The Budget and Reforecast Calendar for a Company Running Its First One

The budget is not the deliverable. The reforecast cadence is.

A company running its first budget usually spends weeks producing a document that is wrong by the end of the second quarter and never opened again. The document is rarely the problem. Nobody scheduled the revisiting, so the revisiting did not happen, and by the middle of the year the plan has quietly become a historical artifact that everyone is slightly embarrassed to mention.

So the position, stated at the front rather than at the end: a mediocre budget on a real reforecast cadence beats an excellent budget with no cadence. If you have the appetite for one of the two and not both, put the reforecast dates in the calendar first and let the budget be rougher than you would like. A rough plan that gets corrected four times is a management instrument. A precise plan that gets corrected never is a decoration, and it cost more to produce.

What follows is the calendar we build for that company. The first half is the build, which happens once. The second half is the cycle, which is the part that actually earns the fee.

Two things that have to be true before you start

Not a long list of preconditions. Two, and both of them are pass or fail.

The last three closes were completed and reconciled. A budget is built on a baseline, the baseline comes from your own actuals, and if the last three months were closed late with unreconciled balance sheet accounts then you do not have a baseline. You have a set of numbers that will move after you have budgeted against them. A company in that position should fix the close and budget next year. That is an unwelcome answer and it is the right one, because the alternative is a full budget cycle whose output nobody can defend the moment the first real variance appears.

The chart of accounts you are budgeting onto is the one you will still be using at the end of next year. Budgets are keyed to account codes. If a rebuild of the chart is anywhere on the horizon, do the rebuild first. Budgeting onto a structure you are about to replace orphans every line on cutover, and the repair job is worse than the original build.

How the dates in this calendar work

Every row below is anchored either to your fiscal year end or to a quarter close. Nothing here is anchored to a month name.

“YE minus 12” means the twelfth week before your fiscal year end. “The first full week after the Q2 close is issued” means exactly that: not a date, but the week following the week in which the second quarter’s reporting package actually goes out.

That convention is not fussiness. A budget calendar written in month names silently assumes a December year end, and a company with a March or a September year end that follows it will start its build in the wrong quarter and schedule its reforecasts against quarters that are not its own. Written in relative weeks, one calendar works for every year end, which is the only way a calendar survives more than one year.

Take the relative weeks as approximate. The sequence matters and the intervals matter. Whether the baseline is extracted in week twelve or week eleven does not.

The build, from kickoff to the lock

Week	What happens	Who is asked	What lands
YE minus 13	Publish the calendar itself, with the lock date on it, and get the owner to agree the lock date before any other work starts. A budget cycle without an agreed end date runs until everybody is exhausted.	Controller, owner agrees	The dated calendar, lock date fixed
YE minus 12	Extract the baseline. Trailing twelve months of actuals restated onto next year’s account structure, with one-off items separated out and individually labelled rather than buried in the run rate. For most companies the third quarter has just closed, so the last quarter of the baseline comes from the Q3 reforecast landing zone described below.	Controller	Baseline sheet, one-offs listed apart
YE minus 11	Decide how revenue will be built. Not the number, the method: by named contract with renewal dates, by pipeline stage with a conversion assumption, by service line against a rate card, or by delivery capacity. Pick one and write it down.	Owner, and whoever runs sales	A written revenue build method
YE minus 10 to YE minus 9	Build the revenue draft by that method. Every assumption written in the cell beside the number it drives, not in a separate memo nobody opens again.	Owner and sales	Revenue by month, by driver, assumptions attached
YE minus 9	The headcount conversation, held on its own, in its own meeting. Roles, start months, fully loaded cost, and the replacements for people you already know are leaving. This is usually the largest single decision in the budget and it does not survive being handled as a line in a cost submission.	Owner decides, controller costs it	A dated hiring plan by month
YE minus 8	Issue the cost submissions, pre-filled with each holder’s own baseline. A blank template comes back late and wrong. A pre-filled one comes back as an argument with the baseline, which is what you want.	Controller issues, budget holders receive	A submission pack per holder, with a due date
YE minus 7	Submissions due back. Chase on the due date, not a week after it. Whatever has not arrived by the end of this week goes into draft one at baseline, with the holder’s name against it.	Budget holders	Submissions in, gaps named and attributed
YE minus 6	Draft one. Consolidate everything received, phase it by month, and circulate it knowing it is wrong.	Controller	Draft one, circulated unpolished
YE minus 5	The draft one meeting. One agenda item: the gap between what the revenue draft produces and what the cost base consumes, and what the owner intends to do about it.	Owner, controller, budget holders as needed	A written list of decisions required, with names
YE minus 4	Rework on those decisions, and convert the result into cash. Monthly cash view, capital plan, debt service, instalment obligations, and covenant tests projected forward to each test date.	Controller	Draft two, with a cash line and a covenant projection
YE minus 3	Draft two review. Decisions taken and recorded in writing this week, because there is no longer room to defer one.	Owner	Decisions closed, owners against each
YE minus 2	Load the budget into the ledger, by account and by month. This is a working week, not a formality, and it is described below.	Controller	Budget in the ledger, properly phased
YE minus 1	Lock and publish. On the same day, publish the reforecast dates for the whole coming year and put them in the calendars of everyone who will be asked to attend.	Controller, owner signs	Locked budget, plus four dated reforecast entries

Thirteen weeks of elapsed time. Considerably less than thirteen weeks of work, because most of those weeks are waiting on somebody else, which is precisely why the calendar has to exist. The elapsed time is the constraint, not the effort.

