GEN AI IN PREDICTIVE ANALYSIS OF START-UP AND BUSINESS PURCHASE RISK
This article examines risk assessment in certain business transactions, primarily on start-ups and purchases of existing businesses. While risk cannot be determined with scientific certainty, attorneys and business owners can approximate a range of foreseeable risks based on available market data, industry trends, and emerging analytical tools including Generative AI. Understanding the statistical realities of business failure, the definitions and scope of small business in America, and the Generative AI methods for risk analysis provides essential context for analysis of risk of the range of any business transaction.
When referencing small businesses, many people envision modest operations with a handful of employees. However, the reality is that many small businesses are not actually “small” in the traditional sense. The Small Business Administration defines small businesses based on industry-specific size standards, generally measured by the number of employees or average annual receipts. Most manufacturing companies with 500 employees or fewer, and most non-manufacturing businesses with average annual receipts under $7.5 million, qualify as small businesses, though these standards vary significantly by industry and North American Industry Classification Code (NAICS) code.1
The Internal Revenue Service defines small business resources as serving taxpayers who file specific forms and small businesses with assets under ten million dollars.2 Of course, this definition focuses on asset size rather than employee count or revenue. For bankruptcy purposes, the debt limitation for a small business filing under Subchapter V of Chapter 11 of the U.S. Bankruptcy Code is $3,424,000 in combined total secured and unsecured debts as of June 2024.3 This amount is adjusted periodically for inflation.
According to the Small Business Administration, 99.9% of all American businesses are classified as small businesses, with over 34 million small businesses operating in the United States.4 These businesses are a significant part of the American economy, yet start-ups face intrinsic, historical challenges.
Below are failure rates over time, derived from the Bureau of Labor Statistics data and cited by the Small Business Administration:
After 1 year: Approximately 20% to 24% of new businesses fail.
After 2 years: The failure rate increases to about 30%.
After 5 years: Approximately 50% of new businesses fail.
After 10 years: The failure rate rises to about 65% to 70%.
(The official source for these failure rates is the Business Employment Dynamics program of the Bureau of Labor Statistics. The specific data on business survival is published in a table that tracks cohorts of businesses born in a specific year. See Table 7. Survival of private sector establishments by opening year, https://www.bls.gov/bdm/).
According to these metrics, only about half of all start-ups survive past their fifth year, and slightly less than one-third make it to their tenth anniversary.
The U.S. Chamber of Commerce synthesized data from sources like the U.S. Bureau of Labor Statistics and highlighted several of the most common issues that lead to failure:
Lack of market demand for the product or service; cash flow problems resulting from poor budgeting and financial management; poor management, including issues with leadership, planning, and hiring; Ineffective marketing strategies; and
Intense competition.5
Potentially overlooked risk in business transactions is the personal guarantee. When a business owner signs a personal guarantee for a business loan, they become personally liable for that debt if the business cannot pay. This means that when a business fails, the owner’s personal assets – including their home (a homestead is exempt from creditors in Texas – other states have their own exemption laws which vary greatly), savings, and other non-exempt property can be seized by various creditor procedures, with or without a court order, to satisfy the business debt. While specific statistics on the exact number of personal bankruptcies filed solely as a result of personal guarantees of failed business loans are difficult to obtain, this type of bankruptcy case appears, at least anecdotally, to be ubiquitous.
SBA loans over $200,000 generally require a personal guarantee, and in many non-government guaranteed business loans the secured lender also requires the owner of the businesses to be personally liable upon potential default. The CARES Act significantly removed the requirement for personal guarantees on SBA guaranteed loans under $200,000 (SBA loans over $200,000 must have been guaranteed by any owner having a 20 percent or greater interest). The CARES Act also removed certain SBA requirements that the borrower not be able to secure credit elsewhere. It also revised the requirement that the borrower have been in business for at least one year provided it was in operation on January 31, 2020.6 That temporary relief to small business owners has expired and is no longer in effect.
When a business fails and cannot pay its secured debts, lenders routinely pursue personal guarantees, often forcing business owners into financial distress because they, in most instances, have already used their non-exempt personal assets in an effort to keep their business solvent and operating.
The COVID-19 pandemic provided a clear example of how unforeseen circumstances can put unsustainable financial pressure on small businesses. During the early pandemic period in March 2020, 41.3% of businesses reported they were temporarily closed because of COVID-19, and 1.8% reported they were permanently closed. The pandemic particularly affected certain sectors. Accommodation & Food Services, Arts, Entertainment & Recreation, Education, and Other Services were among the most affected sectors.7 The COVID-19 crisis demonstrates how immediately just one unforeseen external strong negative factor, or “shock”, can threaten the solvency of even well-established businesses.8
According to Wave, 57% of small business owners have less than $5,000 saved for emergencies.9 Financial experts consistently recommend that businesses maintain three to six months of operating expenses in cash reserves, yet the majority of small businesses fall far short of this target. Without adequate liquid assets or retained earnings, small businesses cannot sustain an unforeseen slowdown in demand for goods or services, or an unforeseen increase in rent, transportation costs, restocking, insurance or labor. This lack of financial cushion means that even temporary disruptions – whether from a pandemic, natural disaster, economic downturn, pressure from competitors, or the loss of a major client – can irreversibly slide even a well-run business into insolvency.
Given the realities of the marketplace, careful risk assessment is crucial for any business transaction. Risk cannot be determined with scientific certainty. A range of risk can be approximated, however by identifying a range of foreseeable risks based on case specific data, available economic indicators city, state and nationwide, industry trends, and careful analysis of specific industry of concern. Generative AI can be a valuable tool in risk assessment – not as the only source of information, but as a predictive resource to consider when starting a new business or analyzing the purchase of an existing business. AI can help determine whether claims made by a seller of an existing business regarding sales volume, costs of goods sold, and fixed and variable costs are reasonable and credible – by comparing them against industry benchmarks, historical data, foot and Internet traffic and sales, the cost of goods sold and other variables.
