By Adeyemi Adebayo, Mfon Akpan, and Scott Dell.
The art of crafting adequate inputs for AI is the foundation that helps automate tedious financial tasks and enhance decision-making. Rather than fearing AI will replace them, finance professionals are learning to partner with AI tools like ChatGPT to boost productivity. Prompt engineering is a critical skill for accountants, focusing on appropriately phrasing questions or instructions to get valuable outputs from generative AI.
By mastering the necessary skills, accountants can save hours on reporting and analysis, reduce errors, and free up time for strategic work. In this context, this article explores practical applications of prompt engineering in finance, shares real-world case studies, and offers best practices to get accurate and valuable results from AI.
Introduction to Prompt Engineering
Prompt engineering refers to designing and refining the questions or commands we give to AI LLM (large language model) systems. According to White et al. it is about how you “ask” an AI like ChatGPT to do something. Just as asking a colleague the right question yields better information, phrasing a prompt clearly for an AI yields better outputs. Meskó states that for accountants, prompt engineering matters because knowing how to ask AI tools properly can dramatically boost productivity and accuracy while reducing the potential for hallucinations or inappropriate answers.
With well-crafted prompts, an AI can draft a report, summarize a 100-page financial policy, or crunch through sales data in seconds. This means tasks that used to take hours, like writing variance analysis commentary or extracting numbers from invoices—can be automated or accelerated with AI assistance. Rather than replacing accountants, Ye et al. argues that these tools act as “virtual assistants” that handle repetitive chores, enabling professionals to focus on analysis, strategy, and decision support.
Prompt Engineering Techniques
Understanding and applying different prompt engineering techniques can improve AI outputs. Thus, it is important to detail prompt engineering techniques. Below are four key techniques used by finance professionals to maximize AI’s effectiveness:
Zero-Shot Prompting
Zero-shot prompting involves asking AI to complete a task without providing any prior examples. According to ICAEW Insights this technique relies on the model’s pre-trained knowledge to generate responses based on the prompt alone.
Example:
Prompt: “Summarize the key financial risks in a company’s annual report.”
AI Output: “Key financial risks in a company’s annual report typically include:
- Liquidity Risk – The risk of not having enough cash or liquid assets to meet short-term obligations.
- Credit Risk – The risk of customers or counterparties defaulting on payments owed to the company.
- Market Risk – Exposure to fluctuations in market variables such as interest rates, exchange rates, and commodity prices.
- Operational Risk – Risks arising from internal failures, such as fraud, system failures, or supply chain disruptions.
- Regulatory & Compliance Risk – The risk of non-compliance with laws and regulations, leading to fines or operational restrictions.
- Cybersecurity Risk – The threat of data breaches, hacking, or IT system failures impacting financial performance.
- Reputation Risk – Potential damage to brand value or customer trust due to negative events.
- Strategic Risk – The possibility of poor business decisions, changing market conditions, or industry disruptions affecting profitability.
- Inflation & Economic Risk – The impact of economic downturns, inflation, or geopolitical instability on financial performance.
- Environmental, Social, and Governance (ESG) Risk – Risks related to sustainability, climate change, and corporate governance issues.
These risks are usually detailed in the “Risk Factors” section of the annual report.” ChatGPT4o
Zero-shot prompting is helpful for general knowledge tasks, but results can vary depending on the complexity of the request.
Few-Shot Prompting
Few-shot prompting provides the AI with a few examples to guide its response. This technique helps improve accuracy by setting a precedent for the model to follow as highlighted by ICAEW Insights.
Example:
Prompt: “Here are examples of budget variance explanations:
- Revenue increased by 10% due to higher-than-expected sales.
- Operating expenses rose by 5% because of supply chain disruptions. Explain this variance: Actual revenue was $1.2M, while budgeted revenue was $1M.”
- AI Output: “Revenue increased by 20% due to higher-than-expected sales.” ChatGPT4o
Few-shot prompting is particularly effective for generating financial commentary, classifying transactions, or structuring reports.
Chain-of-Thought Prompting
Chain-of-thought (CoT) prompting encourages the AI to explain its reasoning step by step before concluding. This technique enhances logical reasoning and is beneficial for complex financial analysis. The addition of “Reasoning Models” such as OpenAI’s o1 and o3 models, has much of the CoT functionality built into its ecosystem.
Example:
Prompt: “A company’s revenue increased by 15% in Q3, but net profit declined by 5%. Analyze why this could have happened, step by step.”
AI Output:
“When a company experiences a 15% increase in revenue but a 5% decline in net profit, several factors could explain this discrepancy. Let us break it down step by step:
1. Increase in Costs and Expenses
- Higher Cost of Goods Sold (COGS): If raw materials, labor, or production costs increased at a higher rate than revenue, gross profit margins would shrink.
- Increased Operating Expenses: More spending on marketing, R&D, salaries, or administrative costs can eat into profitability.
