FP&A explained: what financial planning and analysis does, and how AI is changing it

Quick answer: FP&A (financial planning and analysis) is the finance discipline that turns the organisation's strategy into budgets, forecasts and scenarios, and then explains the gap between plan and actual so leaders can act. A modern FP&A team runs driver-based rolling forecasts on a dedicated planning platform, partners directly with the business, and uses AI to automate variance commentary and baseline forecasts rather than to replace judgement.

Ask ten finance leaders what FP&A means in their organisation and you will get ten answers, from "the budget team" to "the commercial brain of the company". The gap between those answers is the gap between an organisation that forecasts once a year in spreadsheets and one that re-plans monthly against live drivers and uses the forecast to steer.

This guide is for CFOs, heads of FP&A, finance business partners and finance systems leads in large organisations across Australia, New Zealand and the wider region. It defines FP&A properly, sets out what a high-performing function actually does, explains rolling forecasts, driver-based planning and scenario planning in operating terms, and then takes a hard look at where generative and agentic AI in finance are delivering value in 2026 and where they are not.

What is FP&A, and what does it stand for?

FP&A stands for financial planning and analysis. It is the forward-looking half of finance. Where accounting records what happened and controls whether it was recorded correctly, FP&A estimates what will happen, tests what could happen, and explains why the two diverged.

The core cycle has four parts:

• Planning — the strategic plan (three to five years) and the annual operating plan or budget, translated into revenue, cost, headcount, capital and cash.
• Forecasting — periodic re-estimates of the current and next year, increasingly on a rolling basis.
• Analysis — variance analysis against plan and prior periods, margin and profitability analysis, unit economics, and the commentary that goes to the executive and board.
• Decision support — business cases, pricing and investment appraisal, scenario modelling, and partnering with divisional leaders on the trade-offs behind the numbers.

FP&A is not management reporting alone, and it is not the budget process alone. An FP&A analyst who only consolidates spreadsheets is doing the mechanical part of the job; an FP&A manager who can walk a general manager through the three drivers that explain this quarter's margin is doing the whole job.

What does a modern FP&A function actually do?

The difference between a traditional and a modern FP&A function is less about tools than about cadence, ownership and what the output is for.

• Planning cadence — Traditional FP&A: Annual budget, two or three reforecasts · Modern FP&A (2026-2027): Rolling forecast refreshed monthly or quarterly, 12-18 months out
• Model structure — Traditional FP&A: Line-item budgets by cost centre · Modern FP&A (2026-2027): Driver-based: volumes, rates, headcount, conversion, utilisation
• Where the work happens — Traditional FP&A: Spreadsheets emailed around · Modern FP&A (2026-2027): Planning platform with one model, workflow and version control
• Primary output — Traditional FP&A: The budget pack and monthly variance report · Modern FP&A (2026-2027): Scenarios, decision briefs and a reconciled forecast the business owns
• Business partnering — Traditional FP&A: Finance explains variances after the fact · Modern FP&A (2026-2027): Partners sit in divisional planning and own the forecast with the GM
• Analyst time — Traditional FP&A: 70-80% on gathering and reconciling data · Modern FP&A (2026-2027): Majority on analysis, scenarios and partnering
• Use of AI — Traditional FP&A: None, or ad hoc · Modern FP&A (2026-2027): Automated commentary drafts, anomaly detection, statistical baseline forecasts

Two practices separate strong functions from the rest.

Driver-based planning. Instead of asking every cost centre to budget 400 lines, the model is built on the handful of operational drivers that actually move the result — enrolments, bed-days, active customers, billable hours, tonnes shipped, headcount by band — and the financial lines are calculated from them. The business argues about drivers it understands rather than cost codes it does not, and the forecast can be re-run in hours when a driver changes.

Rolling forecasts. A rolling forecast always looks the same distance ahead (commonly 12, 15 or 18 months) and is refreshed on a fixed cadence. It replaces the ritual of a budget that is out of date by the second quarter with a forecast that the executive uses to make resource decisions. The annual budget does not disappear — boards and regulators still need an approved plan — but it becomes one snapshot of the rolling view rather than a separate exercise.

Scenario planning sits on top of both. With drivers and a platform, running a downside case for a 10% volume fall or a 200-basis-point rate rise is a model change, not a six-week project.

What is FP&A software, and how do you implement it?

FP&A software — also called financial planning and analysis software, enterprise performance management (EPM) or extended planning and analysis (xP&A) — is a platform that holds the planning model, connects to the ledger and operational systems for actuals and drivers, manages the workflow of contributors and approvals, and produces forecasts, scenarios and reports from a single version of the data. The established products include Anaplan, Workday Adaptive Planning, Oracle EPM Cloud, SAP Analytics Cloud, OneStream, Planful, Board, Pigment and Vena, with ERP vendors increasingly bundling planning into their suites.

A platform does not fix a weak planning process; it makes a strong one scalable. The implementations that work follow a pattern:

1. Redesign the planning process first. Decide the cadence, the drivers, the level of detail and who owns which numbers. Implementing the current spreadsheet logic in a platform reproduces its problems at higher cost.
2. Fix the data foundation. The platform needs clean actuals from the ledger, a stable chart of accounts and reporting hierarchies, and reliable operational data for the drivers. If the ERP is mid-migration, sequence the planning platform after the chart of accounts is final — the ERP implementation and cloud ERP modernisation guide explains why.
3. Start with one use case and expand. Headcount and workforce planning or a P&L rolling forecast is a typical first release; capital, cash and operational planning follow once the model and the team are proven.
4. Build in-house model ownership. The organisations that get value own and maintain the model themselves. A platform that only the implementation partner can change becomes a new black box.
5. Design the outputs for the audience. Executives and divisional leaders need three drivers and a decision, not forty pages of variances. Build the summary views and the decision briefs as deliberately as the model.

