Somebody spends two hours a week producing a document that changes nothing. How internal reporting accumulates, how to test whether a report earns its cost, and what to keep.

How reports accumulate

Someone asked once. A specific question, a specific week. The report answered it and then never stopped.

A problem happened. The report was the response. ⚠️ The problem was solved, and the report outlived it — this is the most common origin by far.

Somebody left. Their report continues, produced by someone who does not know what it was for.

Nobody ever cancels one. There is no ritual for stopping, only for starting.

The test

Two questions per report:

When did somebody last change a decision because of this? ⚠️ If the answer is "I can't think of one", you have your answer.

If it stopped, who would notice within a month? If nobody, it is not a report — it is a habit.

The stronger test

Stop producing it and say nothing.

If someone asks within two weeks, it matters. If nobody asks, it never did. ⚠️ This feels risky and almost never is; the reports that matter get chased quickly.

What survives

Reports that pass are usually few, and they share a shape:

Someone specific reads them. Not "management" — a person.

They contain something that changes. ⚠️ A report of stable numbers is a report of nothing — if it looks the same every week, replace it with an alert for when it does not.

They arrive before the decision, not after. A Monday report about last week's staffing changes nothing about last week.

The cost being paid

The hours, obviously. Two hours weekly is a hundred a year, usually from someone whose time is scarce.

The attention. Reports fill the space where the missing report would go.

⚠️ The worst version: the report exists and is wrong, and decisions are made from it. Nobody checks it because it has always been there.

What to build instead

One page, weekly, with the four or five numbers somebody actually acts on. Everything else on request.

⚠️ The discipline is in what is excluded. Adding is easy and each addition dilutes the page; a one-page report that grows to three pages has become the thing it replaced.

Then the rule that keeps it honest: anything added must displace something.

Who asked for it originally

Almost every dead report was once a real question, asked once, by one person.

The question got answered
The report kept running
The person left

⚠️ Nobody kills it because nobody remembers owning it. It looks official precisely because its origin is forgotten.

The question that retires it

"Whose decision changes if this number moves?"

⇒ If you cannot name a person, the report has no owner and no purpose.

Retiring one safely

1. Stop sending it. Do not delete it.
2. Wait four weeks.
3. If nobody asks, delete it. If somebody asks, you found the owner.

⚠️ Deleting first turns a cleanup into an incident. Pausing costs nothing and produces the same information.

What "somebody asked" tells you

⇒ Not that the report was valuable — that one person had one use. Ask what they needed. It is usually one number, not the report.

Reports that survive have three properties

One reader who can act
One number that moves week to week
One place it always appears

⚠️ The third is the one people skip. A report that arrives somewhere different each time gets read once.

The weekly review that replaces most reports

15 minutes. Three numbers. One decision.

⚠️ If no decision comes out, the meeting is a report too. Kill it by the same test.

Three numbers, chosen by one rule

"If this moved 10%, how much money changes?"

⇒ Numbers with no answer to that question do not belong in the review.

What to do with the data underneath

Do not throw it away when the report goes. The data was never the problem.

Report  a rendering, for one audience, at one cadence
Data   the record, kept regardless

⚠️ Retiring the rendering is cheap; losing the record is not.

There is a weekly summary that takes somebody two hours. Nobody has acted on it in a year.

It continues because stopping it would require a decision, and nobody has been asked to make one.

Three to hold

Recurring reports produced, and by whom. ⚠️ Most operators cannot list them, which is itself the finding.

Hours spent producing them, monthly. Frequently a surprising number.

Decisions actually changed by a report, quarterly. Small, and the small number is the point.

Three pages on reporting, and what each one is for

This one covers which reports to stop producing.

The Report You Should Read Every Monday — which numbers to put on it
The Report You Open Every Monday — which figures actually change a decision

⚠️ They are different questions. What to put on a report, what actually moves a decision, and what to stop writing altogether are three separate problems, and most venues have all three at once.

Where the record has to sit

Most internal reports exist because assembling the numbers is manual, so somebody's job becomes producing them. Where the underlying record can be queried directly, the report stops being a document somebody makes and becomes a screen somebody opens.

tasteck keeps sales, shifts, and guest history on one record, so the common questions are a view rather than a weekly assembly job.

Nobody decided this report should exist forever. Somebody asked a question in 2023 and the answer has been arriving ever since.

The Report Nobody Reads