AI costing prompts · By role
AI costing prompts for the CFO
You are asked to connect cost detail to strategy, the board and the capital plan, usually with less time than the question deserves. AI can give you a faster first draft of a profitability argument or a board narrative, but it cannot be the source of the numbers you defend in the room. These prompts let AI do the drafting while you keep the deciding.
In short
Use AI as a senior drafting partner, not a calculator of record. The prompts below help you frame which customers and products truly earn their return, prepare a board-ready profitability story, and stress-test a pricing move before you commit to it. Each forces the model to work only from figures you supply, to separate fact from assumption, and to surface what is missing, so the output is something you can interrogate rather than something you have to trust blind.
What a CFO should and should not ask AI to do
AI earns its place in the parts of your week that are about structure and articulation. It can turn a customer profitability table into a clear board narrative, frame the questions a capital decision should answer, draft scenario logic before you put real numbers in, and pressure-test a pricing rationale by playing devil's advocate. Used this way it compresses the time between having the analysis and being able to tell the story, which is often where the bottleneck sits.
The risk is using it as an oracle. Asking AI to "estimate our margin," "assume a market rate," or "tell me which segment is most profitable" without giving it your data invites a confident answer with no foundation, and at board level a wrong number is expensive in more than money. Keep three things firmly yours: the underlying data, the strategic judgement about what the numbers mean, and the final decision. AI drafts; you decide, and you sign.
Three prompts to start with
1. Read the real profitability of customers and products
Use this to turn a profitability table into a clear view of where return actually comes from. It builds on the customer profitability page.
You are a finance analyst helping a CFO interpret customer and product profitability. Work only from the data I give you. Do not invent any numbers or assume market figures. List every assumption, and label anything missing as DATA MISSING with what you need from me. My data: - Customers or segments with revenue, gross margin and cost-to-serve: [paste] - Products or lines with revenue and contribution: [paste] Steps: 1. Rank customers (or products) by net profit after cost-to-serve; show the calculation as a formula before the result. 2. Identify any that are revenue-large but profit-small or loss-making, citing the source row for each. 3. Separate clearly: what you calculated from my data, what you assumed, and what is your suggestion. 4. Pose the three strategic questions these results raise, without recommending an action I have not asked for. 5. Check that segment profits sum to the total I gave you and flag any discrepancy.
2. Prepare a board-ready profitability story
Turns analysis into a narrative the board can act on, with nothing overstated. See board reporting.
You are helping a CFO prepare a profitability section for a board pack. Work only from the figures I give you. Do not invent numbers or causes. If a movement needs an explanation I have not provided, ask me rather than guess. My data: - Profitability by segment, this period vs prior and vs plan: [paste] - Confirmed drivers of any movement: [paste] - The one decision I want the board to engage with: [state it] Steps: 1. Open with the single most important fact from my data, stated plainly. 2. Show the two or three movements that matter, each with the variance as a formula. 3. Attribute causes only where I have confirmed them; otherwise write "cause to confirm". 4. Frame the decision and the trade-off, without inventing supporting figures. 5. Keep every number consistent with my data and flag anything that does not reconcile.
3. Stress-test a pricing move before you commit
Run the downside before the board does. Pairs with pricing decisions.
You are acting as a critical reviewer helping a CFO pressure-test a proposed pricing change. Work only from the data I give you. Do not invent elasticities, volumes or competitor prices. Where a figure is needed and missing, label it DATA MISSING. My data: - Current price, volume, unit cost and contribution for the line: [paste] - The proposed price change: [state it] - Any volume response I can evidence: [paste, or write "none confirmed"] Steps: 1. Show the contribution impact at current volume as a formula, then the value. 2. Compute the break-even volume change that keeps total contribution flat; show the formula. 3. List the assumptions the case depends on and mark which are evidenced and which are not. 4. Argue the strongest case against the change using only my data. 5. State what additional data would make this decision safe.
The one rule
Work only from the data I give you. Do not invent any numbers, rates or volumes. Label anything missing as DATA MISSING and tell me what you need.
At board level the cost of a plausible fabricated number is high. For the full set of safeguards, read how to stop AI inventing your numbers.
A defensible story needs a defensible model
A good prompt sharpens how you tell the profitability story. It does not give you the reconciled, multidimensional model the story should rest on. That is what we build with finance leaders: customer and product profitability you can take to a board, a buyer or a lender and have it hold up under scrutiny. If the decisions are getting bigger, make sure the numbers underneath them are sound.