Rolling Forecast Template (Driver-Based, Forecast Accuracy)

Driver-based rolling forecast: horizon, driver owners, bridges, revenue, cost, headcount and cash outlook, scenarios, MAPE and bias, decisions

財務・投資・会計フルFP&A practice; Beyond Budgeting principles (names); FAST Standard 02c (by reference)rolling forecast math更新 2026/10/24

エディタで開く.mdをダウンロード

使い方: Use it each quarter or month to update the landing view and the next six quarters from operational drivers. Fill the driver table and its owners first, then bridge from the previous forecast. The main pitfall is never measuring forecast accuracy, so optimism or sandbagging goes unnoticed.

プレビュー

Rolling Forecast: [Organisation name] [Cycle, e.g. RF 2027-Q1]

Driver-based rolling forecast that extends a fixed number of periods ahead every cycle: purpose, horizon and cadence, drivers and driver owners, the bridge from the previous forecast, revenue, margin, cost, headcount and cash outlook, scenarios, forecast accuracy, risks and opportunities, and the decisions requested. Written as original practice guidance for financial planning and analysis (FP&A); no standard prescribes a rolling forecast. Replace every [bracketed] value and every row marked Example. This is a document format, not financial, investment, tax, legal or accounting advice.

Document control

Field	Value
Document ID	[RF-2027-Q1]
Organisation and scope	[Group / Segment] Example: Corvid Commerce Group, consolidated
Actuals cut-off	[YYYY-MM-DD] Example: 2027-03-31
Forecast horizon	[Periods] Example: Q2 2027 to Q3 2028 (six quarters, monthly detail for the first two)
Cadence	[Monthly / Quarterly] Example: quarterly full refresh, monthly flash on revenue and cash
Comparison bases	[Budget version], [previous forecast ID] Example: BUD-FY2027 v1.0; previous forecast = budget (first cycle of the year)
Currency and units	[EUR thousand, budget FX rates unless stated]
Version	[0.1]
Status	[Draft / In review / Approved / Superseded]
Owner	[@head-of-fpa]
Approvers/reviewers	[@cfo], [@ceo], [@driver-owners]
Last updated	YYYY-MM-DD
Classification	[Confidential]
Model file	[Path and version]

Approvals

Role	Name	Decision	Date
Chief Financial Officer	[@name]	[Approve / Return]	YYYY-MM-DD
Executive committee	[@ceo]	[Noted / Decisions taken, see section 10]	YYYY-MM-DD
Board (quarterly, for information)	[@chair]	[Noted]	YYYY-MM-DD

Revision history

Version	Date	Author	Change summary
0.1	YYYY-MM-DD	[@name]	Drivers collected from owners
0.5	YYYY-MM-DD	[@name]	Example: First consolidation, commerce volume cut after March trading
1.0	YYYY-MM-DD	[@name]	Example: Approved by CFO, presented to executive committee

Purpose, horizon and cadence
Drivers and driver owners
Changes since the last forecast
Revenue and margin outlook
Cost and headcount outlook
Cash and liquidity outlook
Scenarios
Forecast accuracy
Risks and opportunities
Decisions requested
Process, controls and model governance
Glossary and references

1. Purpose, horizon and cadence

[State the purpose, the decisions this forecast informs and who relies on it.]

Example: The rolling forecast gives the Corvid executive committee a six-quarter view of revenue, EBITDA, headcount and cash so that it can redirect spending inside the approved budget envelope, plan hiring ahead of demand and keep liquidity headroom above the treasury limit. It also produces the full-year landing view used in the quarterly board report and in lender updates.

