How top FP&A teams use Claude: five practices that make the difference
The FP&A teams getting real output from Claude do five things: load the planning context, write the variance process down, draft the heavy Excel work, connect current files, and pressure-test the forecast.
Most FP&A teams have tried AI on a forecast or a variance report and gotten something generic back.
The teams getting real output are not using a better model. They set the work up differently.
Here are the five practices that make the biggest difference.
1. They give Claude a project with the planning context
The budget, driver assumptions, KPI definitions, chart of accounts mapping, and the last couple of board decks all sit in one Claude project.
A question like "why did gross margin drop in March?" then gets answered using the company's own definitions, not a textbook's.
This is the step most teams skip, and it is the reason the first attempt reads like it was written by someone who has never seen the business.
2. They turn the variance process into a skill
The skill spells out:
- What gets flagged. For example, anything over 10% and $25k.
- How each variance is explained.
- In what order: what happened, why, and what it changes in the forecast.
Commentary comes back in the same format every month. Analysts spend their time checking the reasons instead of writing the first draft.
3. They have Claude draft the heavy Excel work
Claude in Excel traces why EBITDA moved between two forecast versions, finds hardcoded numbers, and builds price and volume sensitivity tables.
The formulas stay in the workbook, so anyone can follow the math. Nothing is a black box.
4. They connect the files they already use
Claude reads the Drive or SharePoint folders that hold the ERP exports and department budgets, plus the CRM for pipeline.
Each forecast refresh starts from current files, not the copy someone saved three weeks ago.
5. They ask Claude to pressure-test the forecast
A prompt that comes up a lot before board meetings:
You're our lead investor. Which three assumptions in this forecast would you question first, and what would you want to see?
It surfaces the soft spots early. The team decides what to change.
What stays with the team
Claude takes on more of the drafting and checking. The FP&A team gets more time with the business.
The assumptions are still the team's. So is the sign-off. A person reviews every output before it goes to leadership, a lender, or the board.
Context, a written process, drafted Excel work, current files, and a forecast that gets questioned before the board meeting. None of it needs a new platform. It needs the way your team already works, written down.
That's what AI-native finance looks like.
If you want help setting this up for your FP&A team, the next step is a short introduction call.
Stavros Christias runs Vantage Rock Financial, a fractional CFO firm working with founder-led services, healthcare and multi-entity businesses. Ten-plus years across FP&A, controllership, reporting, forecasting and systems implementation, including PE-backed operators. You talk to the operator, not a sales team. LinkedIn.
15–30 minute Introduction Call.
It is a fit-check, not a sales call. We don't diagnose on the call and you don't leave with a plan. Thirty minutes tells us both whether there's work here worth doing.