Buying a Business, Explained

From 0 to 0.5: what I've learned about paying for a small business

A first-time buyer's notes on the capital stack, the new SBA rules, what investors expect, and why the price matters most.

The debt side of buying a small business makes sense to me: a bank lends against a real company with real cash flow. The hard part is the equity, the cash that has to sit under the loan. And often the cheapest equity is a better price.

Below: where I am, what I did, five lessons with numbers, and two industries I'm testing. Deal numbers come from an illustrative company, not a real deal. The visual explainer draws it all, with a narrated video of about 2 minutes.

Situation

I'm an operator, and I'm studying AI engineering. I'm considering buying a small business and growing its value with both. That's called value creation: making a company worth more after you buy it.

I'd start with the slow, manual work: quoting jobs, scheduling crews, and the back office (invoices, billing, paperwork). My bet is that if those run faster, the same team can serve more customers. It's a bet, not a promise.

A bank looks at a real company with real assets and real cash flow, and says yes or no on those facts. That part makes sense to me. Raising money from investors is a different skill, and it's my weak spot. That's why I'm writing this: to learn in public, and to meet people who have done it.

Task and action

The question: how do you even do this, and where does the money come from?

Here's what I did:

The illustrative company earns $1.00M of EBITDA (earnings before interest, taxes, depreciation and amortization, a rough measure of yearly operating profit) on $5.00M of revenue. After cash taxes (10% of EBITDA in the model) and upkeep spending (2% of revenue), about $800K a year is left to pay the loans. Closing costs and a cash cushion add $250K on top of the price. The bank loan is assumed at 9.75% over 10 years.

A capital stack is the layers of money that together pay for the project. Here there are up to four:

Lenders check the DSCR (debt service coverage ratio): cash available for loan payments divided by the loan payments. Here they want at least 1.25x: cash covers the payments with room to spare. Prices are shown as a multiple of EBITDA, so 4.0x means $4.00M.

Case Price Project cost Buyer cash (5%) Standby seller note (5%) Investors SBA loan Yearly SBA payment DSCR
A: 4.0x $4.00M $4.25M $212K $212K None $3.83M $600K 1.33x
B: 5.0x $5.00M $5.25M $262K $262K None $4.72M $741K 1.08x
C: 5.0x + investors $5.00M $5.25M $262K $262K $647K $4.08M $640K 1.25x
D: 4.28x $4.28M $4.53M $227K $227K None $4.08M $640K 1.25x

Case A passes. In case B the seller asks for 5.0x, the payment jumps to $741K, and the DSCR falls to 1.08x, below what lenders want. Cases C and D are the two fixes, and the heart of what I learned. The visual explainer builds this stack layer by layer.

Result

I'm still early. But after the discussions, the bankers and the math, five lessons stand out.

Five lessons

1. Your own cash sets your ceiling. To buy a whole company with an SBA loan, the buyer side must put in at least 10% of the project cost, called the equity injection (Starfield & Smith). The SBA updated its rulebook for these loans again (SOP 50 10 8.1); one secondary source dates the change to October 1, 2026 (Angel Investors Network). As one SBA loan broker reads the new rules, at least half of that 10%, so 5% of the project, must be the buyer's own money: cash, or a personal loan repaid from income outside the business. A seller note on full standby can cover the other half at most (Whiteford), and on the broker's reading, small investors (each under 20%, with no control) can count toward that half too. Divide your cash by 5% for a rough ceiling: in the worked example, $250K of a buyer's own cash supports a project of about $5.00M. Confirm with your lender.

2. Investor money works best on top of the 10%, not inside it. As one SBA loan broker reads the new rules, if investor money is used to meet the 10%, those investors get only tax distributions (cash to cover the tax on their share of profit) until the bank loan is paid off. Money put in above the 10% can receive regular payouts, within the lender's terms. One public practitioner page also describes a limit on investor payouts (Angel Investors Network). That's why case C puts the investors' $647K above the 10%. The buyer's $262K and the seller's $262K standby note fill the 10%, so the investors can still be paid along the way. Confirm with your lender.

3. Know what investors expect before you ask. In discussions in the Searchfunder community, investors came back to the same four terms:

A public guide for self-funded searchers lands in the same zone: a 10% to 15% preferred return, a 2.0x to 2.5x step-up, and at least a 30% to 35% IRR (Pioneer buy-side brief). In my example, $647K at a 2.0x step-up on a $5.00M price buys 25.9% of the company.

