How many budget versions should you go through before approval?

Two to three substantive versions is the right number for a mid-market annual budget: a first pass built from actuals against a top-down target, one negotiated round that closes the gap, and a final version for board approval. Most teams run more. In our 2027 budget season survey of more than 250 finance leaders, 78% went through three or more full revisions — up from 63% the year before.

That fifteen-point jump is the most useful number in the whole survey, because cycle length barely moved in the same period. Teams are not planning longer; they are planning the same thing repeatedly. Every extra round is not a finance task — it is twenty or more budget owners redoing work, which is why revision count is a better predictor of budget-season pain than any other single metric.

Bottom line: aim for three versions and treat a fourth as a process failure worth diagnosing, not a normal cost of planning. The number of rounds is set almost entirely by two decisions made before templates go out: whether the top-line target is agreed, and whether assumptions are frozen on a published date.

A version is not a revision

Half the confusion in this conversation is vocabulary. Three things get called a version and they cost wildly different amounts.

Only the first one is expensive, because it is the only one that puts the business back to work. A team that runs two versions and fifteen revisions is in good shape. A team that runs five versions is burning goodwill it will need in March. When someone says the budget went through six rounds, the useful follow-up is: how many of those pulled department heads back in?

This distinction is also what makes scenario planning cheap and re-collection expensive. If your tool can only express an alternative by copying the whole model, every what-if becomes a version, and the count inflates for reasons that have nothing to do with disagreement.

Why teams end up with five

Four causes, in order of how often they are the real one.

The target was not settled before templates went out. This is the dominant cause and the most fixable. If leadership has not agreed the top-line number, the bottom-up total will miss it, and the miss triggers a full re-collection. Our survey found hybrid kickoffs — setting top-down targets and then collecting bottom-up against them — rose from 31% to 43% of teams in a year, which is the right direction of travel for exactly this reason.

Assumptions moved mid-cycle. A pricing change, a funding event, a hiring freeze. Some of this is genuinely unavoidable. What is avoidable is not having a published freeze date, so that a late assumption change becomes an explicit decision to re-open the plan rather than something that happens by default.

Nobody said how many rounds there would be. Round count expands to fill the calendar. Teams that state the number at kickoff and hold to it run fewer rounds than teams that leave it open, for the same reason any deadline works.

Consolidation was slow enough that the numbers went stale. If it takes finance two weeks to produce a roll-up, the roll-up describes a business that has moved on, and the next round is partly about catching up rather than deciding anything. This is the cause that tooling actually fixes.

The version-control problem underneath

Round count and version control are the same problem seen from two angles. Ninety percent of the finance leaders in our survey use Excel in their budgeting process, up from 78% the year before, and 97% use a spreadsheet somewhere in the cycle — including 96% of teams that already own dedicated FP&A software. Spreadsheets are not the problem. Uncontrolled copies of them are.

The failure is not gradual. It looks like this: a board pack goes out, someone notices the marketing number is from version three rather than version four, and the pack gets corrected. The cost is not the rework; it is that the next number you present starts from a deficit of trust. That single event does more damage to finance's standing than a whole slow cycle.

Three controls prevent it, and none require replacing the spreadsheet:

• One plan of record with a name and a date, distinct from every working copy. Everything else is explicitly a draft.
• Proposed is not approved. An owner can change their submission as often as they like without touching the plan of record until finance accepts it.
• A change log with attribution. What changed, when, who changed it, and why — which is also what makes a financial model audit trail useful rather than decorative.

How to cap the number of rounds

1. Agree the top-line target before any template leaves finance. The single highest-leverage move available, and it costs nothing but a meeting.
2. State the round count at kickoff. "Two collection rounds and a board version" is a sentence that saves weeks.
3. Publish an assumption freeze date. After it, changed assumptions go into the reforecast, not the budget.
4. Pre-populate templates from actuals with a guardrail per department. Owners reviewing a baseline produce fewer surprises than owners filling a blank sheet.
5. Consolidate continuously rather than at the deadline. If the roll-up is current, a round is a conversation rather than a rebuild.
6. Separate scenarios from versions explicitly. Give leadership a way to ask what-if that does not re-open the plan.

