Three people, three numbers, one meeting

You know the meeting. Someone puts up shrink by region. Someone else says that is not what they have. A third person says both are wrong because neither excludes the damaged-goods write-off. Twenty-five minutes later the group has agreed on a number and made no decision about it.

The cost is not the twenty-five minutes. It is that the meeting has quietly taught everyone that the numbers are negotiable, and a number that is negotiable cannot be acted on quickly. The next time somebody sees a figure they dislike, the cheapest response is to question the source — and that is now a reasonable thing to do, because last time it worked.

Where the divergence actually comes from

Almost never from arithmetic. Nearly always from one of three places:

• Different extracts. Two people pulled the same report on different days and the system backfilled.
• Different exclusions. One workbook drops inter-company transfers. The other does not, because the person who built it did not know they existed.
• Different grain. One is per store per day, the other per receipt, and something got summed twice on the way to a monthly figure.

All three are invisible in the output. A spreadsheet showing 1.8% looks exactly as confident as a spreadsheet showing 2.1%.

Making the definition the artefact

The fix is not better discipline. It is moving the definition out of individual workbooks and into something shared that produces the figure.

In Dashful a measure is defined once, on the dataset, and every dashboard using it means the same thing by it. The exclusions are part of the approved dataset rather than a tab somebody remembered to apply. The grain is stated explicitly — one row per receipt, per store per day — because getting it wrong is how a join multiplies rows and a number silently doubles.

And a published report is a single address over live data, not a workbook that was emailed on Tuesday and edited on Wednesday. Two people opening it are, definitionally, looking at the same thing.

What this does to the meeting

The disagreement does not vanish. It moves — and it moves somewhere much cheaper.

Instead of arguing about whose number is right, somebody says "this excludes transfers, and it shouldn't." That is a concrete, answerable claim about a stated assumption. Somebody changes the dataset, every report built on it refreshes, and the question is closed for good rather than re-litigated next month by whoever happens to pull the extract.

You will still have arguments. They will just be about the business rather than about the spreadsheets.