Using AI comes with its own risks. Confidential and trade secret information can be leaked into the AI database, which should be avoided by using a closed system that prevents the client’s data from becoming part of the larger data pool that Gen AI uses to produce its generated result. Another safeguard against this possibility is to avoid identifying the name of the client or potential seller, any identifying information for all parties in the AI training or data feed.
For transactions of significant scale, the use of a qualified data scientist or engineer to work with the client and client’s counsel will produce the best results. Persons working with Gen AI, an attorney or data scientist, “train” the AI by providing specific information relevant to the business issue at hand. For example, when evaluating a business opportunity, relevant training data might include whether demand for the service or product is sufficient in the target area using publicly available sales figures, population and income statistics in the relevant geographic areas, the number of business bankruptcies in that industry and location, the actual range of costs for goods and services that will be sold, local market conditions including whether there is an inordinate number of business failures in the immediate market area, average commercial rents in the area of the target business, competitive landscape and market saturation, and regulatory requirements, compliance costs, and projections of industry demand.
An attorney versed in Generative AI can assist in this risk assessment process. One of the advantages of Gen AI is that platforms can be selected that take written instructions through natural language prompts, so it is not necessary for the person assisting with Gen AI input to have skills in Python programming or other coding languages. Gathering relevant information regarding the potential transaction is especially important. It stands to reason that the more relevant and accurate information that Gen AI is trained with, the more useful any risk (and potential profit) prediction will be.
To conclude, it is useful to suggest a very old phrase: “Caveat Emptor” or “Buyer Beware.” First coined in 1523, this principle remains as applicable today – and tomorrow – as it was five centuries ago. The use of Gen AI to assist the buyer of a going concern or potential start-up owner is a new way to accomplish a timeless goal, reducing business risk in order to prevent loss and maximize profit.
Copyright@2025 William Paul Rossini
William P. Rossini is the owner and manager of Rossini Law, a law firm in Dallas, Texas. Rossini Law focuses on business litigation and small business bankruptcy. He holds a BA in Economics and in Political Science with Departmental Distinction in Political Science from Southern Methodist University, cum laude and a JD from SMU (Dedman) Law School.
ENDNOTES
1. U.S. Small Business Administration, Table of Size Standards, CFR, Ch.1, part 121, link available at https://www.sba.gov/document/support-table-size-standards.
2. For IRS references to “small business” see IRS Publication 334, Tax Guide for Small Business, state that the SB/SE Tax Center serves “taxpayers who file Form 1040, Form 1040-SR, Schedules C, E, or F, or Form 2106, as well as small business taxpayers with assets under $10 million”. Another example of a specific tax code definition is the “small business corporation” definition in Internal Revenue Code Section 1361(b) for S-corporations. To qualify, a corporation must meet specific requirements, such as not having more than 100 shareholders. This definition, however, only applies to the S-corporation designation and is not a universal rule for all small businesses.
3. Subchapter V uses a single combined debt limit, $3,424,000., that includes both secured and unsecured debt together, rather than separate limits for each type. 11 U.S.C. § 1182(1). This section defines a small business “debtor” as a person engaged in commercial or business activities. The statute also specifically excludes: (1) any member of a group of affiliated debtors with aggregate debts greater than the debt limit, (2) corporations subject to SEC reporting requirements under sections 13 or 15(d) of the Securities Exchange Act of 1934, and (3) affiliates of such corporations. Only noncontingent, liquidated debts count toward the limit. This debt limit had been temporarily increased to $7.5 million during COVID, but this higher limit expired on June 21, 2024, reverting to $3,024,725. Public Law 116-136, Section 1113 (CARES Act).
4. SellersCommerce, United States Small Business Statistics (2025 Data), April 28, 2025, https://www.sellerscomminerce.com/blog/small-business-statistics/. From that web page: “There are 34,836,451 small businesses in the United States as of 2025, accounting for 99.9% of all businesses. The number of US small businesses has grown by 0.28% in the past year and over 9.7% from the start of this decade.”
5. https://www.uschamber.com/co/start/strategy/why-small-businesses-fail.
6. CARES Act, Section 1110, Emergency EIDL Grants.
7. Bartik, A., et al., The Impact of COVID-19 on Small Business Outcomes and Expectations, PNAS, July 10, 2020, available at https://www.pnas.org/doi/10.1073/pnas.2006991117; (“Across the sample, 41.3% of businesses reported that they were temporarily closed because of COVID-19. A far smaller number – 1.8% – reported that they were permanently closed because of the pandemic. By contrast, only 1.3% reported that they were temporarily closed for other reasons; 55.5% reported that they were still operational.”).
8. IMF [International Monetary Fund] Working Papers, COVID-19 and Small & Medium Sozed Enterprises (SME) Failures, September 25, 2020, https://www.elibrary.imf.org/view/journals/001/2020/207/article-A001-en.xml. (“We estimate a large increase in the failure rate of SMEs under COVID-19 of nearly 9 percentage points, absent government support. Accommodation & Food Services, Arts, Entertainment & Recreation, Education, and Other Services are among the most affected sectors.”).
9. https://www.score.org/resource/blog-post/how-much-cash-should-a-small-business-keep-reserve. In this article, SCORE [Service Corps of Retired Executives] acknowledges the widely cited guideline of having a cash reserve equivalent to three to six months of operating expenses. The post also offers important context for this recommendation, noting that it is not a “one-size-fits-all” approach and that the right amount depends on a business’s specific situation.