Using AI to Analyze a Business (What Works & What’s Dangerous)

 

How to Use AI to Analyze a Business Before Buying It: Four Levels Every Buyer Should Understand

Artificial intelligence is changing the way people analyze businesses.

Whether you're evaluating a company you want to buy, reviewing your own business's financial performance, or preparing a business plan, AI tools can help you process information much faster than traditional methods.

I've been experimenting with these tools for the past couple of years, and they've become an increasingly important part of my consulting and business analysis work.

But I've also discovered something that concerns me.

The more powerful these tools become, the easier it is to confuse a convincing AI-generated answer with a reliable business decision.

AI can organize thousands of pages of information, identify unusual patterns, compare financial data, and help prepare questions for your professional advisers.

But it can also misunderstand documents, overlook important context, and confidently produce incorrect conclusions.

And when you're considering investing hundreds of thousands or even millions of dollars into a business, those mistakes can be extremely expensive.

I want to share the four levels of AI usage I've identified through my own experience, explain how I'm using these tools in business acquisition analysis, and show you why human judgment remains essential.

Why AI Is Becoming Such a Powerful Tool for Business Analysis

One of the biggest advantages of artificial intelligence is its ability to work through large amounts of information.

Consider what happens when you're evaluating a business acquisition.

You might receive several years of financial statements, tax returns, customer information, inventory reports, contracts, and a confidential information memorandum.

You may also need to research the company's industry, competitors, market conditions, and operating requirements.

Traditionally, reviewing all this information could require hours or days of work.

AI can help organize the material, identify potential inconsistencies, and prepare summaries much more quickly.

That doesn't mean it has verified the information or determined whether the acquisition is a good idea.

But it can help you understand what you're looking at and identify areas that deserve further investigation.

How AI Helped Me Reduce My Own Business Expenses

Here's a practical example from my own business.

At the end of the year, I downloaded PDF copies of my business credit card statements.

I uploaded them into ChatGPT and asked it to examine the expenses for patterns and potential savings opportunities.

The system categorized the transactions and identified software subscriptions and communication services that might be worth reviewing.

That led us to reconsider several platforms we were using.

Ultimately, we moved away from three separate platforms and consolidated those functions into one less expensive solution.

The AI didn't independently make that business decision.

It organized the information and helped identify an opportunity.

We then evaluated the options and decided what to change.

That's an excellent example of how these tools can support business management.

AI Is Particularly Useful When You Provide the Information

Another task where I've found AI useful is summarizing lengthy conversations.

Sometimes I'll have a ninety-minute conversation with a business owner or buyer.

I can take the transcript, upload it into ChatGPT, and ask for a case study.

The system can quickly organize the conversation into a coherent article.

Then I can review it and request improvements.

For example, I might ask it to include direct quotations from the conversation or provide more context around a particular decision.

The important point is that I'm providing the underlying information.

The AI is helping me organize and present that material.

That is different from asking a chatbot to invent an analysis based on incomplete information.

And that distinction becomes increasingly important as we move through the four levels of AI usage.

Level 1: Asking AI Questions Like a Search Engine

The first level is where most people begin.

They open ChatGPT or another chatbot and type a question.

For example:

How do I value a small business?

What should I look for when buying a plumbing company?

How much working capital do I need?

The AI produces an answer.

It may provide a useful overview, explain common terminology, or organize general information.

But you're still using the tool primarily as a question-and-answer system.

The Problem With Generic AI Answers

Large language models generate responses based on patterns learned during training and, depending on the tool and settings, information retrieved from other sources.

They can produce sophisticated explanations.

But a well-written answer isn't necessarily an accurate answer.

And generic questions often produce generic advice.

Suppose you're evaluating a plumbing company.

You ask AI whether it's a good industry to enter.

You might receive a list of potential advantages and disadvantages.

But that doesn't tell you whether the particular plumbing company you're considering is worth buying.