- Higher Distribution or Logistics Costs: If shipping, supply chain disruptions, or fuel costs increased, this could reduce net profits.
2. Pricing and Discount Strategies
- If the revenue increase was driven by price reductions, promotions, or discounts, the company’s profit per unit could have declined, leading to lower net margins.
3. Unfavorable Product or Service Mix
- If the company sold more lower-margin products or services compared to previous quarters, revenue might grow, but overall profitability could decline.
4. Higher Interest or Financing Costs
- If the company took on more debt, higher interest expenses could reduce net profit.
- Rising interest rates could increase the cost of existing loans, further impacting profitability.
5. Increased Depreciation or Amortization
- If the company acquired new assets or technology, higher depreciation or amortization costs could lower net income.
6. One-Time or Non-Operating Expenses
- Legal fees, settlements, restructuring costs, or write-offs could temporarily reduce net profit.
- Foreign exchange losses from currency fluctuations may also impact net earnings.
7. Tax Rate Changes
- If the company’s effective tax rate increased due to regulatory changes or fewer tax credits, net income would decline.
Conclusion
The revenue increase suggests strong sales growth, but declining net profit indicates rising costs, pricing strategy issues, changes in product mix, higher financing expenses, or external factors like taxes or one-time costs. A deeper look into the income statement and financial disclosures would reveal the exact cause.” ChatGPT4o
Chain-of-thought prompting is particularly valuable for root-cause analysis, audit explanations, and scenario-based financial reasoning.
Chain of Drafts Prompting
According to Xu et al’s article Chain of Drafts prompting structures AI responses progressively, ensuring iterative improvements in depth and accuracy.
Example:
Prompt: “Analyze a company’s profitability trends over time by following these steps:
- Provide a high-level overview of factors influencing profitability.
- Expand on revenue and cost drivers.
- Analyze external (economic conditions, inflation) and internal (cost-cutting, efficiency) influences.
- Compare profitability trends over the last three years.
- Suggest strategic actions to improve profitability while managing risk.”
AI Output:
“To analyze a company’s profitability trends over time, we will follow the structured approach outlined:
1. High-Level Overview of Factors Influencing Profitability
Profitability is primarily driven by the balance between revenue generation and cost management. Key factors influencing profitability include:
- Revenue Growth – Sales volume, pricing strategy, product mix, and market demand.
- Cost Structure – Fixed and variable costs, cost of goods sold (COGS), operating expenses.
- Market Conditions – Industry trends, competition, and customer preferences.
- Operational Efficiency – Productivity, technology adoption, supply chain management.
- Financial Management – Capital structure, interest expenses, and tax strategies.
2. Revenue and Cost Drivers
Revenue Drivers
- Sales Volume – Increased market penetration and customer retention.
- Pricing Strategy – Premium pricing, discounts, and promotions.
- Product/Service Mix – High-margin vs. low-margin products.
- Geographical Expansion – Entry into new markets.
- Customer Base Growth – Acquisition strategies and brand loyalty.
- Digital Transformation – E-commerce and online sales.
Cost Drivers
- COGS – Raw materials, labor, and supply chain costs.
- Operational Expenses – Rent, utilities, marketing, R&D.
- Labor Costs – Wage inflation, benefits, and training costs.
- Technology & Automation – Initial investment vs. long-term savings.
- Regulatory Costs – Compliance and legal expenses.
- Debt Servicing – Interest expenses and capital costs.
3. External and Internal Influences
External Influences
- Economic Conditions – GDP growth, consumer spending, and demand fluctuations.
- Inflation – Increases in input costs and pricing power.
- Interest Rates – Impact on borrowing costs.
- Regulatory Changes – Compliance costs, tax policies, and trade restrictions.
- Industry Trends – Competitive landscape and technological disruptions.
Internal Influences
- Cost-Cutting Initiatives – Streamlining operations and renegotiating supplier contracts.
- Operational Efficiency – Process automation and lean management.
- Strategic Investments – R&D, brand development, and customer acquisition.
- Workforce Productivity – Employee engagement and talent development.
- Financial Restructuring – Debt management and capital allocation.
4. Profitability Trends Over the Last Three Years
To provide an in-depth trend analysis, I will need financial data such as:
- Revenue growth
- Gross profit margins
- Operating profit margins
- Net profit margins
- Return on assets (ROA) and return on equity (ROE)
If you provide financial statements, I can conduct a more detailed analysis, including trend comparisons and key insights.
5. Strategic Actions to Improve Profitability While Managing Risk
Revenue Enhancement Strategies
- Market Diversification – Expanding into new markets and customer segments.
- Product Innovation – Enhancing value propositions and launching new products.
- Dynamic Pricing Strategies – Adapting pricing based on demand and cost fluctuations.
- Strengthening Customer Relationships – Improving loyalty programs and personalized marketing.
Cost Optimization Strategies
- Supply Chain Efficiency – Reducing procurement costs and optimizing logistics.
