← All use cases

Use case #28 · CPA

FP&A CPA tests Claude on a revenue forecast case and compares it with his own model

Jea Hun Shin, CPA (LinkedIn) ·

The short version

What they did

Shin gave Claude a revenue forecasting case (analyze historical data, model customer growth, utilization by pricing tier and revenue per customer, and tie the drivers together) with the case’s simple instructions, the way he would test an interview candidate. He then compared its model with his own forecast. It was a practice case, not client data.

Result (as claimed)

“Claude was able to review the source data in the file and come up with a model in 15 minutes. The first draft had some errors, but after reviewing and pinpointing the problem area, I was able to prompt it once more to allow Claude to fix it.” Compared with his own: “The results were meaningfully different, not because of a difference in calculation, but because of a difference in assumptions around one of the core drivers.”

Prompt

I’m an accountant testing a revenue forecast against a model I’ve already built. I’ll attach [historical monthly data] and [the practice case instructions]. The data includes customer counts, utilization by pricing tier and revenue per customer. Forecast period: [next 12 months or the period in the case].

1. Review the source data and explain how the drivers combine to produce revenue.
2. Build a monthly spreadsheet forecast with formulas linking those drivers to revenue.
3. Put each assumption on a separate line with your reasoning, and explain which assumption most affects the forecast.

Add a short plain-English model guide so I can compare it with my own work. Separate facts from assumptions. Ask about missing units or definitions rather than guessing, and flag data gaps. Don’t present a forecast as a promised result.

Do it yourself

These steps are my suggested way in, not Jea Hun Shin’s setup.

What you need

  • A chat assistant that builds spreadsheets, such as Claude.
  • A practice case or past data where you already built a forecast.
  • Your own forecast to compare against.

Start simple

  1. Pick a case you already solved, as Shin did.
  2. Paste the Prompt and attach the data.
  3. Read its assumptions first. Mark any you disagree with.
  4. Compare its forecast with yours and find where they differ.

Take it further

  • Ask it to fix any errors you find, then re-check the fixed version.
  • Ask for a sensitivity table on the driver it says matters most.
  • Use it on a client forecast once it handles your test case well.

Check your work

  • Check that each driver ties to revenue with a formula you can follow.
  • Focus on assumptions. Shin’s results differed because of one core driver, not the math.
  • A forecast sent to a client is your work. Review it as you would a staff member’s.

Before you rely on it

  • This is general information about how accountants use AI, not professional guidance for your engagements.
  • You stay responsible for the work under your professional standards, including review and sign-off. An AI tool’s output is a draft, not evidence.
  • Before you send client data to an AI tool, read the tool’s data and training terms, and get any client consent you need. For tax return information, that includes IRC section 7216.

Source checked. The link, author, date and quote were confirmed against the source. The result itself has not been reproduced. Prompt adapted for this library, inspired by the case. Business size: Not stated. Results are quoted as the source claimed them and never improved. Spot an error or a broken link? Email charlie@cbarmorecpa.com.

More in Advisory/CAS

New cases go out every Saturday in Inside AI.