Employee stock options (ESOs) are one of the most complex areas of financial reporting under IFRS 2, Share-based Payment. For CFOs and controllers, the pain is acute: you must estimate fair value at grant date using option pricing models, allocate that cost over the vesting period, adjust for forfeiture estimates, and then track modifications, cancellations, and settlements. Each step introduces assumptions—expected volatility, risk-free rate, expected life, dividend yield—that must be defensible to auditors and consistent with market practice. A single miscalculation in the binomial model or a missed forfeiture adjustment can cascade into a material misstatement.
The friction is compounded by the sheer volume of documentation required. IFRS 2 demands that you disclose the fair value at grant date, the model used, the inputs to that model, and the total expense recognised. Many finance teams still build these calculations in spreadsheets, manually linking grant schedules to vesting tables, then to P&L entries. This process is error-prone, time-consuming, and difficult to audit. When an auditor asks, “How did you arrive at the 35% volatility assumption?” you need a clear, traceable rationale—not a footnote that says “based on historical data.”
ChatGPT changes this. By acting as a structured reasoning assistant, it can help you draft IFRS 2 disclosure notes, validate your option pricing inputs against industry benchmarks, build forfeiture-adjusted expense schedules, and even generate auditor-ready documentation. The key is using it with the right prompt structure—one that forces the model to think step-by-step, reference authoritative guidance, and produce outputs you can verify. Below, I show you exactly how.
Why Most ChatGPT Attempts Fail in Accounting
The typical user opens ChatGPT and types: “Calculate the IFRS 2 expense for my stock options.” The model returns a generic answer that ignores your specific grant terms, vesting schedule, and forfeiture rate. Worse, it may invent numbers or cite outdated guidance. The fix is not to ask for a calculation—it’s to ask for a structured process that you control. The two prompts below are designed for exactly this. They follow a “anatomy of a prompt” template that forces ChatGPT to read your context, ask clarifying questions, and produce an execution plan before generating any output. This prevents hallucination and gives you a workflow you can audit.
First, read these files completely before responding:
[grant_terms.md] — Contains grant date, number of options, exercise price, vesting schedule (graded vesting over 3 years), expected life, risk-free rate, expected volatility, dividend yield, and forfeiture rate assumption.
[ifrs2_disclosure_checklist.md] — Lists every disclosure required under IFRS 2 paragraphs 44-52, including fair value at grant date, model used, inputs, expense recognised, and reconciliation of outstanding options.
Here is a reference for what I want to achieve:
Upload reference file as markdown, or describe reference: I want the disclosure note to match the style and completeness of the example in the “IFRS 2 Illustrative Disclosures” published by Deloitte (2025 edition). That example includes a narrative description of the plan, a table of inputs, a reconciliation of options outstanding, and a breakdown of the total expense by vesting tranche.
Here’s what makes this reference work:
Patterns, tone, structure, rules extracted from reference: The reference uses clear, declarative language. It groups disclosures into three sections: (1) Description of the plan, (2) Fair value inputs and model, (3) Expense and reconciliation. It avoids boilerplate—each number is explained. The tone is factual, not promotional. It uses IFRS 2 paragraph references in parentheses.
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: Full disclosure note, approximately 800-1000 words, suitable for a footnote in interim financial statements.
Recipient’s reaction: Our auditor should be able to verify every number against our grant schedule and valuation report without asking follow-up questions.
Does NOT sound like: A generic template from a textbook. It must reference our specific grant terms and assumptions.
Success means: The note passes our internal review and is accepted by the external auditor without material adjustments.
My context file contains my standards, constraints, audience. Read it fully before starting.
DO NOT start executing yet. Ask clarifying questions first.
Give me your execution plan (5 steps max) before you begin.
This first prompt is designed for the disclosure drafting phase—the moment when you need to translate your grant data into auditor-ready language. Notice what it does: it defines the task and success criteria upfront (“so that I can include it in our interim financial statements”), it forces the model to read your specific files, and it provides a reference document to set the standard. The success brief is critical here—it tells ChatGPT that the audience is the external auditor, not the board or investors. This shifts the tone from promotional to technical. The final instruction—”DO NOT start executing yet”—gives you control. You can review the clarifying questions the model asks and the execution plan it proposes before it writes a single word. This is the opposite of a one-shot prompt; it’s a collaborative workflow.
