Shortlist Refresh, Scored — 2026-08-30

Rescored 2026-08-30 PM under two founder rulings. Ruling 1: a business does NOT need to replace his income on day 1 at 1-2 hrs/week. That is a milestone to build toward, not a gating criterion for inclusion. The step-away model below is retained as a milestone-distance metric; no candidate is excluded or demoted for missing the bar today. Ruling 2: an AI-interfacing-with-hardware lane is added — dental labs / CAD-CAM shops, small job shops, print shops. This explicitly reverses the dental-lab lane-kill this note previously carried (which scored labs 0 on axis one by treating them as pure physical production; the CAD-CAM digital workflow was the miss).

Refresh of the [[2026-05-03-tampa-target-shortlist]] against current listings, 119 days later. Every candidate scored on the founder's 5-point fit test, plus a step-away-number model per candidate.

The two-line version

Three of the five May candidates are gone and a fourth is under offer, so the May list is effectively one live name. The refreshed funnel produces two candidates at 8/10 and a lot of 6s. The step-away math stays in the note as a milestone-distance metric (Ruling 1: day-1 paycheck replacement is a target to build toward, not an inclusion gate). With the home build parked, the milestone is $10.9k a month of true recurring living spend ($12.2k on the conservative variant), not the $16k-$19k of raw Monarch outflow this note first used. With the build live it is roughly $15.2k. The best candidate lands at $10.8k-$12.9k a month — already at the build-parked milestone at the top of its range — and the second 8/10 sits $5k-$8k a month short, a gap agent-driven margin expansion is precisely the thesis mechanism for closing.

What happened to the May five

Verified 2026-08-30 by browser. All five URLs refuse programmatic fetch, so each was opened in a real browser session.

The Pasco home health figures are unchanged where they are still visible: asking $314,578, revenue $1,383,868, Seller's Discretionary Earnings (SDE) not disclosed, seller financing $235,933 down plus $78,645 over 36 months at 7%. It is under offer, so it is a comp now, not a target.

The Integrated Medical and Behavioral Health listing is also unchanged: asking $600,000, revenue $1,100,000 in the structured field, SDE $244,200. The title still claims "$2.4M+ Revenue" against a $1.1M revenue field. That contradiction has now sat unresolved on a live listing for 119 days. In May it was the number-one diligence question. Four months of it not being fixed is itself a data point about how the seller keeps books.

The durable lesson is about listing half-life. Sixty percent of a broker-call-ready shortlist evaporated in four months. Any acquisition shortlist older than roughly 90 days should be treated as a lane map, not a target list.

The 5-point fit test

Each candidate scored 0-2 on five axes, maximum 10.

1. Knowledge-work cost base — is 60% or more of the cost base structured knowledge work an agent stack can absorb? Ruling 2 correction: for AI-hardware-lane candidates (dental labs / CAD-CAM, job shops, print shops), the digital workflow — CAD design, quoting, nesting/slicing, case and queue management, customer comms — counts as absorbable structured work. Physical production alone no longer zeroes this axis.
2. Recurring contracted revenue — contracted, recurring, low churn?
3. Regulatory or accreditation moat
4. Retiring owner-operator, weak tech stack, clean books
5. Family-orbit edge — do the healthcare-operations brothers-in-law give a real diligence or reference advantage?

Scored roster

Verified rows were opened in a browser on 2026-08-30. Unverified rows are search-snippet derived; every marketplace blocks programmatic fetch, so those figures are leads to confirm, not confirmed data.

Note the two highest-scoring accounting practices both went under offer, and both explicitly required a CPA buyer. Good small accounting books clear fast.

The AI-hardware lane: dental labs / CAD-CAM, job shops, print shops

Status: OPEN — founder ruling 2026-08-30 reverses the lane-kill this section previously recorded.

The earlier version of this section disqualified dental labs on thesis: "a dental lab's cost base is skilled technicians doing physical production... not structured knowledge work, and it scored 0 on axis one." That reasoning is retired. A modern lab runs a digital pipeline — intraoral scan intake, CAD design of crowns/bridges/prostheses, CAM milling and printing, case tracking, remake management, dentist comms — and that pipeline is exactly the AI-interfacing-with-hardware surface the founder wants exposure to. The same logic admits small CNC/job shops (quoting, CAM programming, scheduling) and print shops (prepress, imposition, quoting, job tracking). This also reconnects to the [[../physical-ai-thesis/2026-05-03-opportunity-map]], which already flagged niche digital-fabrication operators as whitespace.

