A rolling forecast is a 12-month financial projection that you rebuild every month: when one month closes, you load the actual results, extend the horizon one more month, and update the assumptions that drive revenue and cost. It suits US small and midsize business owners who need a current view of where the year is heading, not a stale annual plan. Its biggest limitation: a rolling forecast is only as good as your monthly bookkeeping close, and it answers monthly direction — not week-by-week cash timing and not annual target setting, which are separate exercises.
Quick answer
To run a rolling forecast, keep a 12-month-ahead model built from a handful of operating drivers (volume, price, variable cost rates, fixed overhead), refresh it within a week of each monthly close, and review the gap between forecast and actuals at the driver level. Plan roughly one focused hour per month once the model exists. Pair it with a weekly cash tool for liquidity timing — our free 13-week cash flow template handles the weekly view — and treat any separate annual budget as a different document with a different job. The sections below build a complete worked example: drivers, a seasonal 12-month window, the roll-forward, scenario bands, the owner cadence, and a forecast-error review.
What is a rolling forecast, and why does a small business need one?
A rolling forecast is a living projection with a fixed horizon — usually 12 months — that moves forward as time passes. Instead of building one budget in December and watching reality drift away from it all year, you reforecast monthly, so the projection always starts from your latest actual results and always looks the same distance ahead. The U.S. Small Business Administration (SBA) already pushes small businesses toward this granularity in its business plan guidance (accessed July 28, 2026), which recommends forecasted income statements, balance sheets, and cash flow statements, and says of the first year:
“For the first year, be even more specific and use quarterly — or even monthly — projections.” — U.S. Small Business Administration
The case for continuous updating is that conditions move. In the Federal Reserve Banks’ 2025 Report on Employer Firms, based on the 2024 Small Business Credit Survey of 7,653 US employer firms with 1–499 employees (accessed July 28, 2026):
“More than half of firms cited paying operating expenses (56%) or uneven cash flows (51%) as challenges.” — Federal Reserve Banks, 2025 Report on Employer Firms
Conditions have not settled since. The 2026 report on the 2025 survey (fielded September 3 to November 14, 2025; published March 3, 2026; accessed July 28, 2026) found 77% of employer firms reporting rising costs of goods, services, and/or wages, tariff-related cost challenges, or both, and revenue-growth expectations at their lowest level since the 2020 survey. A projection built once a year cannot absorb swings like that; a monthly reforecast can.
This is not a new idea. In their widely cited 2003 Harvard Business Review essay “Who Needs Budgets?”, Jeremy Hope and Robin Fraser argued that traditional annual budgeting blocks “nimble adjustments to market conditions” — the problem a rolling forecast is designed to solve. Yet adoption is far from universal: in the 2026 AFP FP&A Benchmarking Survey from the Association for Financial Professionals (AFP), conducted in August and September 2025 among 332 corporate finance practitioners, only 43% of organizations used rolling forecasts. As AFP’s Director of FP&A (financial planning and analysis) Practice put it:
“Finance’s response to an unpredictable future must be to maintain multiple points of view of what can happen. Inflexible budgets break.” — Bryan Lapidus, Association for Financial Professionals
How do you build a driver-based 12-month rolling forecast model?
A driver-based model computes the financials from a small set of operating inputs you can observe and change, rather than typing guessed revenue into each month. The method has four steps:
- Pick the window. Twelve monthly columns starting the month after your last close (the example below runs October 2026 through September 2027).
- Choose 3–7 drivers. Typically volume (jobs, orders, billable hours), price (average ticket, rate), variable cost rates (materials, direct labor), and fixed overhead. Fewer, observable drivers beat many vague ones.
- Write the formulas. Revenue = volume × price. Variable costs = rate × revenue. Operating income = revenue − variable costs − fixed overhead.
- Add seasonality. Apply a monthly index from your own prior-year revenue pattern so the window reflects your busy and slow months.
The worked example throughout this article is a hypothetical illustration with made-up inputs for a fictional US heating, ventilation, and air conditioning (HVAC) services company with six field technicians. It is a management-reporting view — the way an owner should plan the business — not a tax-basis or US Generally Accepted Accounting Principles (GAAP) book calculation, and every input is invented to show the method.
Table 1: Hypothetical monthly drivers for a six-technician HVAC company (made-up inputs).
| Driver | Made-up value | Where it comes from in a real business |
|---|---|---|
| Field technicians | 6 | Payroll roster |
| Completed jobs per technician-day | 3.0 | Scheduling or field-service software |
| Working days per month | 21 | Calendar |
| Average ticket per job | $650 | Invoicing history |
| Materials cost | 28% of revenue | Job-cost reports |
| Field labor (fully loaded) | $34 per paid hour | Payroll, including taxes and benefits |
| Fixed monthly overhead | $85,000 | Bookkeeping: rent, vehicles, office, marketing |
Interpretation: the model needs only seven inputs, and each one maps to a record a US small business already keeps — that is what makes a monthly refresh cheap enough to sustain.
