Deal screening in the lower middle market works best in order of cost. Run the free desk checks first: mandate fit, ownership, whether the company still operates, and hard disqualifiers. Use paid data as a range, not a verdict. Save anything that needs a call for the short list that survives, and record why every company was cut.
The list arrives as a spreadsheet. It came from a data subscription, an association directory, a conference attendee export or all three, and it has four hundred rows. The partner wants twenty names worth calling by the end of the week. Somewhere in between is the associate, and the screen they run decides whether the next month of outreach goes to the right owners or to whoever had the best website.
This is a method for that screen. It is written for the person who runs it, and it assumes the mandate is already defined: sector, geography, a size band, and the ownership profile the fund wants. If the list itself is the problem, fix that first. A screen cannot rescue a universe built from the wrong source.
The principle: cheap checks first, expensive checks last
Every check on a target costs something. Reading a website costs a few minutes. Pulling a paid record costs a credit or a seat. A call to the owner costs the relationship, because you only get one first impression and a badly timed first call is hard to take back.
So the screen runs like a funnel with the cheapest filter at the top. Each pass removes companies for a reason you can write down, and nothing reaches the expensive stage until it has cleared everything cheaper. The mistake the screen exists to prevent is spending an expensive check on a company a free one would have removed.
The other rule: a cut needs a reason, and the reason gets recorded. A list that loses two hundred rows with no trail gets rebuilt next quarter from scratch, and the same two hundred companies come back.
Pass 1: does it actually match the mandate?
Start by checking the list against the mandate it claims to match, because source lists drift. A NAICS code in a database is whatever the company or a data vendor chose years ago. An HVAC distributor gets coded as a contractor; a specialty manufacturer gets coded as wholesale. Read what the company says it does on its own site, and cut anything that is adjacent rather than inside the sector.
Geography is the same. A headquarters address is not an operating footprint. A company registered in one state with every job site in another is in the second state for the purposes of an add-on thesis.
This pass is fast, and on a list built from broad codes it often removes more than any other.
Pass 2: who owns it?
Ownership is the check most screens skip and the one that saves the most wasted outreach. Four questions, all answerable from public sources:
- Is it already sponsor-backed? Search the company name against press releases and PE portfolio pages. A company acquired two years ago is not a target; it may be a competitor's platform.
- Is it owned by a strategic? Subsidiaries often keep their original brand and website for years. A small "a member of" line in the footer ends the conversation.
- Who are the officers? State business entity searches, run through each Secretary of State's office, list registered officers and formation dates. A company formed last year is not the founder-owned business the mandate describes.
- Who lends to it? UCC filings, also searchable at the state level, show secured parties. A blanket lien from a lender tells you something about the capital structure before anyone has said a word.
None of this needs a subscription, and it is the pass that turns a list of companies into a list of owners.
Pass 3: is it still operating, and is it roughly the right size?
A surprising share of any list built from a database is companies that have closed, merged or shrunk. Recent customer reviews, current job postings, an active permit history and a website that has been updated in the last year are all evidence the business is running.
Size is harder, because a founder-owned company does not publish revenue. Use proxies, and use several:
- Number of locations, trucks, bays or crews, where the sector has a physical unit of capacity.
- Job posting volume, which tracks whether the company is growing, not how big it is.
- Federal contract awards on USAspending.gov, for any company that sells to government, which also shows how concentrated that revenue is.
- Headcount estimates, read as a floor, for reasons covered below.
The output of this pass is a size band per company, not a number. Tag each survivor as clearly inside the band, clearly outside, or unclear. The unclear ones stay in; the clearly outside ones go.
The disqualifiers that outweigh every positive signal
Some facts end the evaluation no matter what else is true. They are worth checking explicitly, because a company that looks strong on every other dimension will carry the screen past them if nobody asks.
- Franchise ownership. Franchise agreements typically give the franchisor approval rights over any transfer. A franchisee can be a fine business and a hard acquisition.
- Customer concentration. A case study page that names one customer repeatedly, or a contract award history dominated by one agency, is a concentration flag you can see from the desk.
- A credential held personally by the owner. Where the business can only operate under a professional qualification the owner holds in their own name, the value may not transfer with the shares.
- Ownership restrictions. Some states limit who may own a medical, dental or professional practice. If the sector sits under those rules, check the state before the company.
- A recent transaction. A recap, a new partner or a change of control in the last few years usually means the owner has already had the conversation you are about to start.
A single disqualifier is enough. The instinct to keep a company because it scores well elsewhere is exactly what this section is for.
Why a score built on public data mis-ranks LMM targets
Most screening tools offer a score, and the score is useful for sorting. It is dangerous for cutting, and the reason is structural.
The inputs to a public-data score are the signals a machine can read: web traffic, hiring activity, social presence, review volume, headcount from professional networks. Every one of them measures how visible a company is online. In the lower middle market, visibility and quality are loosely related at best. Many of the businesses a sponsor most wants to own are founder-run, referral-driven and decades old. Their customers find them through relationships, so they have never invested in a website. Their workforce is field technicians and drivers who do not keep professional profiles, so modeled headcount undercounts them. Any revenue estimate built from that headcount inherits the same error.
The result is a score that ranks companies by digital footprint. The polished marketing firm with the growth-stage hiring page floats to the top. The fifty-year-old contractor with the one-page site sinks to the bottom, and it may be the better target on every measure the partner cares about.
The practical fix is simple. Use the score to order companies within a tier after the desk checks have run. Never use it to decide which companies get desk checks in the first place.
Pass 4: the checks that need a conversation
What survives the first three passes and the disqualifiers is the short list, and it is the only part of the list that should ever cost a relationship. Owner intent, succession, real financials and timing are invisible from the desk, and they decide whether a company that looks right is actually available.
This is also the point where the stage of the deal changes the next step. An Off-Market company needs an introduction and a reason to talk. A Pre-Market company may already have an advisor preparing materials. An On-Market deal needs a different first question entirely: whether it is still available at all, which is covered in how to tell whether a deal is real. Once a conversation produces real numbers, the work moves from screening to preliminary diligence, which is a separate discipline with its own checklist.
What to hand the partner
The short list is not twenty company names. It is twenty names with a line each: why this company is in, what the ownership check found, the size band and how it was estimated, and the one open question a call would answer. Alongside it goes the cut log, grouped by reason, so the partner can see that the universe was worked rather than skimmed.
That cut log is the asset most screens throw away. Rejections, recorded with a reason, are how a mandate gets sharper. They show which sectors produce nothing but franchisees, which geographies are already consolidated, and which source lists are mostly noise. Next quarter's screen starts from that knowledge instead of from a fresh export.
It is also the design behind the approve-or-reject queue on the OmniSource platform: matched companies land in Targets to Review, and a rejection with a reason teaches the mandate as much as an approval does. Whether the screen runs in a spreadsheet or a product, the rule is the same. Every company that leaves the list leaves with a reason, and the ones that stay have earned the call.