- Process Automation – Investing in technology to lower operational costs.
- Outsourcing & Shared Services – Reducing in-house costs for non-core functions.
Risk Management Strategies
- Hedging Against Inflation – Forward contracts and cost adjustments.
- Diversifying Revenue Streams – Reducing reliance on a single product or market.
- Scenario Planning – Preparing for economic downturns and competitive threats.
Would you like me to perform a profitability analysis based on specific company financials? If you provide revenue, cost, and profit data for the last three years, I can generate visual trends and deeper insights.” ChatGPT4o
Chain-of-drafts prompting is helpful for refining AI-generated financial analyses step by step, ensuring logical progression, deeper insights, and improved accuracy over multiple iterations.
The authors of this article have also developed and shared additional methodologies for effectively structuring the content of the prompt. One such methodology that is actively used and has been presented at multiple conferences and workshops is the R-STAR-QC format of a prompt. The acronym stands for Role, Situation, Task, Action, Result, Question, and Constraints. Having a structured system for prompting, along with implementing the previously mentioned prompting considerations, will optimize your use of these powerful AI systems.
As you can see from the above prompts and responses, how you ask a question can lead to very different results. Each has its benefits and limitations. As highlighted in the following quote by Indira Gandhi, “The power to question is the basis of all human progress,” asking appropriate questions provides the opportunity for maximizing the quality of responses. Your ability to ask the right questions, along with the follow-up dialog/prompting, yields the true power of effectively using these ever-improving AI tools.
References
CFI Team. (2023). Chain-of-thought (CoT) prompting in AI-powered financial analysis. Corporate Finance Institute. Retrieved from https://corporatefinanceinstitute.com/resources/financial-modeling/chain-of-thought-prompting-financial-analysis/
ICAEW. (2025, February). Prompt engineering: Interacting with generative AI tools. Institute of Chartered Accountants in England and Wales. Retrieved from https://www.icaew.com/insights/viewpoints-on-the-news/2025/feb-2025/prompt-engineering-interacting-with-generative-ai-tools
Meskó, B. (2023). Prompt Engineering as an Important Emerging Skill for Medical Professionals: Tutorial. Journal of Medical Internet Research, 25(1), 1–6. https://doi.org/10.2196/50638
Xu, S., Xie, W., Zhao, L., & He, P. (2025). Chain of draft: Thinking faster by writing less. arXiv preprint arXiv:2502.18600. https://arxiv.org/abs/2502.18600
White, J., Fu, Q., Hays, S., Sandborn, M., Olea, C., Gilbert, H., Elnashar, A., Spencer-Smith, J., & Schmidt, D. C. (2023). A Prompt Pattern Catalog to Enhance Prompt Engineering with ChatGPT. ArXiv Preprint, 1–19. http://arxiv.org/abs/2302.11382
Ye, Q., Axmed, M., Pryzant, R., & Khani, F. (2024). Prompt Engineering a Prompt Engineer. Findings of the Association for Computational Linguistics:, ACL 2024, 355–385. http://arxiv.org/abs/2311.05661
====
About the Authors:
Adeyemi Adebayo
Email: deyemidebayo@zoho.com
Adeyemi Adebayo is a chartered accountant and senior research fellow at the University of South Africa. His research, teaching, and consulting interests include organizational governance, reporting, accountability, corporate responsibility regarding environmental, social, and governance practices (ESG), and broader organizational sustainability.
===-=
Scott Dell
Email: ScottDell@ferris.edu
Dr. Scott” Dell is an Assistant Professor of Accounting at Ferris State University. He is an award-winning academic, engaging AI keynoter, successful author, recovering CPA, lifelong learner, enthusiastic teacher, experienced entrepreneur, motivating coach, and proud Navy veteran. He thrives on building skills and knowledge of his students/clients/colleagues, helping them grow personally and professionally. His AI workshops, presentations, and keynotes are not to be missed.
He co-wrote with Mfon Akpan the AI guidebook for accountants, ChatGPT and AI for Accountants: A Practitioners Guide. He also authored “Your COMPLETE Guide to SUCCEEDING with LinkedIn” and co-authored the book/course “Living Your DREAM CAREER”. He has been published in multiple peer-reviewed journals and publications.
Mfon Akpan
Email: akpanm@nsuok.edu
Dr. Mfon Akpan is an assistant professor of accounting at Northeastern State University, specializing in AI and emerging technologies in finance. He has published extensively on generative AI, blockchain, and NFTs, exploring their impact on financial reporting, auditing, and strategic decision-making. Dr. Akpan co-authored *ChatGPT and AI for Accountants* with Dr. Scott Dell and *Future Proof Accounting: Data and Technology Strategies*. His research focuses on AI-driven financial analysis, efficiency optimization, and risk assessment.
Thanks for reading CPA Practice Advisor!
Subscribe Already registered? Log In
Need more information? Read the FAQs