Table 2: Base-month calculation from the Table 1 drivers (made-up inputs, USD).
| Line | Method | Amount |
|---|---|---|
| Jobs | 6 techs × 3.0 jobs/day × 21 days | 378 jobs |
| Revenue | 378 jobs × $650 | $245,700 |
| Materials | 28% × $245,700 | $68,796 |
| Field labor | 6 techs × 8 hrs × 21 days × $34 | $34,272 |
| Gross margin | $245,700 − $68,796 − $34,272 | $142,632 (58.0%) |
| Fixed overhead | per Table 1 | $85,000 |
| Operating income | $142,632 − $85,000 | $57,632 (23.5%) |
Interpretation: this base month is the company’s average month — profitable, with a 58% gross margin and 23.5% operating margin at these invented inputs. Every other month is the same calculation scaled by a seasonal index.
Table 3: The 12-month rolling window, October 2026–September 2027 (made-up seasonal indices that sum to 12.00; revenue = index × $245,700 base month).
| Month | Seasonal index | Forecast revenue |
|---|---|---|
| October 2026 | 0.95 | $233,415 |
| November 2026 | 0.85 | $208,845 |
| December 2026 | 0.80 | $196,560 |
| January 2027 | 0.80 | $196,560 |
| February 2027 | 0.85 | $208,845 |
| March 2027 | 0.95 | $233,415 |
| April 2027 | 1.05 | $257,985 |
| May 2027 | 1.10 | $270,270 |
| June 2027 | 1.20 | $294,840 |
| July 2027 | 1.25 | $307,125 |
| August 2027 | 1.20 | $294,840 |
| September 2027 | 1.00 | $245,700 |
| Total | 12.00 | $2,948,400 |
Interpretation: with materials at 28% ($825,552), field labor held at $34,272 per month ($411,264 for the year), and overhead at $85,000 per month ($1,020,000), full-year operating income is $691,584 — about 23.5% of revenue. The simplification here is deliberate: labor and overhead stay flat all year, so seasonality flows through revenue and materials only. In a real model you add a hiring row when a driver crosses a threshold — for example, bringing on a seventh technician when jobs per technician-day run above 3.2 for two consecutive months — which is exactly the kind of decision the forecast exists to time.
How does the monthly roll-forward work?
The roll-forward is the mechanic that makes the forecast “rolling.” At each monthly close:
- Replace the just-closed forecast month with actual results from the bookkeeping.
- Extend the horizon one month, so you always see 12 months ahead (October 2027 joins the window when October 2026 closes).
- Update the drivers with what the actuals taught you — new average ticket, new jobs-per-day rate, new materials percentage — rather than editing totals.
Table 4: Window before and after the October 2026 close (hypothetical illustration).
| Step | Forecast window | What changed |
|---|---|---|
| Built late September 2026 | Oct 2026 – Sep 2027 | 12 forecast months from Table 3 |
| After October 2026 closes | Nov 2026 – Oct 2027 | October becomes actuals ($225,700 revenue vs. $233,415 forecast); October 2027 is added at the 0.95 index on the updated base month of $234,360 = $222,642 |
Interpretation: the updated base month falls from $245,700 to $234,360 because the average-ticket driver was revised from $650 to $620 (378 jobs × $620); the next section shows why that revision happened. Notice what did not change: the structure, the formulas, and the horizon length. Only inputs move.
How wide should your scenario bands be?
A single-number forecast implies a precision no small business actually has, so carry a band around the base case. AFP’s benchmarking work distinguishes structured scenario planning — considering a range of circumstances and projecting outcomes for each — from simple sensitivity analysis, and finds the structured version associated with better alignment and faster planning cycles. For a small business, three cases built from the same drivers are enough.
Table 5: Base-month scenario band from two driver moves (made-up inputs; labor hours and overhead held fixed).
| Scenario | Jobs per tech-day | Average ticket | Revenue | Materials (28%) | Field labor | Overhead | Operating income |
|---|---|---|---|---|---|---|---|
| Downside | 2.6 | $620 | $203,112 | $56,871 | $34,272 | $85,000 | $26,969 (13.3%) |
| Base | 3.0 | $650 | $245,700 | $68,796 | $34,272 | $85,000 | $57,632 (23.5%) |
| Upside | 3.4 | $670 | $287,028 | $80,368 | $34,272 | $85,000 | $87,388 (30.4%) |
Interpretation: moving just two drivers swings monthly operating income across a roughly $60,000 range — 53% below base in the downside and 52% above in the upside at these invented inputs. The band is a decision tool: size your minimum cash buffer and your hiring and spending gates against the downside case, not the base case. Weekly liquidity timing inside those months is a separate question, which is where the 13-week cash flow template takes over.
