Four studies, and what each one does not show.

What has actually been tested

1. Accountability

   In a randomised comparison of 114 people, accredited human coaches were measured against automated AI coaches on goal attainment, motivation, resilience and wellbeing. Substantial effects appeared, with mid-to-high effect sizes, only for the human coaching.

   Erik de Haan, Nicky Terblanche and Kenneth Nowack, Human Resource Development International, 2026

   What it does not show. 114 people in executive coaching, which is a different setting and a different kind of goal from three things you told a friend you would do this week. It compares accredited coaches with a coaching chatbot, so it says nothing about whether software helps at all, and KeMA is neither of the two things it tested.

2. Accountability

   Of 149 people followed for four weeks, those who only thought about their goals reached or half-reached 43 percent of them. Those who wrote them down, told a friend, and sent that friend a weekly progress report reached 76 percent.

   Gail Matthews, Dominican University of California, 2015

   What it does not show. Presented at a conference rather than peer reviewed, and 267 people were recruited against 149 who finished. It is the closest thing to a trial of this exact mechanism and it is not a large or a published one.

3. Intention

   Across 94 independent tests and roughly 8,000 participants, deciding in advance when and where you will do something produced a medium-to-large effect on whether it got done (d = .65).

   Peter Gollwitzer and Paschal Sheeran, Advances in Experimental Social Psychology, 2006

   What it does not show. This is about naming a moment, not about being watched. It supports asking when in the week a goal happens. It says nothing about whether anybody else should see the answer.

4. Accountability

   People pursuing an identity they have not yet earned will take a shortcut to feeling complete, and announcing an intention to somebody can substitute for acting on it.

   Robert Wicklund and Peter Gollwitzer, Symbolic Self-Completion, 1982

   What it does not show. This one cuts against the product and is here for that reason. Telling somebody can replace doing the thing. It is the argument for why a supporter is sent an answer on Friday and not just an announcement on Sunday.

Four studies is not a literature. Two of these support the product, one argues against a version of it, which is why the Friday answer exists and the Sunday announcement is not the end of the week, and one compares people with software in a setting that is not this one.

Books that changed how this was built

None of these is evidence for anything. Each one changed a specific decision, and the decision is what is listed rather than a review.

1. Atomic Habits

   James Clear, 2018

   That the environment does more work than the intention. Most of this product is an attempt to put the scaffolding outside your head, where it does not cost you anything to keep holding it.

2. Designing Your Life

   Bill Burnett and Dave Evans, 2016

   The idea that a life is worked on in prototypes rather than decided in one sitting. Three things a week is a small prototype. Their later Designing Your New Work Life carries the same method into work specifically.

3. Four Thousand Weeks

   Oliver Burkeman, 2021

   The cap. Burkeman's argument is that productivity culture treats the symptom as the disease, asking how to get through more rather than why so much is being let in. Three is not a limitation on this product. Three is the product.

4. The Trusted Advisor

   David Maister, Charles Green and Robert Galford, 2000

   Their trust equation divides credibility, reliability and closeness by self-orientation, and the denominator is the one nobody optimises. It is why this site cites the study that undercuts it and prints the sample size that weakens it.

5. The Wheel of Life

   Attributed to Paul J. Meyer, 1960s

   The areas themselves, and the ten point scale. An earlier version of this product scored attention instead of satisfaction, on the argument that an area can be going fine and still be starved. That was replaced: needing to explain your own scale is the evidence the scale is wrong, and satisfaction is the question people can answer without being taught it.