Searcher OS

A listing came across my feed last month at $2M asking, $550K SDE. That's a 3.6x multiple. Reasonable, on paper. Then I read the add-back schedule. A "one-time" equipment repair of $45K that had happened twice in three years. $24K of "client entertainment" with no client and no documentation. Strip those two out and the real SDE is $481K. Same asking price, now a 4.2x multiple. The story moved a full half-turn without anyone touching the headline number. This is the part of valuation that actually decides whether you overpay. The multiple is the easy math. The earnings figure underneath it is where sellers and brokers quietly write fiction. First, SDE vs EBITDA, because people use them like synonyms and they aren't. SDE (Seller's Discretionary Earnings) adds back the owner's full comp and is the right metric when you'll be the one showing up to work. EBITDA assumes a management team is already in place. For owner-operated deals under $5M, your SBA lender underwrites on SDE. So that's the number that has to be honest. Here are 3 add-backs I never take at face value: The "one-time" expense that recurs. If equipment breaks every couple of years, the most recent repair isn't a one-time hit. It's a maintenance cycle wearing a costume. The artificially low owner salary. If the owner pays themselves $60K to run a business that needs a $120K operator, the SDE is overstated by the $60K gap. You'll feel it the first month you can't replace yourself for free. Anything that won't tie to the tax return. The P&L is the broker's document. The tax return is what the seller swore to under penalty of perjury. When they diverge, the answer to "which is real?" tells you the whole culture of the financials. The test I run on any add-back: would a reasonable person running this business still incur this cost? If yes, it stays. If it's tied to the current owner's lifestyle, it's suspect. Get the earnings figure right and the rest of the deal math follows. Get it wrong and you're structuring debt on a foundation someone else made up. I built Searcher OS partly so I'd stop re-deriving this by hand on every deal. It runs the multiple and the DSCR as you screen, before you've sunk two hours into a CIM. There's a 14-day free trial if you want to pressure-test a few listings against the real number instead of the stated one.

The best feature I shipped this year came from a searcher complaining about something boring: he was losing 3 hours a week chasing brokers for CIMs they'd promised and never sent. That's most of how the roadmap actually gets written. I watch where the search itself leaks time, then sand down that one spot. The CIM-chasing complaint turned into Marcus. Marcus is an AI agent I built that handles the broker back-and-forth: it submits NDA requests, tracks which CIMs have actually landed, and runs the follow-up so the searcher doesn't have to babysit an inbox. The job it replaced was real and tedious and nobody enjoyed doing it. That's usually the sign something's worth building. I run Searcher OS solo, which forces a kind of honesty about what to build. I don't have a team to throw at a feature just because it would demo well. So the filter is simple: does this give a searcher back time they're currently spending on something a machine should be doing? Most of what I ship comes from three places. A specific email from a frustrated user. A thing that wasted my own time when I was running my search. Or a pattern I notice across the pipeline data, like a step where deals consistently go cold. I'll be honest that not everything I ship is a headline. A lot of it is unglamorous: a filter that finally works the way you'd expect, a screen that loads faster, a number that was off by a rounding error. The stuff that doesn't make a good announcement is often the stuff that makes the tool feel less like a chore. I think the searchers building with AI right now have an edge here that's easy to miss. The gap between "I wish this existed" and "this exists now" has collapsed. I'm not an engineer (I wrote almost none of the code by hand), and I still ship faster than the search tooling I was using two years ago ever did. If there's a part of your search that's quietly eating your week, that's exactly the kind of thing I want to hear about. It's how most of this got built. There's a free 14-day trial if you want to see what's in there now, including Marcus working the broker side for you. Come kick the tires and tell me what's missing.