Design choice	Setting	Reason
Horizon (Example)	Six quarters, always rolling	Covers the next budget year before the budget cycle starts
Granularity	Months for the first two quarters, quarters after that	Detail where it can be acted on
Cadence	Full refresh each quarter; flash on revenue and cash monthly	Balances effort and freshness
Level of detail	Segment, function, entity; not cost centre line items	Forecast what moves the result
Currency	Budget rates for comparison; actual rates for cash and covenants	Separates performance from translation
Lock date	Working day 8 after quarter close	Allows actuals to settle

2. Drivers and driver owners

Driver ID	Driver	Unit	Budget (Q2 to Q4 2027)	Forecast (Q2 to Q4 2027)	Owner	Lines moved
DRV-001 (Example)	New ARR bookings	EUR thousand	11,000	11,000	[@cro]	Subscription revenue, commissions
DRV-002 (Example)	Price uplift on renewals from 1 July	% of renewing ARR	0%	3%	[@cro]	Subscription revenue
DRV-003 (Example)	Gross churn	% of opening ARR, annualised	4.3%	4.3%	[@customer-success]	Subscription revenue
DRV-004 (Example)	Commerce orders	% versus budget	0%	-3.6%	[@commerce-director]	Commerce revenue, cost of sales
DRV-005 (Example)	Commerce average order value	EUR	45.00	45.00	[@commerce-director]	Commerce revenue
DRV-006 (Example)	Billable services days	days	40,000	42,500	[@services-director]	Services revenue and cost
DRV-007 (Example)	Net hires	FTE	55	68	[@chro]	People cost
DRV-008 (Example)	Marketing programme spend	EUR thousand	11,250	12,450	[@marketing-director]	S&M expense
DRV-009 (Example)	EUR/USD rate outlook	USD per EUR	1.10	1.12	[@treasurer]	Translation of US revenue and costs, USD cash, covenant tests
DRV-010 (Example)	Capex phasing	EUR thousand	9,400	9,700	[@cfo]	Cash, depreciation

Notes on the example: DRV-009 moves EUR/USD from 1.10 to 1.12 on US subscription revenue of USD 18,480k for Q2 to Q4 (the budget of USD 24,640k less the Q1 USD 6,160k): 18,480 / 1.10 - 18,480 / 1.12 = 16,800 - 16,500 = 300. CC-US sells subscriptions only and US commerce orders are invoiced in EUR by CC-DE, so commerce carries no translation effect. About one sixth of US revenue is matched by costs paid in dollars (Annual Budget ASM-014), so the EBITDA effect is 300 x 5 / 6 = 250. DRV-010: Q1 capex was 1,900 against a budget of 2,600, so Q2 to Q4 spend of 9,700 gives a full-year 11,600, 400 below budget. DRV-006 adds 2,500 days at the budget day rate of EUR 200, or EUR 500k of revenue. DRV-004 removes 3.6% of the Q2 to Q4 commerce budget of 41,600, or about 1,500. Driver definitions and the model logic that links them to the statements belong in Financial Model Specification (finance/financial-model-specification).

3. Changes since the last forecast

3.1 Revenue bridge, FY2027

Step	EUR thousand	Driver	Owner
Budget FY2027 (Example)	160,000
Q1 actual versus budget	(800)	Actuals, see `finance/variance-analysis`	[@head-of-fpa]
Commerce volume, Q2 to Q4	(1,500)	DRV-004	[@commerce-director]
Subscription price uplift	900	DRV-002	[@cro]
FX translation on US subscriptions	(300)	DRV-009	[@treasurer]
Services days	500	DRV-006	[@services-director]
Forecast FY2027	158,800

Check: 160,000 - 800 - 1,500 + 900 - 300 + 500 = 158,800. Forecast at budget rates (constant currency): 158,800 + 300 (DRV-009) + 100 (Q1 translation, in the Q1 actual) = 159,200.

3.2 EBITDA bridge, FY2027

Step	EUR thousand	Explanation
Budget FY2027 (Example)	21,600
Q1 actual versus budget	(150)	Commerce shortfall partly offset by opex savings
Commerce volume	(450)	1,500 of revenue at 30% gross margin
Subscription price uplift	900	Price falls straight to EBITDA
FX translation	(250)	300 of revenue less 50 of US costs
Services days	150	500 of revenue at 30% gross margin
R&D hiring re-phased	300	Ten Vietnam hires start in October instead of April: 10 x EUR 48k x 6 / 12 = 240, plus 60 of recruiter fees avoided
Marketing reinvestment	(1,200)	Second US campaign, decision D-001
Forecast FY2027	20,900

Check: 21,600 - 150 - 450 + 900 - 250 + 150 + 300 - 1,200 = 20,900.

