The receipt, not the promise: how to tell whether a commitment is alive.

Imagine a neighborhood bakery with a standing rule: the walk-in refrigerator temperature must be checked at the start and end of every day. The rule is written in the safety binder. Everyone agrees it matters. For months, nobody questions it.

Then a carton of cream spoils early. The manager opens the binder and finds the promise exactly where it has always been. What is missing is the record. The last temperature entry is eleven days old.

That gap is common in every kind of organization. A commitment and evidence that the commitment is being kept are two different objects. Most teams build the first and forget the second. They write what should happen, assign a person, and assume the sentence will notice when reality changes. It cannot. A written rule looks equally healthy on a day it is followed and a day it is ignored.

Memory is usually the hidden monitoring system. That works until people get busy, schedules change, or ownership shifts. Then the promise remains on the page while the behavior quietly stops. “We check it twice a day” can survive weeks without anyone checking because the grammar never changes.

Andy Grove offered a distinction that helps: activity indicators and output indicators. An activity indicator tells you what someone did. An output indicator tells you what changed because of it. Activity is often easier to count, and that can make it seductive. A team can perform a ritual perfectly while the result stays flat. Grove's answer was not to ignore activity. It was to pair activity with output so the two cannot drift apart unnoticed.

At the bakery, the activity indicator is whether someone records the temperature at the required times. The output indicator is whether the refrigerator stays inside the safe range and ingredients remain usable. You need both. If the log is current but the temperature keeps rising, the team is faithfully recording a failing refrigerator. If the food seems fine but the log is empty, the team has no reliable warning when conditions begin to change.

The useful question is simple: if this commitment is alive, what artifact does the work produce?

For a temperature check, the artifact could be a dated reading on a sheet attached to the refrigerator. For a weekly cash count, it might be a signed total in the register record. For a monthly supplier review, it might be a decision note with the date and the next action. The artifact should be something you can look for in seconds.

Donald Berwick used a sharp line in health-care improvement work: some is not a number, and soon is not a time. The point is checkability. “We will check more often” cannot be found true or false. “The opening and closing leads record a reading every day” can. A number and a time turn a hope into something reality can answer.

That is not cold management. It can reduce suspicion. Auditing a vague promise feels like questioning someone's character. Looking for today's reading is just looking for today's reading. The evidence carries the conversation. If it is absent, you have a concrete problem to examine rather than a general accusation about reliability.

There are two cautions.

First, the receipt should be produced by the work itself, not added as a separate performance. If the baker reads the thermometer, writing the reading on the sheet takes a few seconds and completes the check. If the team must later open another system, fill out a long form, and send a summary email, the proof becomes a second job. It may stop even when the real work continues, which makes the evidence misleading.

Choose the smallest trace that the work naturally leaves. A receipt is useful because it is cheap and hard to misunderstand.

Second, a stale artifact is information, not a verdict. When the manager sees the eleven-day gap, the answer is not to fill the missing rows from memory or quietly restart the checks so the page looks healthy. The answer is to ask what stopped. Perhaps the opening shift changed, the thermometer moved, or the rule never named a backup. Perhaps twice a day was not the right schedule. The gap starts the diagnosis. It does not finish it.

The output measure matters here too. Suppose the bakery restores the log, but the readings show a slow rise every afternoon. The checking ritual is now alive and the refrigeration commitment is still failing. The team needs a repair, not better handwriting. A receipt proves an action happened. It does not automatically prove the action works.

Try this with three recurring commitments you rely on. They could be a safety check, a weekly update, a recurring conversation, or a piece of routine maintenance. Finish this sentence for each one: “If this is alive, then blank exists.”

Make the blank concrete. A dated reading exists. A short record of the conversation exists. A count has changed. Then go look. If you cannot finish the sentence, the commitment is still an intention. If you can finish it and the artifact is stale, do not rush to make the record green. Ask whether the work stopped, the evidence stopped, or the original promise stopped making sense.

Finally, pair the receipt with the result. “Did we record the temperature?” belongs next to “Did the refrigerator stay safe?” That pairing protects you from confusing a completed routine with a successful outcome.

That is lesson thirty four. A promise tells you what people meant to do. A receipt lets reality tell you whether the promise is still alive.