Draft one is supposed to be wrong

This is the row where a first budget cycle most often loses a fortnight, and the failure looks like diligence.

The submissions arrive. They do not add up. Revenue is optimistic, the cost side has grown more than the revenue side, and the whole thing lands somewhere the owner will not accept. The controller can see all of that, so the controller does the responsible-seeming thing and works on it for a week before circulating, trimming the obvious padding, softening the revenue, absorbing part of the gap.

Do not do that. Circulate the sum, unpolished, in week YE minus 6.

The purpose of draft one is not to be credible. It is to put the size of the gap in front of the one person who can close it, with six weeks left before the lock date and enough runway to act on it. A draft one that arrives already reasonable has usually had the gap quietly absorbed by the finance function, which means the owner was never asked the question, which means the question gets asked in the second quarter instead, when the hiring has already started.

Two numbers make draft one useful and neither of them is the annual total: the size of the gap, and the month in which the cash position is thinnest. Put both on the front page.

Phasing, which is the step that gets skipped

Week YE minus 2 exists because of a specific and very common failure: a budget entered into the ledger as an annual figure divided evenly across twelve periods.

That budget produces a variance in every month with any seasonality in it, and every one of those variances is meaningless. Within a quarter, everyone learns that budget variance is noise. By the middle of the year nobody reads the variance column, and the reporting that the whole cycle was supposed to feed has been trained into irrelevance by a data entry shortcut.

Phase it properly:

• Revenue on whatever driver you chose in week YE minus 11, month by month.
• Payroll on the actual pay calendar, including the months that carry an extra run.
• Anything with a renewal date in the month it renews. Insurance, software, licences, the workplace insurance and benefit plan costs that sit on a calendar year clock regardless of when your fiscal year ends.
• Commission and bonus in the periods they are earned or paid according to your own policy, not spread flat.
• Capital in the month the commitment is made, and the cash effect in the month it is actually paid.

The test is simple. Look at the budgeted result for your seasonally weakest month. If it looks the same as your strongest month, the phasing has not been done.

The lock

Once the budget locks, it does not change for the rest of the year. Not when a large customer leaves. Not when a hire is deferred. Not when the second quarter is uncomfortable.

This is the rule everything else depends on, and it is the one that gets broken quietly. Somebody proposes a small adjustment in the third month to reflect what is now known. It is entirely reasonable. It also destroys the only fixed reference point the year has, and the second amendment is much easier to argue for than the first was. A budget that has been amended twice has no variance history, which means the year’s management reporting has nothing to measure against and the post mortem has nothing to learn from.

When the business changes, the reforecast changes. That is the entire reason there is a reforecast.

There is one exception and it is deliberately narrow: a restructuring, an acquisition or a loss large enough that the locked budget describes a different company than the one now operating. In that case you re-budget, you state in writing that you have done so and why, and you keep both versions with their dates on them so that the year can still be read afterwards.

The reforecast cycle, which is the actual programme

Four anchors, three of them reforecasts, and they are not the same size as each other.

Anchor	Horizon	Depth	Output
First full week after the Q1 close is issued	Remaining three quarters, plus the full-year landing	Full rebuild of revenue on the current contract list and current pipeline. Cost base re-phased onto actual hiring rather than planned hiring.	Reforecast one, with each material movement from budget explained by a decision or an event rather than by an amount
First full week after the Q2 close is issued	Remaining two quarters, plus the landing	The deepest of the three. Everything in reforecast one, plus the capital plan re-approved line by line or cut, the covenant projection re-run to each remaining test date, and an explicit keep or drop decision on every planned hire that has not yet started.	Reforecast two, and a short list of decisions the owner has to take this quarter
First full week after the Q3 close is issued	The final quarter only	Shallow. A landing zone, not a rebuild. What the year will actually finish at, and what has to happen in the remaining weeks for that to hold.	Reforecast three, which is also the closing quarter of next year’s baseline
After the year end close	None	No reforecast. The year is over.	The post mortem, described below

Now the part that makes this a cycle rather than a list.

Reforecast three and the next budget build are the same week. Look at the two calendars together: the third quarter closes roughly a quarter before year end, its package is issued a week or so later, and that lands within a week of YE minus 12, which is the baseline extraction row of the build calendar above. Treat that overlap as the join rather than as a collision. The landing zone that reforecast three produces is the last quarter of the trailing twelve months that next year’s budget starts from.

