For most finance teams, the quarterly segment reporting close is a quiet exercise in controlled frustration. You are not solving a math problem; you are managing a narrative problem. IFRS 8 requires you to report segments based on the “chief operating decision maker” (CODM) view, which sounds straightforward until your CODM reviews a P&L that does not match your general ledger, your shared costs are allocated using a spreadsheet from 2019, and the auditors ask for a reconciliation that ties to the consolidated income statement with zero unexplained drift. The friction is real: you spend hours extracting data from ERP systems, mapping cost centres to segments, drafting explanatory footnotes, and then rewriting everything because a board member asked a question that changes the disclosure.
ChatGPT, when used correctly, does not replace your judgment—it compresses the mechanical and linguistic burden. It can draft the segment note in IFRS 8-compliant language, generate the required reconciliations in plain-English explanations, and even challenge your assumptions about what constitutes an “operating segment” under the aggregation criteria in paragraph 12. The key is not asking it a vague question like “help me with IFRS 8.” The key is feeding it a structured prompt that includes your actual segment data, your internal reporting pack, and the specific disclosure requirements you are trying to meet. When you do that, the model becomes a drafting partner that produces a first pass in minutes, not a junior analyst in days.
This post walks you through two practical applications: first, building the core segment disclosure note; second, drafting the “measurement of segment profit or loss” explanatory paragraph that auditors scrutinise most. Both prompts follow a strict anatomy that forces the model to read your context, ask clarifying questions, and produce output you can actually use—not generic boilerplate.
Why Most ChatGPT Prompts Fail in Financial Reporting
The typical failure mode is not the model’s capability; it is the prompt’s ambiguity. If you write “Draft a segment note for IFRS 8,” ChatGPT will happily produce a generic template that looks professional but contains no numbers, no reference to your internal reporting, and no reconciliation to consolidated totals. That output is worse than useless because it creates a false sense of progress. To get a useful result, you must treat ChatGPT as a new hire who has read IFRS 8 but knows nothing about your company. You need to give it the relevant files, the reference format, and the success criteria—then require it to ask questions before executing. The prompt structure below enforces exactly that discipline.
First, read these files completely before responding:
[segment_data_q3.xlsx] — Contains segment revenue, segment profit, total assets, and liabilities for each of our five operating segments, plus the consolidated totals from the general ledger.
[internal_management_report_q3.pdf] — The monthly CODM pack showing how the CEO and CFO actually review performance, including the column layout, subtotals, and any non-GAAP measures like adjusted EBITDA.
[prior_year_q3_note.docx] — The same note from last year, which was approved by our auditors, so you can see the expected structure and level of detail.
Here is a reference for what I want to achieve:
[Upload a clean, audit-approved segment note from a comparable public company as markdown, or describe its structure in detail.]
Here’s what makes this reference work:
It uses a table format with segment columns, a reconciliation column, and a “Total” column. Revenue is split into external and inter-segment lines. Profit measure is labelled as “segment operating profit” and clearly reconciled to profit before tax by adding back central costs and finance income. No narrative fluff; every line item is traceable to the CODM pack.
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: A full IFRS 8 note in Word-ready format, about 400-500 words plus a data table.
Recipient’s reaction: My audit partner should be able to compare it to the CODM pack and see no unexplained differences. They should not need to ask “where did this number come from?”
Does NOT sound like: Generic textbook language, overly legalistic phrasing, or a note that lists segments but does not show how management measures them.
Success means: The note reconciles exactly to the consolidated income statement, includes all required disclosures from IFRS 8 paragraphs 22-24, and passes an internal review in one round.
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.
Bridging from the Core Note to the Measurement Paragraph
Once the main segment table is drafted, the next hurdle is the narrative that accompanies it. IFRS 8 paragraph 27 requires you to explain the basis of accounting for any transactions between segments, and paragraph 28 demands a description of the nature of differences between segment profit and consolidated profit before tax. This is where most drafters struggle, because the differences are often subtle—shared IT costs allocated on headcount, a central treasury function not allocated at all, or a pension charge that sits only in the corporate column. The prompt below forces ChatGPT to produce that narrative from your actual reconciliation data, not from a template.
First, read these files completely before responding:
[segment_to_consolidated_recon.xlsx] — A detailed waterfall showing each adjustment from the sum of segment operating profit to consolidated profit before tax, including amounts for central costs, share-based payments, amortisation of acquisition intangibles, and finance income.
[accounting_policies_manual.docx] — The section of our accounting manual that describes how inter-segment pricing is set (cost-plus 5%) and which costs are not allocated to segments.
[auditor_review_notes_q2.docx] — The comments from our external auditors on last quarter’s draft, specifically on the wording of these paragraphs.
Here is a reference for what I want to achieve:
[Upload a prior year’s approved annual report section as markdown, or describe the exact phrasing used previously for the measurement basis and reconciliation.]
Here’s what makes this reference work:
It uses clear, declarative sentences. For example: “Segment profit is measured as operating profit before central overheads, share-based payments, and amortisation of acquisition-related intangibles. Inter-segment revenue is priced on a cost-plus-margin basis consistent with third-party transactions.” The reconciliation is presented as a list of line items with brief explanations, not as a single lump sum.
Here’s what I need for my version / SUCCESS BRIEF:
Type of output + length: Two narrative paragraphs (measurement basis and nature of differences) plus a bulleted reconciliation explanation. Total 250-350 words.
Recipient’s reaction: The auditor should be able to tick each reconciliation line to the supporting schedule and find the narrative description matches the accounting policy manual exactly.
Does NOT sound like: Vague phrases like “certain corporate costs” or “other adjustments.” No hedging language.
Success means: The narrative explains each reconciling item in the waterfall, states the inter-segment pricing policy, and explicitly confirms that segment assets are not reported to the CODM, if that is the case.
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.
Practical Tips for Controllers and CFOs
When you run these prompts, expect the first response to be a set of clarifying questions. That is by design. Do not skip that step and ask ChatGPT to proceed anyway. The questions will force you to articulate decisions you may have deferred, such as “Is the CODM the CEO or the CFO?” or “Do we report segment assets, given management does not review them?” Answering these questions in the chat is itself a valuable exercise—it surfaces gaps in your own documentation. Once you provide answers, the model will produce a draft that you can paste into your working paper, then mark up with tracked changes.
One caution: ChatGPT will occasionally invent a plausible-sounding policy that you do not follow. Always verify that the narrative matches your actual accounting manual before sending it to the auditor. The model is a drafting accelerator, not a source of truth. For your next step, try running the first prompt with your real Q3 data, but deliberately change one segment boundary in your input file to see if the model catches the inconsistency in the reconciliation. If it does not, tighten your context file with an explicit instruction that segment totals must tie to the CODM pack. That iterative refinement is where the real value lies.
Published on 4 September 2026 on growwithgpt.com