What monthly cadence should the owner follow?
The forecast fails as a document and works as a rhythm. A realistic US small-business cadence:
- Days 1–5 after month-end: close the books. Load actual revenue, materials, labor, and overhead. No close, no reforecast — this is the dependency most owners underestimate, and it is why the forecast discipline starts with timely remote bookkeeping support or an equivalent monthly close process.
- Days 5–7: reforecast. Replace the closed month with actuals, extend the horizon one month, and update drivers. With a driver-based model this takes under an hour.
- Within a week of close: hold a 60-minute owner review. Three questions: What changed in the drivers? What does the band now say about hiring, pricing, and spending? Does any month in the next 12 breach the downside case?
- Quarterly: refresh the scenario bands and seasonal indices against the latest 12 months of actuals, and push the monthly cash implications into your weekly cash tool.
- Annually: set targets separately. An annual budget is a target-setting exercise with different mechanics; it is deliberately outside this article’s scope.
Owners who want this cadence run for them — model maintenance, driver review, and the monthly decision meeting — typically hand it to an outsourced finance function; see our overview of remote CFO services for growing businesses for how that work is organized, and our financial planning blueprint for startups if you are building your first plan from scratch.
How do you review forecast error and get better each month?
A rolling forecast earns its keep through a short, honest error review. Use two measures per month:
- Error = actual − forecast (dollars), and % error = error ÷ forecast.
- Mean absolute percentage error (MAPE) = the running average of your monthly absolute % errors — your accuracy trend line.
Table 6: October 2026 forecast-error review (hypothetical illustration, made-up actuals).
| Item | Forecast | Actual | Error | Diagnosis |
|---|---|---|---|---|
| Jobs | 359 (0.95 × 378) | 370 | +11 jobs | Volume beat plan (2.94 vs. 2.85 jobs per tech-day) |
| Average ticket | $650 | $610 | −$40 (−6.2%) | Mix shifted toward maintenance calls, away from installs |
| Revenue | $233,415 | $225,700 | −$7,715 (−3.3%) | Entirely a price/mix miss, not a demand miss |
Interpretation: a −3.3% revenue error decomposes into a volume beat and a ticket miss, and the corrective action follows the driver, not the total: the owner reviews discounting and job mix, and the ticket driver moves to $620 for the reforecast (a judgment call, not a formula) while the volume driver stays. Two review rules make this stick. First, distinguish noise from bias: errors alternating in sign are noise, but three or more consecutive misses in the same direction mean a driver is systematically wrong and must move. Second, never force the total to match by editing the output cell — fix the input that reality contradicted, or the model stops teaching you anything.
FAQs
How is a rolling forecast different from an annual budget?
A budget is a fixed annual target set once; a rolling forecast is a 12-month projection rebuilt every month from the latest actuals. They answer different questions — “what did we commit to?” versus “where are we actually heading?” — and many small businesses run only the rolling forecast.
Do I need special software to run a rolling forecast?
No. A spreadsheet with monthly columns and the formulas in this article is enough for most small businesses. The real requirement is clean, on-time monthly bookkeeping, because the model consumes actuals every month.
How far ahead should a small business forecast?
Twelve months suits most businesses because it always captures a full seasonal cycle. Extend the horizon only when a specific decision — a lease, a loan, a large hire — needs a longer view.
How accurate does a rolling forecast need to be?
Directionally right and improving. Track your mean absolute percentage error and investigate driver-level misses and repeated same-direction bias. A forecast that is 3% off and explains why beats one that is 1% off by luck.
What if my books are always two months behind?
Fix the close first. A rolling forecast fed two-month-old actuals is just a stale budget with extra steps. Get bookkeeping to a five-business-day close, then start the reforecast rhythm.
The bottom line
A rolling forecast is a monthly habit, not a document: build a 12-month window from observable drivers, roll it forward at every close, carry a downside band for spending and hiring gates, and review each month’s error at the driver level. Start with the worked structure above, connect the monthly output to weekly liquidity with the 13-week cash flow template, and if you want the model built and the monthly cadence run for you, talk to our Remote CFO team about building your rolling forecast.
This article provides general educational information, not tax, legal, accounting, or investment advice. All worked examples are hypothetical illustrations with made-up inputs; your drivers, seasonality, and margins will differ, so build the model on your own books and consult qualified advisors for decisions specific to your business.