A company that skips the Q3 reforecast begins its next budget from a baseline containing a quarter it has not yet lived and has no view on. It will then discover in week YE minus 5 that the gap it is arguing about is partly an artifact of a stale assumption about the quarter currently in progress.

If you can only sustain one reforecast a year, make it the one after the second quarter closes. By then the year’s actual shape is visible rather than theoretical, and two full quarters remain in which to do something about it. The Q1 reforecast is often too early to say much that is not already in the budget. The Q3 reforecast is too late to change the year, though it earns its place by feeding the next one. Q2 is where a reforecast can still alter an outcome, which is the only test that matters.

Do not reforecast monthly

At this size, a monthly reforecast is the wrong instinct, and it is usually proposed by the most engaged person in the room.

It fails for two reasons. The first is capacity: a genuine reforecast is several days of work involving people outside finance, and doing it twelve times a year means either that finance does nothing else or that ten of the twelve are a spreadsheet copied sideways with two cells changed.

The second reason is worse. If the reference point moves every month, then every month’s actual result is compared against a number set four weeks earlier by someone who already knew most of what was going to happen. Nothing is ever a surprise, nobody is ever accountable to a plan, and the organisation loses its ability to distinguish between a business that is performing and a forecast that is being managed.

The monthly job is scoring, not reforecasting. In the monthly pack, actual runs against two columns: the locked budget, and the current reforecast. Each column answers a different question. The budget column asks whether the year is going as planned. The reforecast column asks whether the most recent correction was any good. Both are useful. Neither survives being merged into the other.

The triggers that override the cycle

A quarterly cadence is the routine. Some events do not wait for the routine.

Event	Response	When
A customer above your concentration threshold is lost, or won	Rebuild the revenue side and re-run the cash view off-cycle. Do not wait for the quarter.	Within the close that follows the event
A planned hire is deferred by more than a quarter, or a role is added outside the plan	Re-phase the cost base and the cash view, and state plainly that the full-year number has moved	At the next close
A covenant projection comes within tolerance of a test	Reforecast forward to the test date specifically, and open the lender conversation before the test rather than after it	Immediately
A financing, lease or capital commitment outside the approved list	Reforecast the cash view and the balance sheet before the commitment is signed	Before signature, not after
A pricing change	Rebuild revenue on the driver, with the volume assumption stated separately from the price assumption so that the two can be scored apart later	The month the change takes effect

The fourth row is the one companies argue about. Reforecasting before a commitment feels like bureaucracy when the decision has effectively been made. It is the only point at which the reforecast can change anything, and a company that only ever reforecasts after signing is producing a record rather than a forecast.

Who is asked for what, and what they are not asked for

The second column of this table is the one that prevents most of the damage.

Person	Asked for	Deliberately not asked for
Owner or CEO	Revenue direction, the headcount decision, and the resolution of the draft one gap	The revenue build itself. An owner who builds the model produces a target, and a target scored as a forecast makes every subsequent variance an accusation.
Budget holders	Their own cost lines, argued against a pre-filled baseline, with a reason for each departure from it	The company total and the revenue number. A budget holder who can see the total negotiates against the total instead of defending their own lines.
Controller	Consolidation, phasing, cash conversion, the covenant projection, the calendar, and the chasing	The revenue assumption. Whoever sets revenue should not also be the person scoring it, and in a small company that separation is the only one available.
Operations or delivery lead	Capacity constraints, and which revenue in the draft cannot physically be delivered with the plan as it stands	Cost allocation methodology, which is a finance question wearing an operations costume

The operations row is the one most first budgets omit entirely. A revenue plan that the delivery side has never been shown is a plan to sell work you may not be able to do, and the constraint surfaces in the second quarter as a delivery problem rather than in week YE minus 9 as a budget problem, which is where it was cheap to fix.

The post mortem, after the year end close

One meeting, held after the year end close is complete, and it produces exactly two things.

The first is a written list of where the budget was most wrong and why, sorted by size rather than by embarrassment. Each item gets a cause: a decision that changed, an event nobody could have seen, or an assumption that was wrong at the time it was written. Only the third category is a lesson. The other two are just what a year is.

The second is a single change to next year’s build calendar. One. A company that leaves the post mortem with eleven process improvements will implement none of them, and a company that moves one submission deadline earlier every year for four years ends up with a build calendar that fits its own business rather than a generic one.

What we do not do

We do not build a budget for a company whose recent closes are unreconciled, because the output would not be defensible and we would be paid to produce something that gets abandoned in the second quarter.

We do not amend a locked budget so that a variance reads better, and we say so at the point the amendment is proposed rather than afterwards.

We do not run a monthly reforecast at this size, and we will argue against one.

We do not load a budget into the ledger as an annual figure divided by twelve.

And we do not write a budget calendar in month names, because a great many incorporated Canadian companies do not end their year in December, and a calendar that only works for the ones that do is not a calendar.