It doesn't establish whether the company's financial statements are reliable.

It doesn't tell you whether its employees are likely to remain after the acquisition.

And it certainly doesn't determine whether the asking price makes sense.

For those questions, you need much more specific information.

That's where Level 2 begins.

Level 2: Using Structured Prompts and Professional Context

At the second level, you stop treating AI as a simple search engine.

Instead, you begin giving it instructions about the role it should play, the information it should consider, and the output you want.

For example, imagine you're evaluating a small manufacturing company.

Rather than asking, Is this a good business to buy?, you could provide a more structured instruction.

You might tell the AI:

I'm evaluating the acquisition of a manufacturing company. I want you to approach this as an M&A advisory exercise. Ask me questions about the business, its financial performance, customers, operations, and risks. Then help me prepare a list of issues to discuss with my accountant and lawyer.

That gives the system a clearer framework.

You're defining the context, the task, and the desired output.

You're also encouraging the AI to ask questions rather than immediately provide conclusions.

Using AI to Prepare for Professional Adviser Meetings

One particularly useful application is preparing for meetings with accountants, lawyers, and lenders.

Suppose you've received a confidential information memorandum from a business broker.

You could upload the document and ask AI to identify the questions your accountant is likely to raise.

You could ask what legal issues your lawyer might want to investigate.

Or you could request a list of financing concerns that a commercial lender may have.

This can help you arrive at those meetings better prepared.

You aren't replacing the professionals.

You're using AI to organize your thinking so you can have more productive conversations with them.

Ask AI What You Haven't Considered

One question I particularly like is:

What important area of concern have we not discussed?

This can be useful because people tend to focus on the subjects they're already familiar with.

A buyer with a finance background may concentrate on earnings and debt.

Someone with an operations background may focus on staffing and processes.

Asking AI to suggest overlooked areas can help broaden the discussion.

Of course, you still need to determine whether those suggestions are relevant.

But it's a useful way to challenge your initial assumptions.

Why Structured Prompts Still Have Limitations

Even with a carefully written prompt, the AI may lack the specialized knowledge necessary to evaluate a particular industry.

You can ask it to act as an experienced acquisition adviser.

But telling a language model to adopt a professional role doesn't make it a licensed accountant, lawyer, or business appraiser.

And it doesn't automatically give the system reliable industry benchmarks.

To produce more meaningful analysis, you need to provide a stronger foundation.

That brings us to Level 3.

Level 3: Building AI Tools Using Expert Knowledge and Industry Data

This is where I believe AI becomes especially interesting.

At Level 3, you're no longer simply asking questions or providing more detailed prompts.

You're building a repeatable analytical framework.

You provide the AI with expert-developed information, benchmarks, scoring criteria, checklists, and methodologies.

Then you instruct it to apply those standards to a particular business or document.

Instead of asking for a general opinion, you're telling the system what to examine and how to evaluate it.

The value comes from combining a well-developed methodology with AI's ability to process information quickly.

I've been developing tools like this for my own consulting work, and the efficiency improvements have been significant.

Example 1: Creating a Customized Business Acquisition Due Diligence Checklist

One of the AI tools we've added to Business Buyer Advantage is a due diligence checklist generator.

The user provides the name of a real business they're considering purchasing.

The tool is designed to research the business, its industry, and its geographic location.

For example, imagine you're evaluating a tile and hardwood flooring company in California.

The tool can investigate the type of business, relevant industry considerations, and potentially applicable licensing or regulatory requirements.

It then uses a due diligence checklist I've developed through years of experience and identifies which items are relevant to that specific acquisition.

The result is a customized checklist intended to help the buyer organize their investigation.

Why This Is Different From a Generic AI Checklist

You could simply ask ChatGPT:

Give me a due diligence checklist for buying a flooring company.

And it would probably produce something useful.

But a specialized tool goes further.

It combines a predefined methodology with information about the actual business being evaluated.