Items four and five are the ones tooling changes. Everything else is process, which is why our advice on shortening the annual budget cycle leads with sequencing rather than software.

What to track next cycle

If you want to improve this deliberately, instrument it. Four measures, all cheap to capture:

The last one is the most useful and the least measured. Unlike forecast accuracy, which moves with business conditions far more than with process, a corrected pack is a discrete countable event that is unambiguously your process failing. It is also the metric that survives scrutiny in an ROI case for FP&A software, precisely because nobody can argue about whether it happened.

Where this is heading

The interesting shift in this year's survey is that the budget is becoming less final. The share of teams that lock the budget for the year fell from 54% to 38%, only a small minority now lock it and walk away entirely, and quarterly or monthly reforecasting rose from 25% to 40%. That changes what a version even means: if the plan is going to move in February anyway, spending a fifth round perfecting it in December is misallocated effort.

The teams handling this well are running fewer budget versions and more frequent reforecasts — treating the annual plan as a target-setting exercise and the rolling forecast as the operating number. That is a genuinely different posture from the traditional annual cycle, and it needs tooling that supports reforecasting without a re-collection each time.

If coordination is what is driving your round count — and for most teams it is — collaborative budgeting with department owners is the mechanism to fix, and agentic budgeting is the version of that where agents handle the collection and reconciliation loop so each round costs finance a conversation instead of a week.

What a round actually costs

Round count feels abstract until you price it. Take a company with twenty-five budget owners. A collection round asks each of them for roughly two to four hours: reviewing the ask, pulling their own numbers, filling in the template, answering follow-ups. Call it three hours. That is seventy-five hours of non-finance time per round, before finance touches it.

Then add finance's side: chasing the last submissions, reconciling formats, rebuilding the roll-up, and re-cutting the comparison to target. For a lean team that is another twenty to thirty hours. So one avoidable round is roughly a hundred hours of company time, and it produces no new information — it re-produces information you already had at a different assumption.

That is the arithmetic behind the survey finding. Teams reporting three or more revisions rose from 63% to 78% year over year while cycle length barely moved, which means the extra rounds are being absorbed by working harder rather than taking longer. It shows up as the 92% of finance leaders who lose evenings or weekends to the season and the 58% who spend half the season or more on busywork.

What a healthy three-round cycle looks like

Concretely, so you can compare it to yours.

1. Round one: baseline review. Owners receive a pre-filled packet built from their trailing actuals with a stated guardrail. They confirm what continues, flag what changes, and name what they need that is not in last year's numbers. Most lines should not change, which is the point.
2. Round two: the negotiated round. Finance has consolidated, knows the gap to target, and goes back to specific departments with specific asks — not a general request for everyone to resubmit. This is targeted, and it is the round most teams accidentally turn into a full re-collection.
3. Round three: the board version. Final numbers, phased, with the narrative built from the decisions on record. No new collection; this is finance assembling.

The discipline that makes round two targeted rather than general is having the gap attributed by department before you go back out. If you know marketing is $400k over guardrail and engineering is on plan, you have one conversation instead of twenty-five. If all you know is that the total misses by $1.2m, you re-open everything.

That attribution is a consolidation-speed problem, which is why continuous checking beats deadline-driven assembly. It is also the practical argument for collaborative budgeting tooling over emailed workbooks: a submission checked against its guardrail on arrival tells you where the gap is while there is still time to be surgical about it.

When more rounds are the right answer

Three situations where a fourth round is a good decision rather than a failure, and it is worth naming them so the cap does not become dogma.

A genuine strategic change. A funding round closes, an acquisition lands, a major customer churns. The business is different, so the plan should be. Re-open deliberately and say so.

The first cycle after a big process change. New tool, new chart of accounts, new leadership. Expect an extra round and budget for it rather than pretending it will go smoothly.

The plan is internally inconsistent. If the revenue plan assumes sales capacity the headcount plan does not fund, another round is cheaper than shipping a plan that cannot happen. This is exactly what the senior review pass is for, and it is the stage most often cut under time pressure.

What is not on that list: a round because leadership wants to see a different number without changing any assumption. That is a scenario, and giving leadership a fast way to ask for one is how you keep it from becoming a version.