The core answer
An annual budget is written at the start of the year with the information available then. When a customer leaves, materials get dearer or a project slips, the budget drifts so far from reality that managers stop opening it. A rolling forecast fixes this by moving the forecast forward every month: close one month, add a new month at the end. The company always looks 12 months ahead.
Three things make it work. Update the assumptions, not just the end-point totals. Review every month at the same time. And split variances into controllable and uncontrollable parts, so the meeting discusses decisions rather than blame.
This guide does not teach how to build a first budget; that is in building a first company budget. This one starts on the day the company already has a budget and must look after it every month. Every figure here is a hypothetical example, not a legal threshold.
Who should read which part
| Where the company is | Read first |
|---|---|
| Before deciding (no forecast yet) | Sections 1 to 3: horizon, level of detail, owners |
| In progress (running monthly cycles) | Sections 4 to 7: calendar, variance comparison, agenda |
| When the numbers are off plan | Section 6, Section 8 and the hypothetical example: separate the causes and see when to decide |
1. How does a rolling forecast differ from the annual budget?
| Annual budget | Rolling forecast | |
|---|---|---|
| Purpose | Agreed targets and spending limits | Decisions from the latest information |
| Period | The fixed financial year | The next 12 months, moved forward monthly |
| Changes | Not changed mid-year unless approved | Updated every month |
| Answers | How far are we from target? | Where are we heading, and what must we do now? |
They do not replace each other. A company that drops the budget and keeps only a forecast has no yardstick for how far it missed. A company with only a budget cannot tell whether cash will last three more months.
2. How far ahead, and how detailed?
- 12 months is a sensible starting point for an SME. Seasonal businesses or long contracts may look further, but very distant figures should not drive large decisions.
- First three months: detail by customer or key project and by actual receipts and payments. For weekly cash use a separate 13-week cash flow forecast.
- Months four to six: revenue and expense categories.
- Months seven to twelve: totals and trends, such as monthly sales and total fixed costs.
Less detail further out is deliberate. If every ledger line must be reworked each month, the cycle takes days and the team abandons it within a few rounds.
3. Which assumptions must be updated every month?
Do not start by editing the total. Start from the assumptions below and let the numbers follow.
| Assumption | Question each month | Who should own the data |
|---|---|---|
| Sales and open quotations | Which customers closed, were lost or slipped? What weight do the remaining opportunities deserve? | Sales; see sales pipeline and quotation follow-up |
| Prices and discounts | Do realised prices differ from plan? | Sales and management |
| Variable costs | Have suppliers changed prices, or scrap and rework rates? | Production or purchasing |
| People and payroll | Any leavers, joiners or changed hiring plans? | HR |
| Receivables and payables | How many days slower do customers pay than assumed? Any supplier needing earlier payment? | Accounting |
| Contracts and committed spending | Any lease, service contract or loan starting, expiring or repricing soon? | Owner of each contract |
| Capital spending | Which projects are postponed, cancelled or added? | Management |
Write each assumption as a short sentence on the first sheet, for example "the largest customer pays in 55 days on average". When an assumption changes, record the date and the reason. Whoever opens the file six months later can see why the numbers moved.
4. How long should the monthly cycle take, and who does what?
A cycle that works must be short enough to repeat. A sample calendar for an SME, adjusted to how quickly the company closes its books:
| Working day | Task | Owner |
|---|---|---|
| Days 1 to 5 | Close last month, release actuals | Accounting |
| Days 4 to 6 | Each category owner updates their assumptions | Heads of sales, production, HR |
| Day 7 | Merge the file; check month-end cash against the bank | Accounting |
| Day 8 | Send a one-page pack to attendees in advance | Coordinator |
| Day 9 or 10 | Review meeting and decisions | Management |
Three rules help: one category, one owner; submit only what changed since last round, not a full re-entry; and one shared file, not private copies edited separately.
5. Variance to budget or variance to the previous forecast?
Keep these two questions apart because they lead to different decisions.
- Variance to budget shows the distance from the agreed target. Use it to assess performance and to decide whether to ask formally for a revised target. The budget does not move, so the comparison holds all year.
- Variance to the previous forecast shows how good the forecasting is. Use it to improve the method. If the three-month-ahead forecast misses by more than your threshold (say 10 per cent) round after round, one of the assumptions is pointing the wrong way; it is not bad luck.
