One place to look
The plan, the quarterly objectives and the weekly operating pack in one private page, with the version list always to hand. The same columns the team already used, so nothing had to be relearned.
Case study · High-growth technology company · Company operating system
A company running a well-known operating framework properly, across five tools that do not talk to each other, with people carrying the numbers between them by hand. The rigour was never the problem. The clerical work was.
This is the normal shape of a company that has grown fast, and it is worth saying that plainly because it is usually described as a failure of discipline when it is the opposite. The discipline was in every one of those tools. They were simply never asked to work together.
The plan lived in one spreadsheet, one tab per version, twenty of them going back to 2020, which is also the company's written history. Quarterly objectives lived in a second spreadsheet. Weekly check-ins lived in a third-party tool. The weekly leadership meeting ran on a board. Revenue actuals lived in the finance team's own sheet. Each one correct, each one separate.
So the same month's revenue figure appeared in several places, and two of them used different definitions, which meant somebody reconciled them by hand every month. Comparing one plan with the next meant opening tabs side by side. Check-in percentages were copied across once a month, so the objectives sheet reflected the last copy rather than the current position. The meeting's own score was posted as a comment on a card, week after week, so the trend was invisible unless somebody read the comments in order.
Most of that work sat with one person, whose stated goal for the engagement was to run the operating system without outside help.
Because it is the same argument as everything else on that page, arriving one layer earlier. An office cannot report the return on anything until the numbers underneath it mean one thing. Here there was no office and no AI program yet: there was a company that could not compare this month with last month without a person doing arithmetic, which is the state every governance structure has to fix before it can do anything useful.
It also answers the question this page gets asked most, which is whether the structure has to be heavy. This one is not. It is a clerk.
The system does not replace the operating framework, the check-in tool, the board or the weekly meeting. It replaces the copying, the hunting through tabs and the reconciling by hand. It reads the sources, keeps one live record with every version, and writes nothing anywhere without a person approving the change first.
The plan, the quarterly objectives and the weekly operating pack in one private page, with the version list always to hand. The same columns the team already used, so nothing had to be relearned.
Exactly one plan is live, drafts are drafts, and a superseded version is never edited again. Twenty versions readable side by side, including what changed between each one and the next, and a cell-for-cell view of the original tab so nothing from the old layout is lost.
A source map saying which sheet, tab and row each number comes from, and a data dictionary saying what each term means and which line is the right one. The reconciliation stopped being a monthly task because the disagreement was resolved rather than repeated.
Every write is shown as a proposed change first. Nothing is overwritten, and when there is nothing to report it says nothing rather than generating a report to prove it ran.
One deliberate restriction: you cannot type a number into the page. Figures arrive through the monthly close and the weekly tasks, after approval, which is the only way the page and its history stay worth trusting.
If your leadership meeting is spent reconciling rather than deciding, the problem is almost never the meeting and almost never the people. It is that the operating picture is assembled by hand every week, and whoever assembles it is doing clerical work at a senior salary. That is a small, unglamorous, extremely high-return thing to fix, and it is usually the thing that has to happen before anything larger can be measured at all.
The fix is usually smaller than anyone expects, and it is what everything else you want to measure depends on.