4. Revenue and margin outlook

EUR thousand	Q1 2027 A	Q2 2027 F	Q3 2027 F	Q4 2027 F	FY2027 F	FY2027 B	Var	Q1 2028 F	Q2 2028 F	Q3 2028 F
Revenue (Example)	36,200	38,400	40,000	44,200	158,800	160,000	(1,200)	41,500	43,000	44,500
EBITDA	3,950	4,700	5,150	7,100	20,900	21,600	(700)	5,000	5,800	6,300
EBITDA margin %	10.9%	12.2%	12.9%	16.1%	13.2%	13.5%	-0.3 pt	12.0%	13.5%	14.2%

A = actual, F = forecast, B = budget. Quarterly budget for comparison: revenue 37,000, 39,000, 40,000, 44,000; EBITDA 4,100, 5,000, 5,300, 7,200.

Segment, FY2027	Budget	Forecast	Variance	Gross margin % forecast
Subscription (Example)	96,000	96,750	750	80.1%
Commerce	54,000	51,400	(2,600)	29.8%
Professional services	10,000	10,650	650	30.2%
Total	160,000	158,800	(1,200)

Subscription: 96,000 + 150 (Q1) + 900 - 300 = 96,750. Commerce: 54,000 - 1,100 (Q1) - 1,500 = 51,400. Services: 10,000 + 150 (Q1) + 500 = 10,650. Gross profit forecast: subscription 76,800 + 70 (Q1) + 900 - 250 (FX net of US costs) = 77,520; commerce 3,290 (Q1 actual) + 12,480 - 450 = 15,320; services 3,000 + 70 (Q1) + 150 = 3,220; total 96,060. Commerce margin is lower than budget because Q1 delivered 29.1% on a weaker mix.

Report revenue also at constant currency (budget rates) and explain the translation effect separately; segment profitability and pricing analysis sit in Unit Economics Analysis (finance/unit-economics-analysis).

5. Cost and headcount outlook

Function, FY2027	Budget	Forecast	Variance	Comment
Sales and marketing (Example)	36,000	36,850	(850)	Q1 underspend 350; reinvestment 1,200 (D-001)
Research and development	24,000	23,950	50	Q1 overspend 250; hiring re-phase saves 300 (240 people cost, 60 recruiter fees)
General and administrative	14,400	14,360	40	Q1 saving 40
Operating expense before D&A	74,400	75,160	(760)

Check: forecast gross profit equals EBITDA plus operating expense: 20,900 + 75,160 = 96,060, against budget gross profit of 96,000. The commerce shortfall of 450 (Q2 to Q4) and Q1 gross profit shortfall of 290 are offset by price (900), services (150) and FX net (-250): 96,000 - 290 - 450 + 900 + 150 - 250 = 96,060.

Headcount (FTE)	Opening 2027-01-01	Q1 actual close	Year end budget	Year end forecast	Variance
Delivery and customer operations (Example)	160	166	180	182	2
Sales and marketing	200	207	220	220	0
Research and development	230	223	260	260	0
General and administrative	95	96	100	98	(2)
Total	685	692	760	760	0

The year-end forecast of 760 is 68 above the Q1 close of 692, matching DRV-007; the budget needed 55 net hires from its Q1 close of 705. Research and development reaches its budgeted 260 at year end because the ten deferred Vietnam hires start in October; the saving is in the months they are not on payroll, not in year-end headcount.