That creates a more focused starting point.

However, any external facts it retrieves still need to be verified.

The tool might identify a licensing requirement, but the buyer and their advisers must confirm whether it applies.

It may suggest investigating environmental issues, but that doesn't mean the company actually has an environmental problem.

The purpose is to help organize due diligence, not to declare that due diligence is complete.

Example 2: Using AI to Evaluate a Business Plan

Another tool we've developed is part of my Cash Flow Forecasting and Business Plan Writing Program.

This one is designed to help students evaluate the quality of their business plans.

The program includes approximately 55 hours of video instruction.

We used the instructional transcripts to help develop a twelve-page analysis document describing what a strong business plan should contain.

The document establishes criteria for evaluating the plan.

It identifies important components, explains what strong performance looks like, and provides a scoring system.

How the Business Plan Scoring System Works

The framework evaluates approximately twenty categories.

Each category can receive a score of zero, one, or two.

Score	Meaning
0	The required element is absent
1	The element is present but incomplete
2	The element is adequately developed according to the framework

The AI uses these criteria to review a student's business plan.

It can identify missing sections, explain weaknesses, and suggest improvements.

For example, if the financial assumptions aren't adequately supported, the tool can highlight that problem.

If a section lacks important information, it may suggest language or additional analysis that would strengthen the plan.

The student can then review those suggestions and decide what changes are appropriate.

Why the Evaluation Framework Matters

The important part isn't simply that AI can read a business plan.

It's that the tool has been given a methodology for evaluating one.

It knows which categories to examine.

It has defined scoring criteria.

And it has examples of what a strong response should look like.

That's a much more useful application than asking a chatbot whether your business plan is good.

Example 3: Analyzing Plumbing Businesses Using Industry Benchmarks

I was recently interviewed on John Wilson's Owned and Operated podcast.

John owns HVAC and plumbing businesses.

During our conversation, he explained that one metric he pays close attention to when evaluating plumbing companies is gross margin.

If a company's gross margin is lower than it should be, he may see an opportunity to improve operations.

Now imagine building an AI tool specifically designed to analyze plumbing businesses.

You could provide the system with reliable industry benchmarks.

You could define the financial metrics that matter.

And you could establish rules for identifying potential opportunities or warning signs.

For example, if a company's gross margin falls below a relevant benchmark, the tool could flag it for further investigation.

It might suggest examining pricing, labor efficiency, material costs, or other operating factors.

But here's the important distinction.

A low gross margin doesn't automatically mean the business is an attractive acquisition.

There could be legitimate reasons for the difference.

The tool helps identify an issue.

The buyer still needs to investigate the underlying cause.

Why Buyers Should Choose an Industry Before Building AI Analysis Tools

This connects to advice I've given business buyers for years.

Choose an industry and learn as much as you can about it.

You don't necessarily need to have worked in that industry.

But you should understand how businesses in that field operate.

Read industry publications.

Study trade association resources.

Learn about the financial metrics that matter.

Understand the typical operating model.

And investigate the problems that commonly affect companies in that industry.

Once you've collected reliable information, you can use it to create more specialized AI analysis tools.

The Gas Station Example

Imagine building a tool to evaluate gas stations.

You could develop a framework that examines fuel volume, merchandise sales, gross margins, staffing costs, and other operating metrics.

You might investigate relationships between fuel sales and convenience store purchases.

You could examine whether having a car wash affects profitability.

Or you could compare the potential economics of a gas station with and without a car wash.

The AI could help organize those comparisons.

But the underlying benchmarks and assumptions need to come from reliable information.

Without that foundation, the tool may simply produce plausible-looking analysis that isn't grounded in the realities of the industry.

Building Your Own Library of Comparable Businesses

Another advantage of specializing in an industry is that you can gradually develop a reference library.

Suppose you evaluate ten plumbing companies over several months.

