Expert Tips for Engineering Effective GenAI Prompts
NACVA recently held a webinar on how to use GenAI “prompt engineering” to obtain better results when using tools like ChatGPT, Copilot, Claude, Perplexity.ai, and Gemini. Colin Brown — the founder and chief technology officer of Syncnet, which helps consultants leverage AI and software — shared a variety of tips, including several that CVAs can use to save time and significantly improve their efficiency.
Understanding GenAI Tools
For valuators’ purposes, GenAI tools are best thought of as very smart interns who can tap trillions of data points, but who don’t understand the particulars of a valuation report. As Brown explained, that means they need a lot of instruction.
Equally important, they’re also like interns in that they want to please you, even when they can’t find a solid answer to your query. In other words, they may return information that’s inaccurate or outright false. So, when it comes to AI-generated content, it’s essential that you verify it before you rely on it.
But don’t let that discourage you from taking advantage of these powerful tools. Prompt engineering can go a long way toward improving your odds of quickly locating useful information.
5 Vital Prompt Engineering Tips
Brown outlined five techniques — or “pillars” — for effective prompts. “If you just follow these,” he said, “your prompts will be better than those of most ChatGPT users.”
- Role/Persona
Define for the GenAI tool whom it should be as it tackles the query. For example, you could tell it: “You are a senior business valuation expert reviewing this Excel file as part of a quality control process.”
- Context
Provide the tool with relevant background information and situation details for accurate responses. For example, “prepare a report for litigation purposes.”
- Task
Specify precisely what you want the tool to do — the desired action or outcome — using clear and actionable instructions. Examples include:
- “Write a professional, 200-word industry overview for a $5M general dentistry practice valuation in the Southeast U.S. Include macro trends, competitive landscape, and demand drivers.”
- “Identify any broken formulas, hardcoded values in calculation cells, or circular references in this valuation model.”
- “Review how valuation assumptions (e.g., discount rate, growth, working capital) are presented in this model. Are they clearly sourced, separated from calculations, and easy to trace?”
- Format
Define the desired structure of the result(s). For example, you could request a:
- Summary
- List
- Table
- Spreadsheet
- Bullet points
- PowerPoint
- Constraints
Set boundaries or requirements for the results the tool provides. For example, “Use only reputable sources” or “Find transactions in the range of $500,000-$1 million in the year 2025.”
Example: Ineffective vs. Effective Prompt
Brown illustrated the difference his five pillars make with two examples.
Ineffective: “Write a report about our company’s performance.”
This prompt is far too vague and ambiguous to produce a useful result.
Effective: “Act as a senior business analyst. Create a Q3 2026 performance report for our SaaS company, targeting C-suite executives. Include revenue growth, customer acquisition metrics, and market position analysis. Format as a 2-page executive summary with key insights, data visualizations, and actionable recommendations. Focus on strategic implications and next quarter priorities.”
This prompt clearly defines the role (senior business analyst). It specifies context (a SaaS company report for C-suite executives) and a timeframe constraint (Q3 2026). And it describes the desired format and content requirements.
While the example isn’t valuation-specific, you can see how including the five pillars greatly improves the odds of getting the results you seek from your initial prompt.
Important: Kicking off his webinar, Brown noted that it was the fourth time he’d given a presentation on this topic — and it changed every time. In other words, today’s tips are likely to be refined or even replaced, making it important to stay up to date with the latest developments in AI capabilities.
Don’t Reinvent the Wheel If You Don’t Need To
Wading into the GenAI waters for valuation purposes can understandably seem daunting. The good news is that you’re not on your own. NACVA’s AI Data University offers a growing library of prompts and other AI tools you can search, filter, and save to use again and again in your valuation practice.