6. Cash and liquidity outlook

EUR thousand	FY2027 budget	FY2027 forecast	Variance	Explanation
EBITDA (Example)	21,600	20,900	(700)	Section 3.2
Working capital movement	(2,100)	(2,700)	(600)	Commerce inventory built for the weaker volume
Tax paid	(3,300)	(3,300)	0	Payments on account fixed for the year
Capital expenditure	(12,000)	(11,600)	400	Part of CAP-002 moves to Q1 2028
Interest, leases, dividend	(5,800)	(5,800)	0
FX on USD cash balances	0	(200)	(200)	EUR/USD at 1.12 versus 1.10
Net change in cash	(1,600)	(2,700)	(1,100)
Opening cash	28,000	28,000	0
Closing cash	26,400	25,300	(1,100)
Less trapped cash (Vietnam deposit account)	(1,000)	(1,000)	0	Not freely transferable
Undrawn revolving facility	20,000	20,000	0	Counts only if more than 12 months to maturity
Liquidity headroom	45,400	44,300	(1,100)	Treasury minimum EUR 35,000k

Check: 20,900 - 2,700 - 3,300 - 11,600 - 5,800 - 200 = -2,700; 28,000 - 2,700 = 25,300; 25,300 - 1,000 + 20,000 = 44,300. Interest in the financing line is net interest (term loan 1,200 plus lease interest 360 less interest received 360), as in Annual Budget (finance/annual-budget) section 9.2.

The revolving facility and the term loan both mature on 2028-06-30. The year-end headroom above assumes the refinancing targeted for 2027-06-30 (Annual Budget ASM-013); without it, the facility would have six months to run at 2027-12-31, would drop out of the liquidity definition and headroom would fall to 24,300, below the treasury minimum. Track the refinancing as a forecast milestone each cycle.

\text{Management leverage} = \frac{30{,}000 - 25{,}300}{20{,}900} = \frac{4{,}700}{20{,}900} = 0.22\text{x} \qquad \text{Covenant leverage} = \frac{30{,}000 + 6{,}400 - (25{,}300 - 1{,}000)}{20{,}900} = \frac{12{,}100}{20{,}900} = 0.58\text{x}

Management leverage excludes leases; covenant leverage follows the facility definition (lease liabilities of 6,400 at year end included, trapped cash excluded) against a maximum of 3.00x.

The weekly cash view for the next 13 weeks is maintained in 13-Week Cash Flow Forecast (finance/13-week-cash-flow-forecast); it must reconcile to the first quarter of this forecast within [tolerance].

7. Scenarios

Scenario	Driver changes	Revenue FY2027	EBITDA FY2027	Closing cash	Management net debt / EBITDA	Covenant leverage
Base (Example)	As section 2	158,800	20,900	25,300	0.22x	0.58x
Upside	Commerce orders back to budget from Q3; new ARR +10%	162,000	22,300	26,600	0.15x	0.48x
Downside	Commerce orders -10% for Q2 to Q4; US churn +2 pt; second campaign kept	153,000	18,500	22,400	0.41x	0.81x

Ratios: (30,000 - 26,600) / 22,300 = 0.15x; (30,000 - 22,400) / 18,500 = 0.41x. Covenant leverage adds lease liabilities of 6,400 and excludes trapped cash of 1,000: (36,400 - 25,600) / 22,300 = 0.48x; (36,400 - 21,400) / 18,500 = 0.81x. Even the downside leaves wide headroom to the 3.00x maximum; the binding constraint in every scenario is the refinancing date, not the covenant.

Trigger ID	Leading indicator	Threshold	Pre-agreed response	Owner
TRG-001 (Example)	Commerce orders, trailing four weeks versus forecast	Below -8% for two consecutive months	Pause second campaign tranche; cut inventory buy	[@commerce-director]
TRG-002 (Example)	US logo churn, trailing quarter	Above 2%	Customer success review; delay US sales hires	[@cro]
TRG-003 (Example)	Liquidity headroom	Below EUR 38,000k forecast	Treasury escalation under the treasury policy	[@treasurer]

For funding stress tests and committed facility planning use Liquidity and Funding Plan (finance/liquidity-funding-plan).

8. Forecast accuracy

Use the following measures, where F is the forecast, A the actual and n the number of periods.

\text{MAPE} = \frac{100\%}{n}\sum_{t=1}^{n}\left|\frac{F_t - A_t}{A_t}\right| \qquad \text{WAPE} = 100\% \times \frac{\sum_t |F_t - A_t|}{\sum_t A_t} \qquad \text{Bias} = 100\% \times \frac{\sum_t (F_t - A_t)}{\sum_t A_t}

MAPE shows the typical size of error; WAPE weights larger periods more; a positive bias means the forecast is systematically too high (optimism), a negative bias means it is too low (sandbagging).

