Three worked examples.

A mid-sized manufacturer closed every month on time and still lost control of its cost base. The monthly pack showed the month — never the direction of travel — so a facilities and energy line that crept up by a few percent each period never triggered anything. Each month on its own looked like a rounding error.

Two weeks mapping where each number actually comes from, then a single model joining the ledger to a proper date table so a figure can be compared with the same month last year, with budget, and with its own trend. The measures are defined once and agreed: overspend counts only the months you went over, because a variance that nets a good month against a bad one hides exactly this.

The pack the board already expects, assembling itself from source rather than by hand. A decomposition view that lets a finance director ask why without asking anyone. And drift detection: a department over budget three months running is surfaced, not averaged away.

Documented, on the client's own Microsoft tenant, with a walkthrough for the finance team and a written definition of every measure. Nothing lives in one person's head, and nothing depends on us staying.

€266,840 of unbudgeted facilities cost identified across 15 months — a €213,000 annual run rate, and a fifth of the company's entire overspend, sitting in a department representing four per cent of the budget.

Beyond the finding: roughly three days a month returned to the finance team, and a close that no longer depends on one person remembering the order of four exports.

Three production lines, one visibly behind, and no agreement about why. Downtime was recorded on paper by shift and typed up weekly, so by the time anyone saw a total it was three weeks old and aggregated past the point of usefulness. The plant knew Line 2 was the problem. Nobody could say what it cost or what to fix first.

Bring production, downtime reasons and delivery promises into one model at the grain they are actually captured — line, shift, day — instead of the grain they were being reported at. Attainment becomes produced against planned rather than a number someone calculates in a spreadsheet.

A daily attainment view by line and shift, a Pareto of downtime reasons that answers what is costing the most hours, and an on-time delivery trend tied back to the shifts that caused the misses.

Running on the client's tenant against their own systems, with the maintenance planner and the shift leads trained on it. Refresh is automatic; the morning meeting starts from the same screen.

Line 2 attainment at 88.5% against 93.6% on the other two lines — 214 excess hours a year, of which 45% is a single cause: unplanned maintenance, materially worse on the late shift.

At the plant's own output rate and contribution margin, that is roughly €109,000 a year of production lost to a maintenance schedule that did not exist — and a specific shift to fix first.

A professional services firm knew its revenue and knew its headcount, and could not tell which work was actually profitable. Fixed-price engagements were quoted from experience, and the hours they really consumed were only visible once the project closed and it was too late to matter.

Join the project ledger to the timesheets — the two systems nobody had put side by side — so hours worked can be read against hours quoted while a project is still running. Utilisation is measured against grade-level targets rather than a single firm-wide number that flatters everyone.

A live view of every open project against its budgeted hours, margin by engagement type, and utilisation by team and grade. Overruns surface in week three, not at close.

Owned by the practice leads, documented, and reviewed monthly for the first quarter so it becomes part of how the firm runs rather than a report someone opens once.

Fixed-price work running at 128% of budgeted hours and earning a 14.7% margin against 33.1% for time and materials — €163,400 of effort given away unbilled across 20 projects.

Per project that is €6,434 against €15,920. The firm was pricing its best work as its worst, and the fix was a pricing decision, not a productivity one.