number of people working separately and pooling their lists afterwards (Mullen, Johnson & Salas, 1991). Two corrections travel with that. It is a finding about the number of ideas, not their worth. And in the same literature, people asked to pick the best idea from their own pile performed no better than chance, so a longer list does not hand you the good one.
Evidence Work that suits you predicts how you feel about it, not what you produce. Across 172 studies, how well a job suits a person correlates with their job satisfaction at .56 and with commitment at .51. Against actual job performance it is .07 (Kristof-Brown, Zimmerman & Johnson, 2005). That is worth knowing in both directions. Finding conditions that suit you is well evidenced for making work feel better, which is not a small thing. It is not evidenced for making you better at it, and anyone selling you the second is citing the first.
The one that is a skill rather than a fact
Hypothesis Ask what a number is a number of. This is the single most useful habit I have picked up doing this, and it costs nothing. Across two research passes, headline figures kept turning out to be something other than an effect: a count of how often researchers mentioned a factor, a list of which accommodations were most commonly used, a median calculated only among the quarter of people who could answer, a spread of averages standing in for a spread of people.
The 1% rule doing the rounds is the everyday version. Improve 1% a day and compounding makes it enormous over a year. The hidden sum is 1.01 multiplied by itself 365 times, which is about 37.8. The math is real and the picture of a person is false, because it assumes daily gains multiply, never plateau, never regress, and that “1% better” is a quantity that exists. When a number impresses you, ask what was counted. It survives that question or it does not.
What this does not tell you
Nothing above says when you should get up, how long to work, or whether you do your best thinking in the morning. Those answers exist and they are yours specifically, and no list can hand them to you. What the list can do is stop you running the search badly, which is most of why people conclude that nothing works.
The honest summary is that the universal part is the method and the specific part is the answer. Anyone offering you both in the same paragraph is guessing at one of them.
Where this stops. The graded labels are doing real work here. The items marked Hypothesis are reasoning, not research findings, and I have marked them that way rather than dressing them up. The three research items each survived adversarial checking, and each is stated in the narrowest form that survived.
What I have not checked. Whether the method items generalise as broadly as I am implying. They are standard reasoning about measurement rather than results from studies of people following advice, and nobody, as far as I know, has run a trial on whether changing one thing at a time actually helps people find their own conditions faster. That is a real gap and it sits underneath this whole site.