A number without a date is a number nobody can trust

Somewhere in your company there is a slide with a number on it that everyone repeats. Customer retention is 92 percent. The close takes six days. Pipeline covers the quarter three times over. The number is quoted in meetings, it appears in the board deck, and it may be in someone's bonus plan.

Ask when it was last true, and the room goes quiet.

A number is a claim about a moment

Every business figure describes a point in time. Retention was 92 percent for the quarter ending in June. The close took six days in August. Without the date, the number is not wrong exactly, but nobody can tell how much to rely on it. A six-day close measured last month is a fact. The same figure measured a year ago is a memory.

Most spreadsheets and planning tools store the number and lose the moment. The cell says 92. When someone updates it, the cell says 94, and the 92 is gone. There is no way to see that the figure sat unchanged for five months before the update, or who changed it, or what it was before.

What overwriting costs

Losing the history matters more than it seems.

You cannot see the trend. A measure at 94 that was 97 three months ago is a different conversation from one that was 88. The single number hides the direction.

You cannot see the silence. If a figure has not been updated since spring, the most important fact about it is that nobody has looked. An overwritten cell shows no trace of that.

You cannot audit it. When a measure feeds an incentive plan, or a board is told it moved, the history is evidence. A number that can be quietly replaced is a number nobody outside the team has reason to trust.

Three rules for numbers you can trust

Record the date the number was true, not just the date it was entered. A reading for June entered in July is still a June reading.

Never overwrite a reading. A correction is a new reading, with its own date and author and a note saying what changed. The old one stays.

Measure progress from where you started. A target of 55 from a baseline of 42 is 13 points of work. At 48 you are under half way there, not 87 percent of the way, which is what dividing the current figure by the target suggests. For measures you want to go down, such as churn or cycle time, dividing the current figure by the target does not work at all.

How Trellis handles it

In Trellis, every metric reading is a new record with its own as-of date and the person who recorded it. Past readings are never edited or deleted; the history is the audit trail, and every change to the plan is written to an audit log as well. Progress is measured as the distance travelled from the baseline toward the target, in whichever direction the measure should move.

Because each reading carries its date, Trellis can tell on its own when a measure has gone quiet, and marks it stale. A leadership meeting can then see, for every number on the page, both what it says and how recently it was true. That second fact is usually the one that changes the conversation.