Your search thesis (lesson 3.3.1) says what you want in sentences. A buy box turns those sentences into filters you can actually run against public data. Most first-time buyers write a buy box as a wish list ("profitable, recurring revenue, great team") and then wait for brokers to send them something that fits. A data-layer buy box works the other way around. It uses only things you can observe from outside the company, before anyone hands you a financial statement, to build the universe of businesses worth scoring. The scoring itself comes in lesson 3.3.3. This lesson builds the filter that comes first.
Principle
Filter before you score. A rubric judges a candidate you already know about. A buy box finds the candidates. Keep the two separate. If you score everything that crosses your desk, you will spend your best hours on businesses that should never have reached you.
Use only what you can observe. At this stage you do not have tax returns or a profit and loss statement. You have public records: industry codes, state business filings, licenses, permits, court records, the company's own website and reviews. Every filter in your buy box must be something you can check from that kind of data, or it is not a filter yet.
A buy box has five layers. This structure comes from a working buy box and generalizes to any industry:
- Industry codes in tiers. The six-digit North American Industry Classification System (NAICS) codes that define your thesis, split into Tier 1 (core, screen first) and Tier 2 (adjacent, screen only after confirming the business really fits).
- Size proxies. Observable signals that a business is probably inside your size band, since revenue and earnings are not public.
- Geography. A universe gate (where you will buy at all) and priority clusters inside it (where you look first).
- Inclusion signals. Public clues that the owner may be nearing a transition.
- Exclusions. Hard exclusions that disqualify on sight, and flag exclusions that require a second look before you continue.
Aim the signals at transition. Many owners will need a buyer and have not planned for one. In a 2026 national survey of small business owners, 40 percent planned to retire within the next decade, 70 percent were in early-stage planning or had no formal succession plan, and 8 percent reported being fully prepared to transfer ownership (X16). Gallup's 2025 survey work found that most employer-business owners nearing retirement plan to sell or transfer their company (X2). Your inclusion signals are a way to find those owners from public records before a broker does.
Case
Coastal Comfort HVAC is the fictional company that runs through this curriculum. Here is the buy box that finds it, built from the thesis in lesson 3.3.1.
Anchors (from the thesis): owner-operated service businesses on Florida's Space and Treasure Coast, $2 million to $8 million in revenue, at least 10 years in business, owner near transition.
Layer 1, industry codes:
| NAICS | Title | Tier |
|---|---|---|
| 238220 | Plumbing, Heating, and Air-Conditioning Contractors | 1 |
| 238210 | Electrical Contractors and Other Wiring Installation Contractors | 1 |
| 561710 | Exterminating and Pest Control Services | 1 |
| 238290 | Other Building Equipment Contractors | 2 |
| 811310 | Commercial and Industrial Machinery and Equipment (except Automotive and Electronic) Repair and Maintenance | 2 |
| 561730 | Landscaping Services | 2 |
Roofing is left out on purpose: the thesis excludes weather-driven boom revenue.
Layer 2, size proxies: headcount, the number of service vehicles visible on the website and in photos, number of locations, number of licensed staff on the state licensing database, permit volume in county records, and the volume and pace of customer reviews.
Layer 3, geography: universe gate = the counties within an hour's drive of home. Priority clusters = the two or three metro areas inside that gate where you already have relationships.
Layer 4 and 5: the signals and exclusions in the steps below.
Now run it. A list built from state contractor licenses and business filings in those counties, filtered by these codes, surfaces a row like this:
| Field | Value | Source |
|---|---|---|
| Company | Coastal Comfort HVAC, LLC, Palm Bay, FL | State business registry |
| NAICS | 238220 | Industry code lookup |
| Size proxy | Estimated $5 million to $8 million revenue from fleet count and headcount | Website, photos, licensing records |
| Years in business | 17 (state filing, March 2009) | State business registry |
| Ownership | Single founder; spouse on administrative filings | State business registry |
| Age proxy | Founder's first contractor license issued 1994, which puts him in his late 50s | State licensing board |
| Successor check | No second licensed qualifier on record | State licensing board |
| Recent activity | A 41-unit townhome project with a local homebuilder in county permit records | County permits |
Every field came from public data. Nobody had to cooperate. And the last row already tells you what to test when the broker's memo later claims a record year: whether that jump was a trend or one project.
Steps
1. Pick your industry codes
Look up the NAICS codes for every industry in your thesis on the U.S. Census Bureau's NAICS site. Use six-digit codes; broader codes mix in businesses you do not want. Put the codes where your edge is strongest in Tier 1. Put adjacent codes in Tier 2. Expect businesses to be miscoded: a company may register under a code that does not match what it actually does, so read the website before you trust the code.
2. Derive your size band from your financing
Your size band is not a preference. It comes from your money:
- Start with the maximum total project from your Runway Sheet (lesson 3.1.1).
- Remember the SBA ceiling: under SOP 50 10 8.1 (Y25), the maximum for a standard 7(a) loan is $5,000,000.
- Remember the coverage floor: under SOP 50 10 8.1, an initial acquisition needs debt service coverage of at least 1.25 to 1, measured on historical earnings, not projections. Debt service coverage is the business's annual cash flow divided by its annual loan payments.
