Claim
Reliability is an accounting fact
Being a man of your word is not a character trait. It is a ratio: things you said you would do, over things you did. Everybody has one. Almost nobody knows theirs, which is why the phrase gets used so freely.
Two men can both describe themselves as reliable while running ratios of nine in ten and four in ten, and both will be sincere, because memory keeps the promises you kept and quietly drops the rest. The ratio carries no verdict on you. It is a number, and numbers move. The first move is always the same: stop guessing what it is.
Evidence
Why intentions fail and instructions do not
Peter Gollwitzer's work separates two things people both call a plan. A goal intention is I will call the bank. An implementation intention names the trigger: when I sit down at my desk on Tuesday, I call the bank before I open email.
The 2006 meta-analysis by Gollwitzer and Sheeran pooled 94 studies and found a medium-to-large advantage for the second form, d = 0.65. Same person, same motivation. The only change is that the promise now carries a time, a place and a first physical move.
The mechanism is unglamorous. A goal intention leaves the hardest part unspecified, so you have to decide again later, and later is when you are tired. An implementation intention spends the decision now, while it is cheap, and leaves a cue that fires without you.
A promise with no time in it is therefore not a promise. It is a preference.
Action
Build the ledger
Open a page. Two columns: what you said, what you did. Go back seven days and no further, because seven days you can still remember.
Write every commitment you made out loud, including the small ones. I will look at it tonight. I will come to the thing. I will get back to you. Include the ones you made to yourself and mark those with a dot, so you can count them separately.
Then fill the second column in one word each: done, late, dropped, renegotiated. Count. That number is your current ratio, measured rather than felt, and it is the baseline for everything after this.
Caution
Overpromising reads as generosity
The most common failure here has nothing to do with lying. You say yes while the other person is still talking, because yes is warm and no is cold and the cost arrives later.
A man who commits to everything is not generous. He is unpredictable, which is worse for the people around him than a man who says no twice a week, because they cannot plan and they will not tell you they have stopped relying on you. They simply stop.
The too-literal failure runs the other way. A man reads this, finds his ratio is four in ten, and fixes it by promising nothing. His ledger goes clean. He is now perfectly reliable and load-bearing for no one, which is tomorrow's problem and a real one. The ratio is a diagnostic. Driving it to one by shrinking the denominator is cheating, and you will know you are cheating.
Aside
Renegotiation is not breakage
A ledger like this treats every broken promise as the same size. They are not. You promised your daughter you would be at the recital and you promised a colleague a document by six. Those go in the same column and they are nothing like each other.
Correct, and the fix is one rule rather than a weighting column: a promise you cannot keep gets renegotiated before the deadline, with the person, out loud. I said Thursday. It will be Monday. Does Monday break anything for you?
Four seconds, and it is the entire difference between a man who is late and a man who is unreliable. Late is a scheduling problem. Unreliable is what happens when somebody finds out at the deadline, because up to that point they were carrying your risk without being told they had it.
Some promises should be broken. Your father goes into the hospital and the wedding goes. The ledger measures reliability, not wisdom, and it will occasionally reward the wrong choice.
Example
Ravi and his two ratios
Ravi is forty-one and runs a small contracting business. He would have described himself, before January, as somebody people can count on.
He ran the seven-day count and came out with fourteen commitments. Nine were to himself, five were to other people. Of the five to other people he had completed four. Of the nine to himself he had completed one.
What he said afterwards was the useful part: So I am reliable to everybody except me, and I have been telling myself the opposite for years.
He has not fixed it. Six months on, the ratio to other people is still around four in five and the ratio to himself is still under half. What changed is that he now knows the two numbers are different, which is more than he had. Promises to yourself have no witness, so dropping one costs nothing on the day, and a man can run that deficit for a decade without anybody raising it.