A simple measure is (actual minus forecast) divided by forecast, recorded every month by category. After six months the company can see which categories it predicts well and which need better information.
6. How do I separate controllable from uncontrollable variances?
Before discussing who should do what, sort each cause into one of three groups.
| Group | What it means | Examples | Response |
|---|---|---|---|
| Controllable | The company can act in the time left | Slow quotation follow-up, rework from internal errors, collecting late | Name an owner and a re-check date |
| Uncontrollable but responsive | Cause is outside the company, but the plan can adapt | Supplier price rise, customer postpones, currency move | Update the cash forecast, renegotiate, find alternatives |
| Timing only | Nothing is lost; it moves to another month | Customer pays after month-end, delivery slips over a month boundary | Move the number to the right month; no plan change |
To sort a cause, ask whether it came from an action or decision by people inside the company. If yes, it is controllable. If not, ask whether the amount is really lost or only delayed. This avoids two errors: blaming the team for every variance, and blaming "the market" while something fixable hides underneath.
One variance can have several causes. Break the amount into parts and give each part its own reason. Do not write a single line saying "market conditions".
7. What should a 60-minute monthly meeting cover?
The meeting exists to decide, not to read reports aloud. Send a one-page pack in advance and use the time as follows. If the company already holds a monthly management meeting, see monthly management meeting agenda and add this item rather than creating another meeting.
| Minutes | Item | Output needed |
|---|---|---|
| 0 to 10 | One-page summary: actual versus budget, actual versus previous forecast | Everyone sees the same picture |
| 10 to 25 | Variances over threshold, split controllable, uncontrollable, timing | Agreed causes |
| 25 to 40 | Changed assumptions and the new 12-month forecast | New numbers management accepts |
| 40 to 55 | Decision triggers (Section 8): what must be decided today | Decisions, owners, due dates |
| 55 to 60 | Review last round's actions | Open items carried forward |
Keep the minutes short: what was decided, who does it, by when, and which assumption changed.
8. When should the forecast trigger a decision?
A forecast without decision rules becomes a report people read and forget. Agree the triggers in advance, while there is no pressure. The figures below are hypothetical; adjust them to the size and risk of the business.
| Trigger (hypothetical) | Decisions to put on the table |
|---|---|
| Lowest cash in the next 6 months below 2 months of fixed costs | Postpone capital spending, speed up collections, negotiate supplier credit, approach financing before cash runs out |
| Sales forecast for the next 3 months more than 10 per cent under budget for 2 months running | Slow new hiring, review costs that do not move with sales, review customer acquisition |
| One customer above 30 per cent of sales with warning signs | Prepare a replacement plan, check contract terms, follow up more often |
| Overdue receivables above one fifth of all receivables | Check credit before shipping more, tighten payment terms |
| Expected gross margin more than 3 percentage points below target | Review selling prices, material costs and discounts |
| Forecast misses by more than 10 per cent several rounds in a row | Go back to the assumptions and data sources |
When a trigger is hit, record what you chose to do, or why you chose not to. Not deciding is also a decision and deserves a note.
9. Hypothetical example: an office equipment supplier and installer
This is a hypothetical example, not a real case. All figures are illustrative to explain the method and belong to no actual company.
The hypothetical company budgets annual sales of THB 12,000,000, or THB 1,000,000 a month. After closing June, actual sales for the half year are THB 5,400,000 against a budget of THB 6,000,000: THB 600,000 short.
Step 1: split the THB 600,000 variance
| Cause | Amount (THB) | Group |
|---|---|---|
| A customer moved to another supplier after two late deliveries | 350,000 | Controllable |
| Several quotations not followed up within 7 days | 100,000 | Controllable |
| Two customers postponed projects because of their own budgets | 150,000 | Timing (expected back in the second half) |
| Total | 600,000 |
Step 2: update assumptions and roll forward. The company removes the old assumption that "existing customers buy the same as last year", because one has gone. It places the postponed amounts in July and August and sets a new assumption of following up quotations within 3 days. The second-half forecast becomes THB 5,400,000 (THB 900,000 a month). Expected full-year sales are 5,400,000 plus 5,400,000, or THB 10,800,000, which is THB 1,200,000 below budget. This does not rewrite the budget. It is the latest picture for decisions.
Step 3: check last round's forecast accuracy. The forecast made at the end of May for June was THB 900,000; actual was THB 870,000. The miss is THB 30,000, about 3.3 per cent, inside the 10 per cent threshold, so the method needs no review yet.