- Ask your lender for the annual payment on the loan your stack implies. Add the annual payments on any seller note that will pay while the SBA loan is outstanding. A seller note on full standby, with no payments for the life of the SBA loan, adds nothing. Multiply the total by 1.25. That is the minimum annual earnings a target must have produced, measured after a market wage for whoever does the owner's job (lesson 5.5.1 covers that adjustment).
- Convert earnings to revenue using a margin you have verified for your industry from real financial statements or trade data. Do not guess it. If you cannot verify a margin, use a wider revenue band and let the scoring rubric sort it out.
Check the band against a real structure. The Coastal Comfort deal that closes later in this curriculum is a $3,500,000 price inside a $4,200,000 total project. Its SBA loan is $3,030,000, under the $5,000,000 maximum. At an illustrative 10 percent variable rate over ten years (your lender quotes the real one), the loan costs about $481,000 a year and the one paying seller note about $43,800, so the earnings floor is about $655,000. Recurring earnings of about $920,000, the figure the quality of earnings review leaves standing, clear it. A business that size fits this buyer's band only because the Runway Sheet reaches a $4,200,000 project: $210,000 of the buyer's own cash, with committed investors supplying the other half of the injection.
3. Build your size proxies
Revenue is not public, so choose proxies you can observe and set a rough in-band range for each: headcount, vehicles, locations, licensed staff, square footage of the facility in county property records, permit volume, review volume. No single proxy is reliable. Two or three that agree are a signal.
Write down what is out of band too. A business with three employees is probably too small to carry an SBA loan and pay you. A business with multiple regional offices is probably too large.
4. Set the geography
Write the universe gate: the exact counties or the drive time you will accept. Then list priority clusters inside it, ranked. A business outside the gate is excluded. A business inside the gate but outside a cluster is not excluded; it simply sits lower in the queue.
5. List your inclusion signals
Each one nudges a business up the list. Adapt these to your industry:
- Founder's surname in the company name, or the founder visibly running it.
- Long tenure: the entity registered 10 or more years ago in the state registry, at the same address for most of that time.
- An age proxy: an early license date, a graduation year, or military service dates in a public bio that imply the owner is in their 60s or near it.
- No visible successor: a single license qualifier on record, no family member or manager in a leadership role.
- Family-run signals: a spouse on filings, "family-owned since" language on the website.
- One location, no parent company branding, no franchise.
- No recent announcements of new outside investors or an acquisition.
- Steady reputation: reviews that are good and answered by the owner.
None of these proves an owner wants to sell. Together they tell you whose door to knock on first.
6. List your exclusions
Hard exclusions (disqualify on sight):
- Owned by a private equity firm or already part of a roll-up.
- A public company, or a subsidiary of one ("a division of" branding).
- An active bankruptcy or receivership (search PACER, the federal courts' public records system).
- A business type the SBA does not lend to (13 CFR 120.110 lists the ineligible types (Y31), such as passive businesses and lending businesses).
- On the federal exclusions list at SAM.gov, if the business sells to the government.
- Outside your geographic gate.
- Anything your thesis excludes.
Flag exclusions (look twice before continuing):
- Liens found in the state's UCC (Uniform Commercial Code) filing search beyond normal equipment financing.
- Lawsuits in county or federal court records.
- A recent acquisition announcement or ownership change.
- A license that only the owner holds, with no one else in the company licensed to qualify the business after the sale. A license belongs to a person and does not transfer with the business; many states require a licensed qualifier for the company to keep contracting. Check your state board's rules.
- A Tier 2 code as the business's only line of work.
7. Build your public-database reference table
List every source you will use, what it shows, and how to search it. A starting set:
| Source | What it shows |
|---|---|
| Census NAICS lookup | Industry code definitions |
| Census County Business Patterns | Counts of businesses by industry code and county, to size your universe |
| Census Annual Business Survey | Owner characteristics by industry and geography; the 2018 survey (reference year 2017) includes how and when owners acquired their business (Y15) |
| State business registry (Secretary of State) | Formation date, officers, registered agent, status |
| State licensing boards | License holders, issue dates, qualifiers, discipline |
| County permits and property records | Project activity, facility size, ownership of the real estate |
| State UCC filing search | Liens on business assets |
| PACER and county court records | Bankruptcy and litigation |
| SAM.gov | Federal registration and the exclusions list, for businesses that sell to government |
| The company's website and reviews | Services, fleet, staff, history, owner's voice |
8. Run the build in order
- Pull every business in your Tier 1 codes inside the geographic gate.
- Drop the hard exclusions.
- Keep those whose size proxies fall in band.
- Tag the inclusion signals on each.
- Sort by geography priority, then by number of signals.
- Hand the result to the scoring rubric (lesson 3.3.3).
Refresh the list every quarter. Lesson 4.2.1 covers working the public data in more depth.
Deliverable
A Buy Box Specification, two pages, plus a list:
- The anchors from your thesis.
- Tier 1 and Tier 2 NAICS codes, with titles.
- Your size band and how you derived it from your financing, and your size proxies with in-band and out-of-band ranges.
- Your geographic gate and ranked priority clusters.
- Your inclusion signals and your hard and flag exclusions.
- Your public-database reference table.
- The first run of the list, in a spreadsheet with one row per business and one column per filter, ready for scoring.
This lesson is education, not legal or lending advice. Public records can be incomplete or out of date; verify anything that matters with the owner's documents and your advisors before relying on it.