The annual planning process: a blueprint that doesn't kill credibility

Three structural forces have broken the old process:

1. The forecast half-life has collapsed. McKinsey's 2025 Global CFO Survey found that the average "useful life" of a Fortune 500 revenue forecast dropped from 9 months in 2015 to roughly 11 weeks in 2024. For companies with exposure to semiconductors, energy, or consumer discretionary, it's closer to 6 weeks.

2. Capital allocation cycles no longer align with calendar years. When Microsoft committed $80 billion in AI infrastructure capexcapexCapital Expenditure (CapEx) is money spent to acquire, upgrade, or extend long-lived assets like equipment, property, or software that deliver value over multiple years.Voir la définition complète → for FY25 (announced January 2025), Amy Hood didn't run that through a traditional October-December AOP cycle. It moved through a rolling capital committee. Most companies still pretend their CapExCapExCapital Expenditure (CapEx) is money spent to acquire, upgrade, or extend long-lived assets like equipment, property, or software that deliver value over multiple years.Voir la définition complète → slots neatly into a Q4 deck.

3. Regulatory volatility now lives inside the plan. Pillar Two's 15% global minimum tax, fully effective in most EU jurisdictions and the UK in 2026, means a plan built without country-by-country effective tax rate scenarios is structurally wrong. Add CSRD's double materiality reporting on roughly 50,000 EU-active companies, and finance teams are now planning around emissions intensity targets that flex with revenue mix.

The credibility trap

Here's the dynamic that destroys CFOs: the board approves a plan in December. By March earnings, the CFO is already explaining variances. By Q2, the plan is a footnote. By Q3, the CFO is defending why mid-year targets were "sandbagged" or "aggressive." Each cycle erodes the perceived quality of finance leadership.

Julie Sweet, Accenture's CEO, put it bluntly on a 2024 earnings call: "We don't budget anymore. We commit, and we reforecast every 90 days against the commitment." Accenture's planning team, under CFO Angie Park, runs a rolling 18-month model with quarterly resets that flow directly into segment compensation. The variance discussion becomes about operating decisions, not forecasting failure.

A redesigned blueprint: the four-layer plan

The CFOs who have rebuilt their planning process, at Unilever, Accenture, Maersk, ING, and Equinor, converge on a similar architecture. Think of it as four layers, each with a different cadence and a different audience.

Layer 1: the strategic frame (annual, 6 weeks)

This is *not* a budget. It's a 3-year directional view set by the CEO, CFO, and CSO. It contains:

• Revenue ambition by segment (ranges, not points)
• Capital envelope and M&A appetite
• Margin floor and ceiling commitments
• 4-5 explicit strategic bets with capital tags

At Maersk, CFO Patrick Jany runs this exercise in 5 weeks every September. The output is a 12-page document. That's the entire "strategic plan." It defines the *guardrails*, not the line items.

Layer 2: the commitment plan (annual, 4 weeks)

This is what goes to the board and external markets. It's the next 12 months expressed as 3-5 financial commitments: organic revenue growth, operating margin, free cash flowfree cash flowFree Cash Flow is the cash a company generates from operations after funding the capital expenditures needed to maintain and grow its asset base.Voir la définition complète → conversion, ROIC, and net debt/EBITDAEBITDAEBITDA (Earnings Before Interest, Taxes, Depreciation, and Amortization) measures a company's operating profitability before financing and accounting decisions, used to compare core performance across firms.Voir la définition complète →.

Crucially, these are *commitments*, not bottom-up aggregations. The trap most CFOs fall into is building the AOP from 400 cost centers up, then "reconciling" to a top-down target. That reconciliation is where credibility dies, because everyone in the room knows the bottom-up was real and the top-down was political.

Accenture, ASML, and L'Oréal all run top-down commitments with bottom-up *execution plans* that don't need to footoot exactly. The variance is the operating cushion.

Layer 3: the rolling operational forecast (quarterly, 6 quarters out)

This is the workhorse. A 6-quarter rolling forecast, refreshed every quarter, with the first two quarters at line-item detail and quarters 3-6 at driver-based aggregation.

The driver-based piece matters. At Unilever, the rolling forecast is built from roughly 30 drivers (raw material indices, FX pairs, volume by category, promo intensity, A&P as % of sales), not from 4,000 GL accounts. When palm oil moves 18%, the forecast updates in days, not weeks.

Layer 4: the operating cadence (monthly/weekly)

Below the forecast sits the actual decision-making rhythm: monthly business reviews, weekly cash forecasts, and, increasingly, daily sales dashboards in B2C businesses. This layer doesn't get "planned." It gets *managed*.

Putting numbers on it

When ING's CFO Tanate Phutrakul redesigned the bank's planning process in 2023-2024, the measurable outcomes were:

• Planning cycle reduced from 19 weeks to 9 weeks
• FTE effort down 40% (roughly 14,000 hours saved annually)
• Forecast accuracy (rolling 4-quarter MAPE) improved from 11.2% to 6.8%
• Time spent in business reviews on *forward decisions* vs. *backward variance* shifted from 30/70 to 65/35

That last metric is the one to watch. If your monthly business reviews are still 70% variance explanation, your planning process is broken regardless of how elegant the framework looks.