You collect financial information, operating metrics, and observations from each opportunity.

Over time, you may begin identifying patterns.

You might notice differences in gross margins, staffing structures, customer concentration, or revenue per employee.

You can then use those observations to refine your analytical framework.

As your reference library grows, your AI tools may become more useful.

Of course, you need to protect confidential information and make sure you're authorized to use the data.

And you need to remember that a small collection of companies may not represent an entire industry.

But the general principle is valuable.

The more relevant, reliable information you provide, the more focused your analysis can become.

Example 4: My Business, Industry, Market, and Model Report

One of the tools I've developed for my consulting work is something I call a BIM report.

BIM stands for Business, Industry, Market, and Model.

When someone hires me for a Buyer Insight Analysis or a Most Probable Selling Price evaluation, I need to understand the industry and market in which the business operates.

Historically, I'd spend roughly two hours visiting websites, gathering data, and organizing the information.

I wanted to understand what a business in that industry should look like before analyzing the specific company.

Now I've developed a framework that directs an AI tool to research particular sources and collect relevant information.

It produces an organized industry and market summary much more quickly.

What previously took me around two hours can now take only a few minutes to assemble.

I still need to examine the information.

I still need to consider whether the sources are reliable.

And I still need to apply my own professional judgment.

But the time saved on collecting and organizing information is substantial.

That's the kind of efficiency I'm excited about.

Level 4: Human Skepticism and Independent Judgment

The fourth level may be the most important.

And unlike the previous levels, it isn't about giving AI more responsibility.

It's about understanding when not to trust it.

The more experienced you become with AI tools, the easier it may be to develop excessive confidence in their outputs.

You create a sophisticated prompt.

You provide detailed information.

The system produces an impressive report.

And because the report looks professional, you begin assuming the conclusions must be correct.

That's dangerous.

As I explained:

"AI can help you think, it can help you analyze vast amounts of data, but it cannot do your thinking for you" —David C. Barnett

The purpose of AI should be to improve your analysis, not replace your responsibility for making decisions.

The Danger of AI-Generated Confidence

I recently came across an article on Neuroscience News discussing research about confidence and the use of large language models.

The article described a concern related to the Dunning-Kruger effect.

The basic idea is that people with limited knowledge may underestimate the complexity of a task.

As they learn more, they begin to recognize how much they don't know.

But the research discussed in the article raised a different concern about AI.

Frequent use of language models may encourage some users to become increasingly confident in AI-assisted judgments.

That confidence can become problematic when people stop independently checking the system's work.

The Hot Air Balloon Example

Imagine I invited you to my house this weekend to build a hot air balloon.

At first, the idea might sound simple.

You need a large bag.

You fill it with hot air.

And the hot air creates lift.

How difficult could that be?

But once you begin the project, you start discovering problems.

What material should the balloon use?

How strong does it need to be?

How will you control the altitude?

What heating system is required?

How do you ensure the equipment is safe?

Suddenly, a project that seemed straightforward becomes much more complicated.

That's what happens when you begin developing genuine expertise.

You discover the details that weren't obvious at the beginning.

Business acquisitions work the same way.

And AI can sometimes create the opposite impression.

It can make a complicated problem look deceptively simple because it produces a polished explanation.

AI Can Sound Certain Even When It's Wrong

A language model can produce a convincing analysis that contains serious mistakes.

It may misunderstand a financial statement.

It may overlook an important contractual provision.

It may rely on outdated information.

It may make assumptions that aren't supported by the documents.

And it may present those conclusions with considerable confidence.

As I explained:

"AI can make mistakes. AI can misunderstand documents. AI may sound confident while being wrong." —David C. Barnett

That's why you need to verify important findings.

If the AI identifies a licensing requirement, confirm it with the appropriate authority.

If it calculates a financial ratio, check the underlying numbers.

If it identifies a legal concern, discuss it with your lawyer.