It takes roughly 291 CIMs reviewed to land 1 closed acquisition. When I first saw that number I assumed it was a typo. It isn't. The full funnel comes from Athena Simpson, founder of AcquiMatch, who walked through it on stage in April. These are observed conversion rates from real searches, not a model someone built on a whiteboard. Here's the shape of it: Roughly 28,000 listings scanned. About 1,300 matched to a buy box. 650 NDAs signed. 291 CIMs reviewed in depth. 81 deals taken seriously. 9 to 14 LOIs sent. 2 to 3 accepted. 1 close. Sometimes zero. Every stage compresses the universe by about an order of magnitude. That's not because searchers are bad at filtering. It's because the SMB market is full of mismatches: wrong size, wrong geography, owner-dependent, declining revenue, a price the seller pulled out of the air. Most listings are a no before you sign anything. Here's the part that quietly kills searches. Most people underestimate the volume by about 10x and never notice. The typical first-timer commits 5 to 10 hours a week, signs 1 or 2 NDAs in a good week, reviews a handful of CIMs a month, and fires off an LOI every couple of months when something looks promising. Feels like real effort. Run that pace against the funnel and the math says you close in 4 to 6 years. Most people quit first. The searchers who actually close are running about 10x that. 7 to 9 NDAs a week. 16 CIMs a month. They didn't get there on willpower. They built a system: sourcing that runs on its own, a pipeline that remembers everything, templates written once and reused 600 times, a CIM review that takes 2 hours instead of 2 days. The honest gut-check is at the very top. If you're not scanning 200+ matched listings a week, every later stage is starved for input and no amount of discipline downstream fixes it. Sourcing is the bottleneck for most people, and it's the one they notice last. That top-of-funnel bottleneck is the one The Prospect chips at. More matched listings landing in your inbox each week, so raw volume isn't the thing that quietly ends your search. It's free to get. Credit to Athena Simpson and AcquiMatch for the numbers.

The best deals get a phone call before they ever hit BizBuySell. The question every searcher should be asking: how do I become the person who gets that call? Brokers are gatekeepers. They represent the seller, they run the process, and they decide which buyers hear about the good one first. Most buyers treat them like a search engine, query in, listing out. That's the mistake. The buyers who get the early call are the ones the broker believes will actually close. That belief gets built through a few unglamorous habits. Start with proof you can finance a deal. A pre-qualification letter from an SBA lender is the strongest signal you can send. It says you've already sat through a lender's process and you know what you can actually buy. Get one before you start requesting CIMs, not after. Be specific about your buy box. "Looking for service businesses" is useless to a broker. "B2B services, $1M to $3M asking, SDE above $300K, Southeast, under 20 employees" is something they can act on. The clearer your criteria, the more accurately they filter for you over time. Respond same day. If a broker sends you a deal and you surface 4 days later, you've been quietly deprioritized. Quick, thoughtful replies make you memorable in a way that's almost unfair given how low the bar is. And close the feedback loop when you pass. Most buyers go silent on a no. Send the one-line reason instead: "revenue trend was declining, didn't clear our SDE floor." That tells the broker what you really want, beyond what you said you wanted, and they calibrate. The compounding move is to introduce yourself to 15 to 20 brokers in your industries and geographies, then check back every 6 to 8 weeks. Don't ask for anything. Just stay present. A broker you met in month one might send you the right deal in month 18, but only if you're still in their head. That's the catch. Once you're working 20 or 30 broker relationships, memory stops being a system. You need to know who sent what, when you last touched base, and what you told each of them. I track every broker relationship in Searcher OS next to the deal pipeline, so when a CIM lands I can see our whole history in one click instead of digging through email. There's a free 14-day trial if you want to stop running your network out of your inbox.

4 states hold 40.9% of every small-business listing in the US: Florida, California, Texas, and New York. That one number quietly decides how hard your search is going to be. I pulled this from our own scraper data. We track 2,473 active US broker firms and roughly 42,878 listings, and the geographic clumping is hard to miss. The deals aren't spread evenly across the map. They pool where the population and the businesses are, which is mostly the Sun Belt and a few big coastal metros. So here's the uncomfortable part. Your buy box has a geography baked into it, and you probably set it for reasons that had nothing to do with deal supply. You want to stay near family. You like a certain climate. You know an industry that happens to cluster somewhere thin. All reasonable. All also a tax on your top of funnel. If your search is anchored to Florida, Texas, or California, you have more shots on goal than you can review. Your whole job becomes filtering. If your buy box sits in a thinner state (and plenty of good ones are thin) you've got a different job. The deals exist, there are just fewer of them surfacing through brokers each month, so you can't run a passive search and wait for inventory to come to you. A couple of things that help if you're in a thin market. Widen the radius before you widen the criteria. A 3-hour drive often doubles your listing count without changing the kind of business you're buying. Lean harder on relationships, because in a thin market the off-market deal matters more. The broker who calls you first is worth more where public inventory is scarce. And watch the whole country's flow, not just your corner of it, so you actually know when something rare shows up in your zone. That last one is why I built The Prospect. If your buy box sits in a thin state, it's a low-effort way to catch the good ones the week they surface, anywhere in the country, instead of hearing about them a month too late. It's free.