What is still true from the earlier read — the availability facts stand:

Dental supply distributors: zero confirmed Florida listings. The two that surfaced fell out — one had no retrievable state or price, and the other resolved to San Bernardino County, California.

Dental labs: structurally thin. BizQuest's entire statewide "Medical and Dental Labs" category holds 22 listings, and it is diluted by dental practices. The Tampa Bay metro cut of that category returns exactly one listing, which is a $7.5M Brandon dental practice, not a lab. Roughly five to seven identifiable Florida labs exist across all sources; one is sold, one is sale-pending, two publish no financials, and the only one cleanly in the SDE band was confirmed as a stale 404.

Two structural reasons, and the second matters more:

• Labs are technician-founded owner-operator shops that mostly transact off-market.
• Trade commentary points to an active in-state consolidator buying these shops directly, which would mean bidding against a dedicated proprietary sourcing function rather than other individual buyers. This is UNVERIFIED — a specific company and ticker surfaced in research but the pairing could not be independently confirmed, so it is recorded as a lead to check, not as a basis for the conclusion. The lane-kill stands on the other two arguments without it.

There is also a thesis problem independent of supply. A dental lab's cost base is skilled technicians doing physical production of crowns, bridges, and prostheses. That is not structured knowledge work, and it scored 0 on axis one. The lane fails the fit test on its merits, not just on availability. Struck per Ruling 2 — kept visible so the reversal is auditable. The corrected read: the digital CAD-CAM workflow is absorbable structured work (axis 1 = 1, not 0), and the dental adjacency carries the family-orbit edge (axis 5 = 2, via the OMFS brother-in-law). The St. Petersburg lab rescored 3 → 6; a live equivalent is a mid-table candidate, not a dead lane.

The channel is direct outreach, and it is now an active sourcing project, not a conditional: Florida Dental Laboratory Association membership rolls plus Sunbiz entity search for labs; local CNC/job-shop and print-shop equivalents via trade directories and Sunbiz. Marketplace scraping demonstrably does not reach these shops — they transact off-market. Job shops and print shops currently have zero scored candidates in this funnel; that is a sourcing gap to fill on the next scan, not a verdict on the lane.

The step-away number — now a milestone metric, not a gate

Ruling 1 (2026-08-30): a candidate does not need to hit this number on day 1 at 1-2 hrs/week of his time. It is the milestone the acquisition builds toward — via agent absorption of the cost base — and this section measures each candidate's distance from it. Nothing below excludes or demotes a candidate from the shortlist.

Definition, as the founder approved it: business cash to him after debt service and after paying a General Manager, sustained two quarters, at or above his recurring monthly living spend as Monarch shows it, plus a household runway floor.

Note the test changed. The earlier version of this note measured against take-home pay, an income-replacement test. This one measures against recurring spend. That swap is a second driver of the reversal below, independent of the decomposition, and it is the friendlier of the two tests.

Model and assumptions

The General Manager band is built on the vendored Bureau of Labor Statistics May-2025 Occupational Employment and Wage Statistics (OEWS) data at ~/Documents/phdata-projects/organizational-intelligence/value-model/data/oews-may-2025/:

Healthcare-lane candidates carry a licensed-administrator premium and should use the 11-9111 row, which pushes the loaded band to roughly $150k-$196k.

Working capital, capital expenditure, and taxes are not modeled. Including them would reduce every figure below.

The threshold, and the basis

Founder ruling 2026-08-30 15:12 ET: model against Monarch only. His income and his Monarch-visible expenses. Any expense not in Monarch is assumed covered by Michelle's income and treated separately. Michelle's income is out of the model entirely — this replaces the earlier framing that carried it as an input.

The bar: $10,900 a month of true recurring living spend, or $12,200 a month on the conservative variant. Confirmed 2026-08-30 via Monarch decomposition of the June-August 2026 window. Read "confirmed" narrowly: one Monarch pull, self-bucketed, unreplicated against a second window, and not reviewed by an accountant. The $16,000-$19,000 figure this note first carried was raw Monarch outflow, and it was inflated. Raw outflow ran about $20,900 a month across the window, and four buckets inside it are not recurring living spend.

Raw $20,900 minus $8,628 leaves about $12,300 a month on those rounded inputs; the decomposition's own conservative figure is $12,200, and the roughly $70 difference is rounding in the raw monthly average. (The per-month column is each aggregate divided by three and rounded to the dollar, so the rows sum to $8,629 while the true total is $25,884 / 3 = $8,628.) $12,200 is the conservative variant because it still carries Ray Data Co business expenses that sit mixed into Monarch (Gusto and similar, roughly $1,300 a month, taken here as the residual between the two variants rather than itemized separately). Strip those and true recurring household spend is $10,900 a month.

