Practical guide
AI fluency for FP&A analysts: review a forecast revision
Work through a fictional forecast update with driver assumptions, a variance bridge and an inspectable recommendation.
An FP&A analyst can demonstrate AI-enabled work by tracing a forecast revision from source inputs through assumptions to a decision-ready explanation. This original exercise focuses on that specific workflow rather than the whole finance function. All figures and business conditions are fictional, and the activity is educational practice, not financial advice or a prescribed accounting treatment.
Define the revision being requested
Fix the period, unit and decision before changing the forecast.
The fictional manager asks for an updated quarterly revenue outlook after a delayed customer launch. The pack includes an earlier forecast, a pipeline note and a revised launch email. The analyst must identify whether the request concerns the whole quarter, the remaining months or a particular customer segment. Those are not interchangeable revisions.
Write a short scope note with the decision owner and expected output. State which supplied figures are actuals and which are assumptions. Do not let AI fill a missing reporting period from context. A clear scope prevents a technically correct calculation from answering a different question.
Build a driver table before drafting the story
Connect each changed assumption to a source and a calculation.
For the fictional exercise, expected units fall from one hundred to eighty while an illustrative unit value remains ten. The simple forecast component therefore moves from one thousand to eight hundred, a reduction of two hundred. These numbers are a toy example for tracing a driver, not measured company data.
Keep the delayed-launch assumption separate from the arithmetic. The email may support a timing change without proving the final number of units. Ask AI to organize the driver table, then verify that it has not converted a planning assumption into a confirmed customer commitment.
Explain the movement without double counting
A variance bridge should reconcile the change and preserve the mechanism behind it.
Compare the previous component with the revised component and show how the stated driver accounts for the difference. If another supplied assumption changes as well, isolate its effect using a clearly described calculation order. Do not let a narrative mention both timing and volume as separate losses when they refer to the same delayed units.
Check that the bridge reconciles to the revised total and uses consistent units. A fluent explanation cannot repair a mismatched denominator or duplicated adjustment. Keep the calculation available so the reviewer can reproduce the movement without relying on the assistant's summary.
Test the forecast under a different launch assumption
Use a sensitivity case to expose dependence on uncertain timing.
Ask what changes if the delayed launch moves outside the quarter rather than merely later within it. Mark this as a hypothetical case. Identify the source confirmation needed to choose between the base assumption and the alternative. Avoid presenting an unsupported probability as if it came from the pipeline.
The analyst should explain which parts of the forecast remain unchanged and which depend on the launch. This keeps the discussion focused on consequential uncertainty. More decimal places do not resolve uncertainty about whether an event will occur inside the reporting window.
Deliver a reviewable forecast note
The final note should connect the revision, its cause and the decision it supports.
Include the original and revised figures, the driver bridge, source references, unresolved assumptions and the next confirmation owner. State whether the forecast is ready for review or remains conditional. Do not imply approval merely because the worksheet and narrative agree with each other.
A reviewer can inspect scope discipline, arithmetic, assumption handling and response to the changed launch case. Use any gap to select a focused follow-up exercise. Keep conclusions limited to this forecast task; the broader finance guide covers other work such as reconciliation and assurance support, which require different evidence.
Sources and scope
NIST addresses AI risk management. Skills England describes workplace AI foundations.
These sources provide background, not endorsement of this exercise. The worked example and suggested review method are original illustrative guidance. They are not customer results, validated benchmarks or evidence of a particular product capability. Adapt the exercise to the task and use qualified review where consequences require it.
Sources: [1] [2]
Sources
- 1.AI RMF Core · NIST
- 2.AI foundation skills for work benchmark · Skills England