95% of small-business listings sit under the $5M SBA loan ceiling. So price almost never kills the deal. The eligibility rules do, and most buyers find out the expensive way. I see people fall in love with a business, sign the LOI, then watch the lender quietly walk because something in the structure was never going to fly. Worth knowing what trips the wire before you spend 90 days on it. On the business side, the SBA screens out a short list. Prohibited industries are the cleanest one: gambling, lending and investment firms, real estate speculation, multi-level marketing, life insurance companies. The list is narrower than people expect (restaurants, HVAC, distribution, healthcare practices all qualify), but if your target is on it, there's no appeal. It's a hard no. The other business-side traps: it has to be an existing operating business with a real track record, not an idea with a revenue projection. Lenders underwrite the trailing earnings, not the seller's "you could double this." And the deal has to result in a genuine change of ownership, at least 51%. Minority buy-ins don't qualify. On the buyer side, a few things disqualify outright. A FICO below 660 (most lenders want 680+). A bankruptcy in the last 3 years. Any unpaid federal debt, including old student loans, which flags you in a database called CAIVRS and stops the loan cold until you clear it. You also have to be a citizen or permanent resident and plan to actually run the thing day to day. The SBA doesn't finance absentee ownership. Then there's the number that decides everything: DSCR. Debt service coverage, your SDE divided by annual debt payments. The floor is 1.25x, and you cannot talk a lender past a failing one. If the business doesn't throw off 25% more cash than the loan costs, the deal is dead at the asking price. The fix is structural: lower price, more down, or a bigger seller note. That last one is where I'd spend your screening energy. Price you can negotiate. A 1.10x DSCR you cannot. The fastest way to know if a deal clears is to run it before you ever call a lender. The free SBA calculator inside Searcher OS shows you DSCR, the monthly payment, and total cash to close in under a minute. There's a 14-day trial if you want to pressure-test your current shortlist against the math.

By year 3, the two paths to buying a business look almost nothing alike, and most people only realize it after they've already picked one. Here's the fork. A traditional search fund means raising $400K to $600K from 10 to 20 investors, searching full-time for 18 to 24 months, then handing those investors 70 to 80% of the equity when you close. You run a $5M to $30M business as CEO and keep maybe 20 to 30% after it vests. The upside lands in a single exit, 7 to 10 years out. Self-funded search means putting in $150K to $300K of your own capital, financing the rest with an SBA 7(a) loan, and owning 100% of a $800K to $5M business the day you close. Cash flow starts immediately. The downside is yours too: that loan is personally guaranteed. So by year 3, here's the actual difference. The search fund founder owns a slice of a bigger thing and answers to a board. Investor approval on the deal, quarterly reporting, committee input on big moves. Some people find that structure steadying. Some find it a leash. The self-funded buyer owns the whole smaller thing and answers to no one but the lender (who mostly cares that you keep clearing debt service). I left corporate in August 2025 to run a self-funded search for exactly that reason. I have capital, I want cash flow, and I want to own what I buy. That's not a knock on the fund model. For a 28-year-old out of HBS with the investor network and no family to support, the math leans the other way. The stipend covers rent, the network opens doors, and 25% of a big exit is real money. The honest decision lens: do you want maximum terminal value (fund) or cash flow plus control starting now (self-funded)? Those are different bets, not better and worse ones. I'd just say this. Whichever path you pick, the screening discipline is identical. You still have to triage hundreds of listings to find the one that pencils, and that's the part that quietly eats a search. I built Searcher OS to handle that part: one feed of broker listings filtered to your buy box, a pipeline that tells you what's stuck, the SBA math in one place. There's a free 14-day trial if you want to run your next 20 deals through it.