How is AI being used in FP&A?

The hype around AI in finance has outrun the evidence, but by 2026 some uses of AI in FP&A are established enough to plan around. They fall into three tiers.

Tier 1 — proven and worth doing now

• Statistical and machine-learning baseline forecasts. Time-series and ML models produce a baseline forecast for revenue, volumes or operating cost lines from historical actuals and external signals. The analyst's job shifts to adjusting the baseline for what the model cannot know (a contract win, a price change, a regulatory shift). Most planning platforms now include this natively.
• Anomaly detection in actuals. Flagging unusual postings, cost spikes or margin shifts before the close is finalised, rather than discovering them in variance analysis two weeks later.
• Generative AI for first-draft commentary. Large language models connected to the planning data produce the first draft of variance explanations and board-pack narrative. Analysts edit rather than write from scratch. The time saving is real, and the quality is acceptable when the model is grounded in the actual numbers and the analyst remains accountable for the final text.

Tier 2 — maturing, pilot with care

• Natural-language querying of the planning model. "Why is Queensland's margin down this quarter?" answered with a drill-through rather than a request to the analyst. Useful, but only as good as the data model and the metadata behind it.
• Scenario generation. Using AI to propose and populate scenarios (supplier failure, demand shock, rate change) from drivers and external data. Valuable for breadth; the finance team still has to judge plausibility.

Tier 3 — agentic AI in finance, early and overpromised

Agentic AI — systems that take multi-step actions such as pulling actuals, re-running the forecast, drafting the commentary and routing it for review — is being demonstrated by every planning and ERP vendor. In practice the organisations getting value in 2026 are using agents for bounded, auditable workflows (reconciliation follow-ups, data collection from contributors, report assembly) with a human approving outputs. Letting an agent change a forecast that drives resourcing decisions without review is not a 2027 practice for any organisation that answers to a board or a regulator.

Three disciplines apply across all tiers:

• Ground every model in governed data. AI on a spreadsheet estate produces confident nonsense. The planning platform and a governed finance data layer are prerequisites.
• Keep accountability with a named person. Commentary, forecasts and board papers remain the responsibility of the FP&A lead who signs them, however they were drafted.
• Document the controls. For regulated entities, APRA's CPS 230 expectations around critical operations and APRA's broader interest in model risk mean AI-assisted forecasting needs the same documentation and review as any other material model. Treasury's voluntary AI safety standard and the Australian Government's guidance on AI assurance are useful reference points for non-regulated organisations.

How does FP&A connect to the rest of finance and operations?

FP&A does not work in isolation, and the strongest functions are built on three connections.

• To the ledger and the close. Forecast accuracy and speed depend on timely, trusted actuals. A five-day close feeds a monthly rolling forecast; a twelve-day close does not. The finance transformation roadmap places close improvement and planning in the same early phase for that reason.
• To operational planning. In organisations with physical supply chains, the demand plan produced by supply chain planning and the financial forecast produced by FP&A are frequently two different numbers. Integrated business planning (IBP) exists to reconcile them, so that the S&OP volume plan and the financial forecast agree. The guide to AI demand planning and S&OP covers how that reconciliation works from the supply chain side.
• To workforce planning. Headcount is the largest controllable cost in most service and public-sector organisations. FP&A and HR planning on a shared driver model (roles, bands, start dates, attrition) removes the perennial mismatch between the people plan and the budget.

What skills does an FP&A team need in 2027?

The FP&A analyst role is changing faster than most finance roles. Hiring and development should target:

• Commercial literacy — understanding the business's drivers well enough to challenge a general manager's forecast.
• Modelling discipline — building and maintaining driver-based models on a platform, with version control and documentation, rather than individual spreadsheet craft.
• Data fluency — comfort with data models, SQL or platform query languages, and the judgement to spot bad data.
• Communication — turning a forecast into a one-page decision brief for a non-finance audience.
• AI literacy — knowing what the forecasting and commentary tools do, where they fail, and how to review their output.

The role that disappears is the consolidator. The role that grows is the partner who owns a division's forecast alongside its leader.

Key takeaways

• FP&A is the forward-looking half of finance: planning, forecasting, analysis and decision support, not just the budget.
• Driver-based rolling forecasts on a planning platform are the operating model that separates strong functions from spreadsheet consolidators.
• FP&A software scales a good process; implement after redesigning the process and stabilising the chart of accounts.
• AI in FP&A delivers now in baseline forecasting, anomaly detection and draft commentary; agentic workflows belong in bounded, human-approved use cases.
• Connect FP&A to the close, to operational S&OP via integrated business planning, and to workforce planning, or the forecast will always disagree with someone else's number.
• Hire and develop for commercial literacy, modelling discipline, data fluency and AI literacy; the consolidator role is disappearing.

Join your peers at the Finance Technology and Transformation Summits 2027

Clutch Events runs free-to-attend, practitioner-led summits for finance and finance technology leaders in large enterprises and government. Upcoming: Brisbane Finance Technology and Transformation Summit 2027 — 14 April 2027 · Sydney Finance Technology and Transformation Summit 2027 — 12 May 2027 · Melbourne Finance Technology and Transformation Summit 2027 — 16 September 2027. See all upcoming events. More guides at the Clutch Events insights hub.