4. Raise from people you know first, and screen every investor. One first-time buyer in the community listed every contact he had and asked each person who said no to introduce someone else. A big part of what he raised came through those introductions. His seller walked away before signing, so it's one story. Other buyers posted one clear public ask and drew replies from investors and funds. Members also share lists of investors in small acquisitions; match each one's check size, industry and region. On timing, a searcher who was raising money said most investors would rather be asked after a signed LOI (letter of intent: the written offer that sets price and main terms). So build the list now and ask after the LOI. Screen early: since March 1, 2026, every owner of an SBA borrower, direct or indirect, must be a US citizen or national living in the US (Dickinson Wright). Keep each investor under 20% with no control. In a full purchase, anyone owning 20% or more must personally guarantee the loan (Whiteford), and lenders may see board seats or vetoes as inconsistent with the buyer being the one who runs the company (Acquisition Stars). Confirm with your lender.

5. A lower price often beats raising more money. Cases C and D end in the same place for the bank: a $4.08M SBA loan, a $640K yearly payment and a 1.25x DSCR. In case C the buyer pays $5.00M, puts in $262K, brings in $647K from investors and owns 74.1%. In case D the buyer pays $4.28M, puts in $227K, raises nothing and owns 100%. Sell after 5 years at the same multiple with flat earnings, and the buyer gets back $1.53M in case C against $2.25M in case D. The investors in case C get $1.40M back on their $647K, a 23.8% IRR, short of the 30% to 35% they target. A seller won't always say yes, but it's worth asking before you raise.

Thesis 1: industrial electrical contractors in New England

I'm exploring two very different industries. The first is industrial electrical contractors: the firms that keep critical power running for hospitals, labs, plants and data centers.

Nationally, the headline is data centers. The US Bureau of Labor Statistics expects electrician jobs to rise 9% between 2025 and 2035, much faster than the average job, with about 72,700 openings each year. It counts AI and data centers among the reasons (BLS).

I went in believing contracts were signed out to 2035. That's only partly right. Long-dated commitments exist, mostly on the power side. Meta's January 2026 nuclear deals support up to 6.6 GW (gigawatts, a unit of power) of clean energy by 2035 and include 20-year purchase agreements (Meta). AEP Ohio's rate rules for data centers (its tariff) allow a first contract term of up to 12 years (AEP Ohio). The contractors doing the work see a shorter window. Quanta Services reports a record $63.7B backlog (signed work not yet done) and expects about $38.4B of it to turn into revenue within 12 months (Quanta).

I know the topic feels taboo, partly because the AI boom may not last: the Bank of England has warned that lenders who funded data centers with debt could take big losses if AI markets drop sharply (Network World), and one research firm expected 30% to 50% of the large data centers planned for 2026 to open late (Latitude Media).

New England is a different market. It already has about 100 data centers, according to the Data Center Coalition, an industry group quoted by WBUR, with the largest cluster around Greater Boston. Most are colocation sites, buildings that rent space, power and cooling to many companies, and they are small by national standards: some use under 1 megawatt (MW), the larger ones up to about 40 MW, and the region has no large AI training campuses yet. New giants are rare too: ISO New England's 2026 forecast counts only two proposed large-load projects (huge new power users), 285 MW at most (ISO Newswire). The steadier growth story is electrification: ISO New England expects the region's electricity use to grow about 9% over 10 years, mostly as heating and transportation switch to electric (ISO New England).

So my target is narrower: an industrial electrical contractor that already maintains critical power for hospitals, labs and plants. Think backup generators, UPS systems (batteries that keep the power on until the generators start), transfer switches and switchgear testing. That work pays with or without an AI boom. The region's existing data centers depend on the same equipment, so they are upside, not the bet. The skills overlap, but data centers raise the bar on redundancy, maintenance without downtime and vendor qualifications, so I'd plan to add someone with data-center operations experience. What I'm still testing: which contractors hold these maintenance contracts today, how much of their revenue repeats every year, and whether a small one wants to sell.

Thesis 2: decorated apparel in New England

The second is decorated apparel: screen printing and embroidery, which put a logo on shirts, hats and uniforms. Businesses buy it, and so do schools, colleges and sports teams.

The wider market is big. US promotional products distributors sold $27.1B in 2025, up 1.3% (PPAI). And New England has a lot of potential customers: 4,570 public K-12 schools in 2022-23 (NCES), about 1,530 private K-12 schools in fall 2021 (NCES), 224 degree-granting colleges and universities in 2022-23 (NCES), and 466,313 high-school sports participation entries in 2025-26, where a two-sport athlete counts twice (NFHS).

Some of this work comes on multi-year contracts. Salem State University in Massachusetts put its athletic apparel, uniforms and supplies out to bid on a 5-year contract, plus two optional 1-year extensions. Every bidder had to offer both screen printing and embroidery (Salem State RFP). The plan would be to bid on contracts like that. One honest gap: so far, the multi-year examples I've found are colleges. I haven't found a K-12 district buying apparel on a multi-year term yet.

That's where I am. Thank you to the Searchfunder community, and to the bankers who took my calls, for helping me go from 0 to 0.5, with a long way still to go. If you've raised equity for a self-funded deal, I'd like to compare notes: what investors asked for, where their money sat, and what you'd change. You can find me in the Searchfunder community.

Sources