17 firms list half the US broker market. Not 17%. 17 firms. I pulled this from 90 days of Searcher OS listings data, and the concentration surprised even me. There are 2,473 active business broker firms in the US. You'd think deal flow would be spread thin across all of them. It isn't. Just 17 of those firms carry 50% of all listings. Stretch it to 254 firms and you've covered 80%. The other 2,200-odd firms are fighting over the last fifth. It gets more lopsided. Only 4 firms (Sunbelt, Murphy, Transworld, First Choice) operate in all 50 states, and those 4 alone carry 24% of the entire pipeline. So here's the thing I keep coming back to. Where you point your search matters more than how hard you look. Most searchers crowd the same handful of national brands, because that's where the volume is and that's what shows up first. Which means that's also where the competition is thickest. Every funded searcher and PE group is refreshing the same Sunbelt feed you are. The long tail is the interesting part. Those 2,200 smaller regional brokers list fewer deals, but the deals are fresher and the buyer pool around them is thinner. These are the off-market Outposts, the listings that don't get 40 NDA requests in the first week. I'd tend to think that's where a patient self-funded buyer actually has an edge. (And 95% of all these listings sit under the $5M SBA 7(a) ceiling. The market is built for buyers like us, even if the volume hides in the wrong places.) The hard part is that watching the long tail means tracking hundreds of small brokers, which nobody wants to do by hand. So I let software do it, and the best of what surfaces goes out in The Prospect each week, weighted toward the quiet brokers most buyers never check. It's free. (If you just want to see the firms themselves, the free broker directory inside Searcher OS maps all 2,473.)

A broker once told me he sends his best off-market deals to maybe 5 buyers before he ever posts a listing. You want to be one of those 5. Here's the part searchers underrate: brokers gatekeep. They sit on deal flow and steer it toward buyers who are responsive, organized, and easy to deal with. You're not competing with the market. You're competing with search funds, PE groups, and every other self-funded searcher for the same broker's attention. Good news is, most of them make it easy to stand out by being slow and disorganized. So do the opposite. Respond within 24 hours. Always. If there's an NDA, sign it today, not next week when you've "had a chance to look it over." Speed reads as seriousness, and brokers notice who moves. Have your About Me ready before you need it. Who you are, your background, how you're funding the deal, why this industry. A broker who can size you up in one paragraph is a broker who keeps talking to you. (Searcher OS lets you save this in your profile so it auto-fills on NDA requests, which sounds small until you've typed the same bio for the 30th time.) Ask smart questions early. Don't ask "what's the EBITDA," it's sitting right there in the teaser, and now you've told him you didn't read it. Ask about customer concentration. Ask how dependent the business is on the owner. Those questions signal you actually know what kills a deal, and that you're worth his time. Kill deals fast, and tell him why. "The financials didn't fit my criteria" is a complete and respectable answer. It costs you nothing and it teaches him your buy box. He'll remember the buyer who passed cleanly and send the next one that actually fits, instead of the buyer who went quiet for three weeks. None of this is clever. It's just doing the obvious things, consistently, when your competition won't. The search is a relationship game. Play it well. If you want to stop retyping your bio on every NDA, that's one of the first things I'd set up. Searcher OS has a free 14-day trial; build your profile and About Me, and it travels with you to every broker after that.

In his first month, Marcus made up a revenue number on a deal and missed a red flag a first-week intern would've caught. Marcus is a feature inside Searcher OS. He's the deal associate, the part that chases brokers for NDAs and CIM follow-ups so you don't have to babysit your inbox. The thing that does the boring follow-through every searcher swears they'll stay on top of and never quite does. I built the first version of him as a loose AI agent. Give him a goal, hand him the tools, let him figure out the rest. It was genuinely clever to watch. It was also flaky in ways that matter when real money's involved. He'd pull an EBITDA figure off the wrong line and state it with total confidence. He'd breeze past owner-dependency signals that were sitting right there in the teaser. Clever most of the time, wrong just often enough that I couldn't trust him with anything I actually cared about. So I rebuilt him. Not as a smarter agent, as a smaller one. I migrated Marcus into the platform and bolted declarative code around every part that has to be exact. The numbers, the deadlines, the status of each broker thread, all of that runs on rails now. Deterministic. Same input, same output, every time. I kept the AI judgment only where judgment actually helps: drafting the follow-up, reading the tone of a reply, flagging what looks off for a human to check. That's the whole lesson, and I suspect it applies far past my little corner. An AI agent with no rails is a demo. An AI agent with rails is a tool. The work isn't making the model smarter; it's deciding where you let it improvise and where you absolutely don't. Marcus is dependable now. He's a product feature, not a science project, and the difference is that I no longer hold my breath when he sends something. If you've got broker threads going cold while you're heads-down reading a CIM, that's exactly what Marcus is for. Come see what he handles so you don't have to. There's a free 14-day trial.