And if it suggests that a business is an attractive investment, investigate the assumptions behind that conclusion.

The more consequential the decision, the more important independent verification becomes.

AI Doesn't Know Whether the Seller Is Telling the Truth

This is especially important in business acquisitions.

Suppose a seller provides financial statements showing strong revenue growth.

You upload those statements into an AI tool.

The system calculates financial ratios, identifies trends, and produces a positive assessment.

But what if the seller's numbers are inaccurate?

What if some revenue isn't collectible?

What if expenses have been improperly classified?

What if a major customer is about to leave?

The AI may analyze the information you've provided very effectively.

But that doesn't establish whether the underlying information is true.

Analyzing a seller's claims is not the same as verifying those claims.

That's why accountants, lawyers, lenders, appraisers, and other experienced professionals remain important.

AI can help you prepare for their work.

It cannot automatically replace it.

What AI Can and Cannot Do in Business Due Diligence

Here's a useful way to think about the distinction.

AI can help you	What still requires verification or human judgment
Organize financial statements	Confirm the records are accurate and complete
Identify unusual expense patterns	Determine the actual reasons behind those expenses
Summarize contracts	Interpret legal obligations and enforceability
Compare industry benchmarks	Determine whether the benchmarks apply
Generate due diligence checklists	Verify that the investigation is complete
Prepare questions for advisers	Obtain qualified professional advice
Model financing scenarios	Confirm financing availability and terms
Identify potential operational risks	Investigate actual conditions at the business
Review business plans	Validate assumptions and commercial feasibility
Compare acquisition opportunities	Decide whether the investment fits your objectives

AI can be an excellent analytical assistant.

But it doesn't assume responsibility for the consequences of your acquisition.

You do.

The Biggest Mistake: Analyzing a Business Entirely From Your Computer

There's another concern I want to emphasize.

Even before AI became widely available, I regularly encountered buyers who spent too much time analyzing businesses from behind a computer screen.

They'd review spreadsheets.

They'd study financial statements.

They'd examine presentations.

And they'd develop increasingly sophisticated financial models.

But they sometimes neglected to investigate what was actually happening inside the business.

AI could make this problem worse.

If you can generate detailed reports and financial analyses in minutes, you may feel that you've developed a deep understanding of a company.

But a business exists in the real world.

It has employees, customers, equipment, facilities, and operating problems.

And not all of those realities appear in a spreadsheet.

The Plumbing Company Example: What You Might Discover by Watching Operations

Let's imagine you're evaluating a plumbing company with five service vans.

The employees typically arrive at the office around 7:30 in the morning.

Their first service calls begin around 8:00.

You could review the company's payroll records, service schedules, and financial statements.

You could ask AI to calculate revenue per employee or examine labor efficiency.

Those analyses might be useful.

But I'd also want to see what actually happens in the morning.

With appropriate access and respect for privacy and applicable laws, observe how employees arrive, prepare for work, and depart for service calls.

Are the vans ready?

Are employees organized?

Does the operation appear efficient?

Are there obvious delays or problems?

These observations can help you identify questions that may not emerge from the financial statements alone.

The objective isn't to replace formal due diligence with casual observation.

It's to combine documentary analysis with a real understanding of how the business operates.

The Property Management Lesson: Sometimes You Need to See Things Yourself

A friend of mine operated a property management company for many years.

Part of his regular routine involved driving around to inspect the properties he managed.

He'd look for unusual conditions.

Maybe mail had been accumulating on a porch.

Perhaps there were unexpected signs of activity at a property.

These observations could suggest that something required further investigation.

The important point is that he wasn't relying exclusively on reports.

He was looking at the actual properties.

That kind of practical observation is difficult to replace with an AI-generated summary.

And it illustrates a broader principle.

The numbers tell part of the story. The real-world operation tells another part.

You need both.

How to Use AI Responsibly When Buying a Business

If you're considering using AI during a business acquisition, I recommend treating it as a structured support tool rather than an independent decision-maker.