$12,200 is the number to trust. $10,900 is not a Monarch output; it is $12,200 minus a $1,300 residual that was never itemized.

Two caveats stay attached to this bar. First, Monarch returned only a three-month aggregate for the window, with no per-month split, so every monthly figure above is an average and a single heavy month is invisible inside it. Second, home-improvement "one-times" are not one-time during an active build: while construction is live, bucket (a) recurs. That is not a small caveat. Bucket (a) is $4,187 of the $8,628 haircut, so half the correction rests on the one bucket the live build reinstates. Strip only buckets (b), (c) and (d) and the bar is about $15,200 a month, back inside the range this note just called inflated. Two smaller notes in the same direction: the $5,000 IRS lump is treated as one-time, but self-employment brings quarterly estimated payments, and dropping Florida Prepaid from the bar means the step-away scenario stops funding college.

Everything below is stated monthly rather than annually. The earlier gross-compensation framing ($205k-$215k of base plus cert escalators) is superseded by the Monarch-only ruling and should not be carried forward.

The one remaining blank

• [BLANK] The household runway floor — months of Monarch-visible cash to hold before any step-away is safe.

The model runs with the runway floor at zero. Every figure below is the version before any reserve is set aside; setting a real floor raises the bar, and the top candidate only straddles it as things stand.

Also unmodeled: the benefits buyback. Take-home is already net of his current payroll deductions, but leaving phData means buying coverage on the open market and losing the employer 401(k) match. That is a real subtraction from every row below and it is not in the numbers. It is not a labeled blank because the ruling reduced the blanks to one, but it is a known negative.

Per-candidate result

Debt service uses the listing's own published terms where they exist, and the modeled assumption otherwise.

The Palm Beach row is the only one built on real published financing. That listing states a third-party SBA lender is approved at 10% down, with terms of $810,000 over 120 months at 10.00%, monthly payment $10,704.21.

The monthly column is an ESTIMATE: pre-tax business cash taxed at a single assumed 27.5% effective rate (Florida has no state income tax), not reviewed by an accountant. A single rate is used deliberately — pairing a low rate with the high end and a high rate with the low end would imply that higher income is taxed more lightly. One asymmetry to disclose: the 27.5% rate is applied only to positive cash. Negative bounds are shown pre-tax, since an operating loss generates no tax benefit here. That is why the two loss-making rows read -$3.3k and -$3.1k rather than -$2.4k and -$2.2k.

The General Manager band splits by role, not by industry label: the higher 11-9111 band applies only where a licensed clinical administrator is required (the provider practice), not to the back-office billing shop, which uses 11-1021.

The finding

Measured against the corrected build-parked bar, the top of the funnel is in range. The Palm Beach CPA practice models at $10,800-$12,900 a month against a bar of $10,900 true recurring to $12,200 conservative. It reaches the bar at the top of its range and sits under it across the rest. That is a candidate worth diligence, not a candidate that clears. Against the build-live bar of roughly $15,200, the earlier conclusion still holds and nothing in this funnel clears.

This is a reversal, and it is worth naming as one. The earlier version of this note measured against raw Monarch outflow of $16,000-$19,000 and concluded that no candidate replaced the paycheck. The decomposition above shows that bar was inflated by roughly $8,600 a month of one-times and mislabeled savings, and the conclusion that rode on it does not survive the correction.

Both 8/10 candidates are worth a broker call, and under Ruling 1 both are full diligence candidates. The medical billing and revenue cycle management firm models at $3,900-$6,100 a month today — $5k-$8k short of the milestone — but it is no longer demoted for that: it has the strongest agent-absorption surface in the funnel, which means it has the most milestone-closing headroom of any candidate. Day-1 distance from the bar is now a trajectory input, not a disqualifier.

Three subtractions are still not in any row above, and each one alone would erase the straddle: the household runway reserve (still blank, modeled at zero), the benefits buyback (open-market family coverage plus the lost 401(k) match, plausibly $2,000-$3,000 a month, not modeled), and the build-live version of the bar. Under Ruling 1 these subtractions push the milestone further out; they no longer strike anyone from the list. Read the state of play as "two candidates are worth full diligence, one starts at the milestone and one has the most headroom to build toward it" — not as "the acquisition path ends the day job."

