I've read over a hundred buy box viability reports. The single most common reason a search stalls has nothing to do with outreach, capital, or luck.
The list was bad.
Bad lists produce bad response rates, and bad response rates convince people that off-market sourcing doesn't work. It works. But a list is a machine, and if you build it wrong it produces nothing no matter how good your email is.
Here's how I build one.
Step one: write the buy box before you touch data
Not after. Before.
A buy box is not "profitable business, $1M to $3M EBITDA, anywhere in the Southeast." That's a filter, not a thesis. It doesn't tell you which companies to put on a list and it doesn't tell you why you'd win.
A usable buy box answers five things:
- What industry, specifically. Not "home services." Commercial HVAC service and maintenance. Not "manufacturing." Precision machining for aerospace suppliers. The narrower you go, the better your outreach reads, because you can speak the language.
- What geography, and why. Where you'll live, where you'll operate, or where the industry actually clusters. Pick for a reason you can say out loud.
- What size, in the metric that matters. Revenue is easy to find and tells you little. EBITDA or SDE is what you're buying. Set a band and know why the bottom and the top are where they are.
- What the business has to look like to work. Recurring or repeat revenue. A manager in place. Customer concentration under some number. Licensing you can hold or hire. These are the make-or-break attributes.
- Why you win it. What do you bring that the next buyer doesn't. Operating background, an industry relationship, a specific plan for the first year. If you can't answer this, you're going to keep losing to people who can.
Write it down in a page. If you can't fit it on a page, you don't have one yet.
Step two: size the market before you build the list
This is the step everyone skips, and it's the one that saves you a year.
Before you spend money on data, count how many companies could possibly match. Take your industry, your geography, and your size band, and get to a number.
Then ask whether that number can produce a deal.
Every step of the funnel is a multiplier under one. A fraction of the list is reachable. A fraction of those reply. A fraction of those have a conversation worth having. A fraction of those transact on a timeline that fits your life.
If your buy box produces forty companies nationwide, you don't have a search. You have a hobby. Widen the geography, widen the size band, or add an adjacent industry. If it produces eleven thousand companies, you're too broad to write anything specific and your response rate will show it.
You want a universe big enough to survive the funnel and narrow enough to speak to. Find that number first.
Step three: build the universe
There is no single database with the answer. You assemble it.
Start with classification data. NAICS and SIC codes at the six-digit level, run against a business database, filtered by state, metro, or county. This gets you a raw universe fast. It will also be dirty. Codes are self-reported and often wrong.
Layer in trade sources. Industry association member directories, licensing boards, certification lists, franchise rosters, trade show exhibitor lists. These are usually more accurate than classification data because somebody had to actually qualify to be on them. A state contractor licensing database will tell you more about who actually does commercial electrical work than any NAICS filter.
Add local sources. Chamber directories, city business licenses, Google Maps scrapes for a defined radius. Tedious, but for a geography-constrained search they fill gaps nothing else does.
Cross-reference. A company that shows up in three sources exists. A company that shows up in one might be a shell, a duplicate, or a business that closed in 2019.
Step four: enrich and filter down
Now cut the universe to a target list. This is where the work is.
For each company you want: legal name and DBA, address, phone, website, estimated employee count, estimated revenue, year founded, owner name, owner age if you can get it, and how long the owner has held it.
Then filter hard on the things that actually predict a transaction:
- Years in business. Twenty-plus years usually means an established base and an owner in a different life stage than he was at the start.
- Owner tenure and age. The strongest signal on any list. An owner who has run the same company for twenty-five years is closer to a conversation than one who bought it last year.
- Size sanity check. Employee count against your revenue band. Fifty employees in a service business is not a $900K revenue company, regardless of what the database says.
- Signs of life. A website updated in the last two years, current reviews, active hiring.
- Ownership structure. Kill anything already owned by a PE platform, a strategic, or a public parent. They're not your conversation.
Verify the owner name by hand on the ones that survive. Every one. A message addressed to the wrong person tells the reader exactly how much care went into it.
Step five: score and sequence
Don't email eight hundred companies in the same week.
Score each name against your buy box, sort into tiers, and work the top tier first with your best writing and your most careful research. Learn from what comes back. Fix the message. Then go to tier two.
A list isn't a static asset. It's the thing you keep correcting as the market answers you.
What good looks like
A good target list is small enough that you know why each name is on it, and specific enough that you could write a sentence about any given company without looking it up.
Build that, and outreach becomes a volume question instead of a hope.