Finding one
Identity moves behaviour more than intention does.
In a set of US voting experiments, people asked questions framed as a noun — how
important is it to be a voter — turned out at meaningfully higher rates than people asked the same
questions framed as a verb, about the act of voting. The effect was found across separate elections. A related
set of studies found that asking people not to be "a cheater" reduced cheating more than asking them not to
"cheat".
What we do with it: promises are written as things a person does, and the language throughout
the product is about who you are becoming rather than what you are hitting. Nobody here has a target weight.
Bryan, Walton, Rogers & Dweck (2011), PNAS 108(31), 12653–12656.
Bryan, Adams & Monin (2013), Journal of Experimental Psychology: General 142(4), 1001–1005.
The limit: these are single-action studies about voting
and cheating, not about maintaining a habit for ninety days. The mechanism is well supported; the size of the
effect over months is not established.
Finding two
Monitoring your progress reliably improves it, and recording it works better still.
A meta-analysis pooling 138 studies and 19,951 participants found that prompting people
to monitor their goal progress produced a reliable improvement in goal attainment. The effect was larger when
progress was physically recorded, and larger again when it was reported to somebody else.
What we do with it: daily sealing is the recording, and it goes to a named person rather than
into a private void. Those are the two conditions the analysis found strengthened the effect.
Harkin et al. (2016), Psychological Bulletin 142(2), 198–229.
The limit: average effects across a very wide range of
goals. Monitoring helps; it is not a guarantee for any one person.
Finding three
Commitment devices work, and they keep working after you stop paying attention.
Bank clients in the Philippines who were offered a savings account they could not
withdraw from until a goal was met increased their savings balances by roughly 81% after twelve months relative
to a control group. In a separate trial, smokers offered a deposit contract — put money aside, take a nicotine
test in six months, forfeit it to charity if you fail — were about three percentage points more likely to pass,
and the effect was still there in a surprise test at twelve months.
What we do with it: the signature, the locked seal, and the fact that a named person has your
promises are all soft commitment devices. We deliberately do not take your money if you fail, because punishment
changes who stays rather than who succeeds.
Ashraf, Karlan & Yin (2006), Quarterly Journal of Economics 121(2), 635–672.
Giné, Karlan & Zinman (2010), American Economic Journal: Applied Economics 2(4), 213–235.
The limit: both involve money at risk, which we do not
use. And in the savings study the effect had faded to a statistically insignificant level by two and a half
years, which is a real and important caveat about how long any of this lasts on its own.
Finding four
Coaching produces measurable improvements in goal attainment.
A meta-analysis of 18 studies found workplace coaching improved goal attainment with a
large effect size, along with smaller improvements in performance, wellbeing and coping. A second meta-analysis
across 17 studies and 2,267 participants found a positive overall effect, with the strongest effects on results
rather than on attitudes.
What we do with it: a Keeper is the smallest useful version of this — not a full coaching
relationship, but a trained human with your goals in front of them and a duty to respond.
Theeboom, Beersma & van Vianen (2014), The Journal of Positive Psychology 9(1),
1–18. Jones, Woods & Guillaume (2016), Journal of Occupational and Organizational Psychology
89(2), 249–277.
The limit: these are studies of professional coaching
in organisations, mostly with more contact time than a weekly message. They support the principle, not our
specific format.
Finding five
Roughly 66 days to automatic, and one missed day does not undo it.
In a study following people adopting a new daily behaviour, the median time to reach
automaticity was 66 days, with an enormous range from 18 to 254 days depending on the person and the behaviour.
Missing a single day did not measurably harm the process. Sustained inconsistency did.
What we do with it: rule two, in its entirety. This is why a miss is not punished and why the
streak you can lose is the honesty one, not the perfection one.
Lally, van Jaarsveld, Potts & Wardle (2010), European Journal of Social Psychology
40(6), 998–1009.
The limit: a small study, 96 participants, self-selected,
and the "21 days" and "66 days" numbers are both wildly over-quoted. Treat 66 as a rough middle of a very wide
range, not a deadline.
Finding six
Beating yourself up after a lapse makes the next lapse more likely.
Students who forgave themselves for procrastinating before a first exam went on to
procrastinate less before the next one. The mechanism is straightforward: self-criticism after a slip adds a
layer of bad feeling to the task itself, which you then avoid harder. Related work on the "what the hell"
effect describes the same collapse — one broken rule, so the day is written off entirely.
What we do with it: owned days are neutral, not red. There is no shame colour anywhere in the
product. A miss is grey, it costs nothing, and your Keeper's response to three of them is to make the promise
smaller.
Wohl, Pychyl & Bennett (2010), Personality and Individual Differences 48(7),
803–808. See also Marlatt & Gordon's relapse-prevention work on the abstinence violation effect.
The limit: the forgiveness study is correlational and
specific to procrastination in students.
Finding seven
You work out who you are by watching what you do.
Self-perception theory holds that when internal signals are weak or ambiguous, people
infer their own attitudes from their own behaviour, much as an outside observer would. You do not decide you are
disciplined and then act on it. You watch yourself keep your word thirty times and conclude something about
yourself from the evidence.
What we do with it: this is the argument for the Receipt existing at all. The point of a
permanent, unfakeable record is that it is evidence you can be shown, about yourself, by something other than
your own memory — which is a famously unreliable witness on this subject.
Bem (1972), self-perception theory, in Advances in Experimental Social Psychology, vol. 6.
The limit: a foundational theory rather than a single
trial, and like most of mid-century social psychology, parts of the surrounding literature have not replicated
well. We rely on the general principle, not on any specific figure.
The one everybody quotes
About that "76% versus 43%" accountability statistic.
You will have seen it: people who wrote their goals down and sent weekly progress
reports to a friend achieved far more of them than people who only thought about their goals. It comes from a
study by Gail Matthews at Dominican University of California, and the direction of the finding is consistent
with the meta-analytic evidence in finding two.
We are not putting it in our marketing, because as far as we can establish it was presented at a conference
rather than published in a peer-reviewed journal, and it gets quoted with a confidence the original design does
not support. It is the single most-cited number in this entire industry and almost nobody who cites it has
checked where it came from.
We are telling you this because if we will flag a statistic that helps
our argument, you can reasonably trust the ones we did keep.