Step 4: compare with the triggers. Lowest expected cash, in October, is THB 380,000, below the two-month fixed-cost threshold (assume THB 400,000). Management makes three decisions: postpone hiring a new installer by one quarter; have accounting chase overdue receivables; and ask the main supplier for 15 extra days of credit. Each has an owner and a re-check date at the next meeting.
What the example shows: most of the variance was something the company could act on, and a timing shift should not be booked as lost sales. Had everything been filed under "slow market", none of the three decisions would have been made.
10. What must be checked on the contract and legal side before acting on the numbers?
A weak forecast does not mean costs can be cut at once. Many items carry commitments, so check before acting.
- Leases and service contracts: early termination terms, penalties and notice dates. See building a first company budget, which covers contracts that have already locked in costs.
- Loan or facility agreements: some contain financial conditions set by the bank. A forecast approaching such a condition is a reason to read the agreement again.
- Contracts with key customers: if the forecast says a customer may leave, look at renewal, termination and any warranty that continues after delivery.
- People: reducing staff, hours or pay is subject to employment law and the employment contracts. Take advice before announcing anything; do not let a forecast cell dictate the instruction.
- Tax: a forecast is not a tax calculation. Where a decision affects tax, such as shifting revenue between months or setting up provisions, have the company's accountant or auditor check it first.
11. Mistakes that make rolling forecasts fail
- Too much detail: the cycle takes weeks and the team stops.
- No owner for assumptions: everyone assumes someone else updated them.
- Editing only totals: the end-point changes with no record of which assumption moved.
- Using the forecast as punishment: the team pads the numbers or hides bad news.
- No decision triggers: the report is read and nothing happens.
- Mixing cash and profit: a profitable business can still run short of cash; see profitable but short of cash.
12. Checklist: set up the cycle within three months
- Agree the 12-month horizon and the level of detail for each part
- Name an owner for each assumption category and a coordinator for the cycle
- Write the main assumptions as short sentences on the first sheet, with the date last changed
- Fix the monthly calendar (close, update, merge, pre-read, meeting)
- Keep the original budget unchanged and add a column for the previous forecast to measure accuracy
- Set the variance threshold that needs an explanation and the decision triggers (Section 8)
- Prepare a table splitting variance causes: controllable, uncontrollable, timing
- Prepare the one-page pack and a short meeting record form
- List contracts that commit spending, with expiry and notice dates
- Ask the accountant or auditor to review the tax-related assumptions
- Review the process after three rounds and drop what nobody uses
Common misunderstandings
"With a rolling forecast we no longer need a budget." Not so. The budget is the agreed yardstick for judging performance; the forecast supports decisions on the latest information. They do different jobs.
"A good forecast must be accurate every month." Not so. The aim is to see the trend and decide in time. Accuracy shows in the variance to the previous round, which should narrow as the habit settles.
"Every variance is the fault of the responsible team." Not so. Some come from outside factors or are just timing. Separate them first and the decisions become sound.
"If the cause is external, there is nothing to do." Not so. The cause may be beyond control, but cash plans, purchasing and pricing can still be adjusted.
"A weak forecast means costs must be cut immediately." Not so. Check contracts and commitments first; some changes engage employment law or tax, so consult the right people.
"Follow this guide and the business will survive." Not so. A forecast brings problems into view sooner and puts evidence behind decisions, but it guarantees no business result.
Further reading
- Building a first company budget: sales, expenses and cash
- 13-week cash flow forecast for SMEs
- Monthly management meeting: agenda and reports
- Sales pipeline, quotation follow-up and sales forecast
- Profitable but short of cash: working capital
A rolling forecast that works is a one-page set of numbers each month that separates the causes of variance and ends in decisions with an owner and a due date.
We help companies structure business-plan documents and spending approval routes, review the contracts that commit costs or revenue shown in a forecast, and coordinate with the company's accountant. We are not accountants, auditors or tax advisers, we do not prepare or certify budget or forecast figures, and we cannot guarantee any business result. Our fees depend on the stage of the matter and the work required: send us the documents first, and we will confirm the scope of work and the fees in writing before we start. Call +66 92 254 2045 or send us the details. See also business plan and accounting and tax.
This guide is published by Suwanvara Law Firm — a Khon Kaen law firm established in 1986. General information only, not legal advice on a specific matter.