Month	Forecast at lag 3	Actual	Error F - A	Absolute % error
Oct 2026 (Example)	11,800	12,050	(250)	2.07%
Nov 2026	12,400	12,150	250	2.06%
Dec 2026	12,900	12,700	200	1.57%
Jan 2027	12,100	11,850	250	2.11%
Feb 2027	12,150	11,900	250	2.10%
Mar 2027	13,300	12,450	850	6.83%
Total	74,650	73,100	1,550

Results for revenue: MAPE = 16.74% / 6 = 2.79%; WAPE = 2,050 / 73,100 = 2.80%; bias = 1,550 / 73,100 = +2.12%, so the forecast has been too high in five of six months. Q1 actuals (11,850 + 11,900 + 12,450 = 36,200) match the variance analysis.

Measure	Target	Latest	Status	Action
Revenue MAPE, lag 3 (Example)	Below 3.0%	2.79%	Met	None
Revenue bias, lag 3	Between -1.5% and +1.5%	+2.12%	Not met	Commerce driver review; owner to show order pipeline evidence
EBITDA error, full year at Q2	Within 5%	[Next cycle]	[Status]	[Action]
Closing cash error, quarter	Within EUR 1,500k	[Next cycle]	[Status]	[Action]

9. Risks and opportunities

ID	Description	Type	Probability	EBITDA effect FY2027	Cash effect	Owner	Review date
RO-101 (Example)	Large US customer delays roll-out	Risk	Medium	(900)	(900)	[@cro]	YYYY-MM-DD
RO-102 (Example)	Supplier rebate renegotiation	Opportunity	Medium	600	600	[@commerce-director]	YYYY-MM-DD
RO-103 (Example)	Excess commerce inventory write-down	Risk	Low	(300)	0	[@financial-controller]	YYYY-MM-DD
RO-104 (Example)	Services day-rate increase for new contracts	Opportunity	Medium	200	200	[@services-director]	YYYY-MM-DD

Probability-weighted net effect (medium = 50%, low = 20%): -450 + 300 - 60 + 100 = -110. Use the weighted figure for discussion only; do not add it to the base forecast. Enterprise-level risks go to Risk Register (compliance/risk-register).

10. Decisions requested

Decision ID	Decision	Options considered	Recommendation	In forecast?	Owner	Date
D-001 (Example)	Release second US marketing campaign	Release now; release in Q3; cancel	Release now (EUR 1,200k) to protect FY2028 ARR	Yes	[@ceo]	YYYY-MM-DD
D-002 (Example)	Start ten Vietnam R&D hires in October instead of April	Keep plan; re-phase	Re-phase (saves EUR 300k: people 240, recruiter fees 60)	Yes	[@cto]	YYYY-MM-DD
D-003 (Example)	Defer part of warehouse automation (CAP-002)	Full spend; defer EUR 400k to Q1 2028	Defer, subject to capex request	Yes	[@cfo]	YYYY-MM-DD
D-004 (Example)	Commerce inventory buy for Q4	Budget volume; forecast volume	Buy to forecast volume	No, saves working capital if taken	[@commerce-director]	YYYY-MM-DD

11. Process, controls and model governance

Single model. The budget and the forecast use the same model and chart of accounts; structural changes are version-controlled under Financial Model Specification (finance/financial-model-specification).
Input log. Every driver change carries the owner, date and evidence (pipeline report, order data, hiring tracker). Manual overrides are logged as OVR-NNN with reason and approver.
Reconciliations. Opening balances agree to the closed ledger; Q1 actuals agree to the management accounts; the first-quarter cash agrees to the 13-week forecast.
Review. FP&A challenge session with each driver owner; CFO review of the bridges; executive committee for decisions.
Separation from incentives. Bonus targets stay on the budget, so owners have no reason to bias the forecast.
Retention. Keep each locked cycle for at least [three years] for accuracy analysis.