The practical mechanics: what a CFO does monday morning

Frameworks are easy. Implementation is where most redesigns die, usually because the CFO underestimates the political economy of the old budget. The annual budget is a power structure: it allocates resources, fixes accountabilities, and, in most companies, drives bonuses. Changing it threatens everyone.

Here's the sequence that has worked at the companies that have actually pulled it off.

Step 1: decouple compensation from the annual budget

This is the prerequisite. If managers are paid against a December-set number, no amount of "rolling forecasting" will produce honest forecasts. They will sandbag in Q4 of the prior year, ride the easy target, and reforecast becomes performance theater.

Equinor, and before that, Statoil under Bjarte Bogsnes, moved compensation to relative performance metrics: performance vs. peer group, vs. external benchmarks, or vs. ambition-based KPIs that aren't locked at a point in time. At Maersk, roughly 60% of senior management variable comp now ties to ROIC vs. peer set, not budget achievement.

If you can't fully decouple, at minimum widen the achievement band (e.g., target payout at 90-110% of plan) and add forward-looking modifiers (next-year pipelinepipelineAll active sales opportunities across the stages of the sales process, together with their combined potential value and probability of closing.Voir la définition complète →, customer NPSNPSNet Promoter Score (NPS) measures customer loyalty by asking how likely customers are to recommend a brand, then subtracting detractors from promoters.Voir la définition complète →, capexcapexCapital Expenditure (CapEx) is money spent to acquire, upgrade, or extend long-lived assets like equipment, property, or software that deliver value over multiple years.Voir la définition complète → discipline).

Step 2: force driver-based modeling

Most planning models are sophisticated-looking spreadsheets that are, underneath, just last year + X%. A driver-based model expresses every material P&L line as: Volume × Price × Mix × Cost driver.

The rule of thumb: if you can't explain 80% of your revenue forecast with fewer than 25 drivers, you don't have a forecast, you have an extrapolation. The same applies to gross margingross marginGross margin is the share of revenue left after subtracting the direct cost of producing goods or services, expressed as a percentage of revenue.Voir la définition complète → (raw materials, labor cost per unit, capacity utilization) and SG&A (headcount, comp inflation, T&E per FTE).

Step 3: build scenarios into the plan, not around it

The mistake most teams make: they build one base case plan, then "stress test" it with sensitivity analysis. By the time the board sees scenarios, they're presented as edge cases.

The better approach: present the board with three integrated scenarios, base, downside, and disruption, each with its own capital allocation logic. PepsiCo's CFO Jamie Caulfield has publicly described running three full forecast versions through 2025-2026, each with pre-committed action triggers (e.g., "if North America Beverages volume declines >3% for two consecutive quarters, we accelerate the $230M productivity program from FY27 to FY26").

Pre-committing trigger actions is the difference between scenario planning and scenario theater.

Step 4: embed the regulatory stack

In 2026, every plan must explicitly model:

• Pillar Two top-up tax exposure by jurisdiction, with effective tax rate by entity
• CSRD/ESRS reporting commitments that tie emissions intensity to revenue growth (because growing 8% while committing to a 42% Scope 1+2 reduction by 2030 has real capexcapexCapital Expenditure (CapEx) is money spent to acquire, upgrade, or extend long-lived assets like equipment, property, or software that deliver value over multiple years.Voir la définition complète → implications)
• IFRS 16 lease modifications as you reshape real estate footprints, a topic many CFOs still treat as accounting noise rather than a planning input

This isn't compliance theater. The CFOs who model these inside the plan look prepared to boards and analysts. The ones who model them in footnotes look like they're improvising.

Step 5: kill one meeting

This sounds small. It isn't. Every redesign of the planning process has to remove something to be credible. Most commonly, the casualty is the multi-day budget review where 40 business unit leaders present 80-slide decks to the CFO and CEO.

Replace it with: a 90-minute scenario session per business unit, with pre-read driver models. ASML's CFO Roger Dassen famously refuses to look at any planning deck longer than 15 pages.

The CFO's action checklist

If you're sitting on a planning cycle starting in Q3 2026, here's what to put in motion now:

1. Audit your forecast half-life. Pull the last 8 quarters of forecast vs. actual at the segment level. Calculate when forecast error exceeds 5%. If it's under 4 months, your annual plan is structurally fiction, and the data will give you political cover to redesign.

2. Separate the commitment from the budget. Define the 3-5 external commitments (revenue growth, margin, FCFFCFFree Cash Flow is the cash a company generates from operations after funding the capital expenditures needed to maintain and grow its asset base.Voir la définition complète →, ROIC, leverage) and treat the underlying budget as the *execution plan*, not the deliverable. Communicate this distinction to the board explicitly in the next planning cycle preview.

3. Move to a 6-quarter rolling forecast with driver-based mechanics. Start with one business unit as a pilot in Q1