Start by defining what you're trying to learn.

Provide the relevant documents and explain the context.

Ask the system to organize information, identify potential concerns, and prepare questions.

Then evaluate the results critically.

For example, you might follow this process:

1. Organize the Information

Use AI to summarize the confidential information memorandum, financial statements, and other materials provided by the seller.

2. Identify Potential Risks

Ask it to highlight unusual trends, inconsistencies, missing information, and assumptions that require verification.

3. Apply Industry-Specific Knowledge

Use reliable industry benchmarks and expert-developed methodologies to evaluate the company's performance.

4. Prepare for Professional Reviews

Generate questions for your accountant, lawyer, lender, and other advisers.

5. Verify Important Findings

Check calculations, source documents, regulatory requirements, and significant assumptions.

6. Investigate the Business in Person

Visit the operation where appropriate, examine equipment, meet relevant people, and observe how the company functions.

7. Make Your Own Decision

Use the information gathered through AI, professional advisers, and direct investigation to decide whether the acquisition makes sense.

AI can help make the process more efficient.

But the responsibility for understanding the risks remains with you.

Why AI Should Make Business Buyers More Curious, Not More Reckless

The most valuable outcome of using AI isn't necessarily getting an answer faster.

It's being able to investigate more questions.

You can review more information.

You can explore alternative scenarios.

You can identify areas that deserve further examination.

And you can prepare more effectively for conversations with experienced professionals.

That's how I want buyers to approach these tools.

Instead of asking AI to tell you whether a business is worth buying, use it to help determine what you still need to understand.

What assumptions haven't been tested?

What documents are missing?

What financial trends need explanation?

What industry risks haven't been considered?

What questions should you ask the seller?

And what information would change your decision?

Those are productive uses of artificial intelligence.

Final Thoughts: AI Can Improve Your Analysis, but It Can't Replace Experience

Artificial intelligence is becoming an increasingly useful tool for business owners, buyers, and advisers.

I've personally experienced the benefits.

It's helped me identify opportunities to reduce business expenses.

It's made it easier to summarize lengthy conversations.

It's helped me develop customized due diligence tools and business plan evaluation frameworks.

And it's significantly reduced the time required to gather and organize industry research for consulting projects.

But the biggest lesson I've learned is that better technology doesn't eliminate the need for sound judgment.

As I explained:

"AI can help you examine all that more carefully, all of it. But it should make you more cautious and more curious, not more reckless." —David C. Barnett

When you buy a business, you're not simply purchasing numbers on a spreadsheet.

You're acquiring customers, employees, systems, equipment, contracts, relationships, risks, and the expectation of future cash flow.

Those things need to be understood in context.

AI can help you process information faster.

It can help you identify questions you might otherwise overlook.

And when combined with reliable data and expert-developed methodologies, it can become a powerful analytical tool.

But it cannot guarantee that the seller is telling the truth.

It cannot replace professional due diligence.

And it cannot take responsibility for your financial decisions.

Use AI to become a better-informed buyer, not a more overconfident one.

The objective isn't to let the machine make the decision.

It's to use the machine to help you make a better decision yourself.

Learn How to Analyze a Business Before You Buy

If you're considering buying an existing business, my Business Buyer Advantage program provides online training, group coaching, and consulting services designed to help you evaluate acquisition opportunities, understand financial performance, identify risks, and structure transactions.

Explore Business Buyer Advantage

If you're developing a business plan or want to improve your cash flow forecasting skills, explore my Cash Flow Forecasting and Business Plan Writing Program, including the business plan analysis framework discussed in this article.

Explore Business Plan School

For more information about my books, videos, and consulting services, visit my website.

Visit DavidCBarnett.com

👉 Want deeper dives like this? Join my email list at DavidCBarnettList.com for early access to videos, insights, and 7 free bonus gifts.