Override ID	Line	Change	Reason	Approved by
OVR-007 (Example)	Commerce Q2 revenue	(200)	April promotion cancelled after the driver lock	[@head-of-fpa]

11.1 How to complete and review a cycle

Start each cycle from the locked actuals, not from the previous forecast file. Roll the model forward so that the first forecast month is the month after the cut-off, load the closed ledger into the actual columns and confirm that opening cash, debt and working capital balances agree to the balance sheet. Only then send the driver templates to owners. A cycle that starts from last quarter's file tends to carry stale assumptions forward without anyone deciding to keep them.

Ask each driver owner for three things: the new value, the evidence behind it and the reason it changed since the last cycle. Evidence means data that someone outside the team could check: the order pipeline, signed contracts, the hiring tracker with offer dates, supplier confirmations or the treasury rate view. "Management judgement" is acceptable only when it is labelled as such and the owner signs it. Record every driver change in the input log so that the bridge in section 3 can be rebuilt from the log alone.

When the model has run, review the outputs in this order. First, check the bridges: every bar must have a driver ID and an owner, and the sum must equal the change in the result. Second, compare the forecast with the run-rate of the last three actual months; a forecast that jumps above the run-rate needs a named reason (a price increase, a campaign, a contract start). Third, check the cash forecast against the profit forecast: working capital, tax, capex and financing lines must follow from the drivers, not from the budget. Fourth, test liquidity and covenants on the facility definitions, including lease liabilities and the treatment of trapped cash, and check the dates on which committed facilities stop counting towards headroom.

Common errors to look for:

Translation and performance mixed in one bar. Keep FX in its own bar at a stated rate, and give the constant-currency result alongside the reported one.
Headcount savings that are really permanent cuts. If hires are delayed, show the months saved and the year-end position separately, as in the R&D example, so that the next budget starts from the right run-rate.
Risks hidden in the base case. The base forecast is a best estimate; contingencies and stretch items belong in section 9 with a probability, never inside a driver.
A horizon that stops at a facility maturity or a refinancing date without saying what is assumed after it. State the funding assumption explicitly and test the case where it slips.
Accuracy measured only on revenue. Track EBITDA and closing cash errors as well, because those are the numbers lenders and the board act on.

Close the cycle by archiving the locked file, the input log and the bridges together under the cycle ID, and by updating the accuracy table in section 8 once the next month closes.

12. Glossary and references

12.1 Glossary

Term	Meaning in this template
Rolling forecast	Forecast that keeps a constant horizon by adding a period each cycle
Driver	Operational variable that determines financial results, such as orders or hires
Lag	Number of periods between the forecast date and the period forecast
MAPE	Mean absolute percentage error, the average size of forecast error
WAPE	Weighted absolute percentage error, total absolute error over total actual
Bias	Signed total error over total actual; positive means over-forecasting
Bridge	Reconciliation of one result to another by named causes
Constant currency	Results translated at budget FX rates
Liquidity headroom	Unrestricted cash (excluding trapped cash) plus undrawn committed facilities with more than 12 months to maturity
Covenant leverage	Borrowings plus lease liabilities less freely available cash, divided by EBITDA, on the facility definition

12.2 References

Reference	Version or date checked	Use
Beyond Budgeting principles, Beyond Budgeting Round Table (bbrt.org)	Principles document dated 2022, checked 2026-10-25	Context: the principle names Targets, Forecasts, Resource allocation, Performance evaluation, Rewards and Coordination; names only
FAST Standard, FAST Standard Organisation	Version 02c, July 2019 (CC BY 4.0), checked 2026-10-25	Modelling discipline for the forecast model, by reference
[Group FP&A manual]	[Version]	Calendar, materiality and override rules

Standard and licence note

Structure aligned with common financial planning and analysis practice for driver-based rolling forecasts; no standard prescribes the format. The Beyond Budgeting principles and the FAST Standard (version 02c, CC BY 4.0) are cited by name only; no text from either is reproduced. Accuracy formulas are general statistical measures. Corvid Commerce Group and every figure are fictional. This template is original text written for this collection. Check the current versions of cited sources before use. Template licence: MIT.