Building the Forfeiture-Adjusted Expense Schedule
Once the disclosure note is drafted, the next pain point is the expense recognition schedule. IFRS 2 requires you to recognise the fair value of options over the vesting period, adjusted for actual and estimated forfeitures. This is where spreadsheet errors are most common—a missed cell reference in the forfeiture adjustment can throw off the entire P&L line. The second prompt addresses this by asking ChatGPT to build a structured calculation that you can audit cell by cell.
First, read these files completely before responding:
[grant_terms.md] — Same file as above: 100,000 options, exercise price $50, grant date 1 Jan 2026, vesting 1/3 each year over 3 years, fair value at grant $15 per option.
[forfeiture_assumptions.md] — Contains our expected forfeiture rate of 8% per year, actual forfeitures observed in Q1 2026 (2,000 options), and the policy to adjust estimates at each reporting date.
[ifrs2_expense_methodology.md] — Explains graded vesting treatment: each tranche is treated as a separate grant, expense recognised on a straight-line basis over each tranche’s vesting period.
Here is a reference for what I want to achieve:
Upload reference file as markdown, or describe reference: I have a spreadsheet template from our previous year’s audit that shows the correct format: columns for Grant Date, Tranche, Options Granted, Expected Forfeiture Rate, Vesting Start, Vesting End, Fair Value per Option, Total Fair Value, Cumulative Expense Recognised, Current Period Expense, and Cumulative Forfeiture Adjustment.
Here’s what makes this reference work:
Patterns, tone, structure, rules extracted from reference: The reference uses monthly periods. Each tranche is on its own row. The forfeiture adjustment is calculated as (Actual cumulative forfeitures minus Expected cumulative forfeitures) times fair value per option, applied prospectively. The cumulative expense column uses a formula that references the vesting completion percentage.
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: A table with 12 rows (one per month from Jan to Dec 2026) showing the expense for each of the three tranches and the total.
Recipient’s reaction: Our controller can copy these numbers directly into the trial balance without recalculating.
Does NOT sound like: A theoretical explanation of IFRS 2. It must be a practical, numbers-driven schedule.
Success means: The total expense for 2026 matches our external valuation report’s expected expense, and the forfeiture adjustment is traceable to actual headcount data.
My context file contains my standards, constraints, audience. Read it fully before starting.
DO NOT start executing yet. Ask clarifying questions first.
Give me your execution plan (5 steps max) before you begin.
This second prompt targets the operational side of IFRS 2 accounting. The key is the reference file—you are telling ChatGPT to replicate the format your auditor already approved in a prior year. This dramatically reduces review time. The success brief specifies that the controller should be able to “copy these numbers directly into the trial balance.” That is a measurable outcome. When ChatGPT returns its execution plan, it will typically propose steps like: (1) Calculate the fair value per tranche, (2) Apply the expected forfeiture rate to each tranche, (3) Determine the monthly expense using straight-line recognition, (4) Calculate the cumulative forfeiture adjustment for Q1 2026 actuals, (5) Output the table. You can then approve or modify each step before the model generates the final schedule.
One practical tip: after ChatGPT generates the schedule, ask it to produce a reconciliation back to the total fair value at grant. For example, if total fair value at grant was $1.5 million (100,000 options × $15), and the year-one expense after forfeitures is $450,000, the model should show that this represents 30% of the total—consistent with a three-year graded vesting with an 8% forfeiture rate. This reconciliation is exactly what an auditor will ask for. If ChatGPT cannot produce it, you know there is an error in the logic.
What to try next: use these two prompts as a template for other IFRS 2 scenarios—cash-settled share-based payments, modifications to option terms, or equity-settled plans with market conditions. The structure (task, files, reference, success brief, execution plan) works for any complex accounting judgment. The more specific your reference file—whether it is a prior-year disclosure, a valuation report, or an auditor’s comment letter—the better the output. Over time, you will build a library of these prompts for recurring quarterly tasks, cutting your close time by hours while improving audit defensibility.
Published on 18 June 2026 on growwithgpt.com