A budget over time with a reserve

A ceiling says how much you may spend in total; it doesn't tell finance when the money leaves. A budget over time does: give each line of work its full budget and an owner, then spread that budget over the months when the work happens.

Klasna's physics launch has a 600,000 UAH ceiling. The six work lines add up to 510,000, planned as 45,000 in October, 160,000 in November, 190,000 in December and 115,000 in January. December is the heaviest month: the expert review, the practice sets and the first ads land there. By the end of December, 45,000 + 160,000 + 190,000 = 395,000 should have left the account.

The reserves aren't in the cash plan, because they aren't planned spending: they are money kept back for things that may happen. Hanna can now line up the project's cash with the months when families pay for spring packages.

Of the 90,000 kept back at Klasna, 50,000 is a contingency reserve: money for risks the team can already name. A second round of demo lessons if fewer than eight tutors pass, extra days from the outside reviewer, diagrams redone, more ads if the waiting list is weak.

Two things define it. It is sized from the risks: what each would cost and how likely it looks, not a round percentage. And it is spent by the lead when one of those risks happens, without a new decision, so the project can react in days rather than wait for a meeting.

That speed is why the list matters. If six candidates pass instead of eight, the second round of demo lessons is a listed risk: the lead pays for it from the contingency reserve and notes it. Planned work, by contrast, is paid from its own line, never from a reserve.

Besides the contingency reserve, Klasna keeps 40,000 as a management reserve: money for surprises nobody can name today. It sits with the sponsor, outside the lead's plan, and Leo releases it only on a written request that says what happened and what was tried first. So the lead controls 560,000 (work plus contingency); the last 40,000 is Leo's call.

The split matches who can judge. A named risk was already weighed when the plan was made, so the lead can act on it fast. A true surprise, like the video vendor changing its terms, needs the sponsor's view of the whole business.

Neither reserve pays for new scope. If someone asks to add olympiad preparation, which is on the out list, that is a separate decision with its own money and time, or the reserves vanish into extra work and are gone when a real risk arrives.

Comparing money planned with money spent says nothing about how much work the money bought. A lightweight progress check uses three numbers.

• Planned value: the budget of the work scheduled by now (at Klasna on 30 November, 205,000).
• Actual cost: what was spent (201,000).
• Earned value: for each line, its share done times its full budget, added up.

The curriculum has a full budget of 120,000 and is 70 % done: 0.7 × 120,000 = 84,000 earned. Danylo's diagnostic test has a full budget of 85,000, is 40 % done and has cost 52,000: it has earned 0.4 × 85,000 = 34,000 for 52,000 spent.

Always multiply by the full budget of the task. Multiplying by the money spent (0.4 × 52,000 = 20,800) or by the plan to date (0.4 × 45,000 = 18,000) gives a number that no longer means "work done at its planned price", and every comparison built on it is wrong.

Once you have the three numbers, two ratios make them easy to read.

• Earned ÷ planned shows schedule: how much of the work planned by now is actually done. Below 1 means behind.
• Earned ÷ actual shows cost: how much planned work each hryvnia bought. Below 1 means the work costs more than planned.

At Klasna on 30 November: 171,500 ÷ 205,000 ≈ 0.84, so about 84 % of the planned work is done; and 171,500 ÷ 201,000 ≈ 0.85, so each hryvnia bought about 85 kopiykas of planned work.

Read the two separately. A project can be behind but on cost (slow, cheap), or on schedule but over cost (fast, expensive). Klasna is both behind and over cost, which is why the forecast in the next step matters. Keep the ratio to two decimals; more precision than that pretends the share-done figures are more exact than they are.

At Klasna's 30 November check, spending looked fine: 201,000 against a plan of 205,000. Earned value tells a different story. At 171,500 it is below planned value, so the project is behind schedule (171,500 ÷ 205,000 ≈ 0.84), and below actual cost, so the work costs more than planned (171,500 ÷ 201,000 ≈ 0.85).

A simple forecast follows: if the rest of the work costs the same, 510,000 ÷ 0.853 ≈ 597,700. That is above the 560,000 the lead controls (work plus contingency), and it would eat nearly all of Leo's management reserve, though it is still under the 600,000 ceiling. Nothing has been breached; it is an early warning.

So the next move is a question, not a panic: where is the gap? Here most of it is the diagnostic test, where scoring took twice the hours expected. Re-estimate what is left there before asking for any reserve.

The most common slip in an earned value check is the multiplier. Earned value must be the share done times the task's full budget, because it measures work done at its planned price. Only then can you compare it with what the work actually cost.

Multiply by the money spent instead and the measure turns circular. Danylo's test, 40 % done with 52,000 spent, would show 0.4 × 52,000 = 20,800: the more he overspends, the more "progress" the number shows, and the comparison with actual cost means nothing. With the full budget of 85,000 it shows 34,000 earned for 52,000 spent, a cost ratio of about 0.65, which is the truth Hanna needs.

The same goes for the plan to date: 40 % of what was scheduled by now (45,000) isn't earned value either.

Earned value is only as honest as the share-done figures behind it. Each line's owner reports it, but against something checkable: chapters accepted by the reviewer, question sets signed off, candidates who finished a demo lesson. Not hours worked, not money spent, and not "almost done", which isn't a percentage at all.

When Vira says the curriculum is 70 % done, the useful follow-up is "which units are accepted?". If seven of ten units have passed review, 70 % is a fact; if all ten are drafted but none reviewed, it is a hope.

The check itself stays light. Spending alone can look perfect while the work falls behind, as it did on 30 November, so once a month send the three numbers and one sentence on the cause. No more ceremony than that.

How it fits together: a budget by month, two reserves with two owners and one honest progress check let you say where the money stands. The case next puts it all together: how long earlier launches really took and the date you can honestly give Leo.