What size would actually clear it

To clear $10,900-$12,200 a month after tax requires roughly $180,000-$202,000 a year of pre-tax business cash. Hold price at 3.5x SDE, 90% financed at 10.5% over 10 years, so annual debt service equals 0.510 x SDE, with a $144,000 General Manager:

SDE ~= $662,000-$706,000   ->   asking ~$2.3M-$2.5M   ->   down payment ~$232,000-$247,000

That is roughly 1.6x-1.7x the median candidate's earnings and 1.7x-1.8x the median asking price in this funnel (medians across the live rows, excluding the Miami row because it reports EBITDA rather than SDE and the Pasco row because it publishes no SDE: $418,618 SDE, $1.35M ask; note the Palm Beach ask feeding that median is itself the ~$900,000 implied estimate, not a published figure).

The 3.5x is the load-bearing assumption here, and it is unsourced. The live rows in this funnel trade from 1.18x to 4.01x SDE. Re-solve at 2.0x and the required SDE drops to about $458,000, which is below the Palm Beach practice. So "he needs a bigger business" is largely an artifact of the multiple, not a finding. The real reason Palm Beach works is its published SBA terms: $128,450 of annual debt service on $468,837 of SDE is 0.27x, against the 0.51x the model assumes elsewhere. The sourcing mandate that follows is hunt cheap multiples with published financing, not hunt bigger businesses.

The down payment is still the binding constraint, and it does not comfortably fit. At $232,000-$247,000 it consumes 73-88% of his visible cash of $285,000-$320,000 (cash figure founder-disclosed, August 2026), before closing costs, the SBA guaranty fee and post-close working capital, and with the runway floor at zero - and it is the same cash the construction-to-perm loan needs. Keep two numbers separate: the recommended Palm Beach deal needs roughly $90,000 down on the implied ~$900,000 price. The $232,000-$247,000 figure belongs to the hypothetical resize target.

Recommended top three

1. CPA and accounting practice, Northern Palm Beach County — 8/10

Asking on request; SBA terms imply roughly $900,000. Revenue $688,247, SDE $468,837, seven years established, two employees plus the CPA owner, reason for sale is downsizing, 1,500 square feet at $2,128/month on a lease through 2029-12-31, sixteen weeks of training included at no cost. Transworld Business Brokers of South Florida, listing ID 3968292, contact Max Crescenzi.

Revenue is unusually granular for a listing: $150,000 monthly accounting across 20-30 clients, $263,739 individual tax across 245 clients, $166,110 business tax across 129, $38,835 estate across 33, $22,979 consulting.

Why it ranks first: it scores 8, it is the only candidate whose modeled cash reaches the corrected build-parked bar at all ($10.8k-$12.9k against $10.9k-$12.2k, touching it at the top of its range), and it is the best-documented deal in the refresh. Its published debt service is 0.27x SDE against the 0.51x the model assumes elsewhere, which is the whole reason it gets there. An SBA lender has already underwritten it, which is a third party staking money on the books being real.

What Ray's agents absorb in the first 18 months: monthly close and bank reconciliation across 20-30 accounting clients, individual and business return preparation and review queues, client document chase, and extension and deadline tracking — the bulk of a 245-client individual tax book is structured, rules-bound production work.

Single next diligence step: ask Max Crescenzi for the published asking price and, in the same call, whether the buyer must be a licensed CPA.

2. Medical billing and revenue cycle management firm, behavioral health and addiction — 8/10

Asking $2,200,000, revenue $628,801, SDE $548,378, established 2017, three employees (one full-time billing specialist and two part-time billers), owner works one to two hours per week, 21 active providers on $30,000-$70,000 annual contracts, monthly operating expense $6,700-$7,000. Website Closers, listing ID 3987885, seller reference WC 4081.

Two corrections worth carrying: the business is listed as "Florida" with no city — the Tampa attribution came from Website Closers' own headquarters, not the business. And the $1.8M / $484,808 figures circulating from earlier snapshots are superseded; that listing (118226) is a 404, and the deal was re-listed higher at $2.2M with the average contract restated from $40,000 to $50,000.

Why it ranks second: the tie with Palm Beach on score is now broken only by milestone distance and price risk, not by a pass/fail economics test (Ruling 1). Its modeled $3,900-$6,100 a month is $5k-$8k short of the milestone today, but it owns the strongest agent-absorption surface in the funnel — the most room to build toward the milestone — and the strongest family-orbit edge: behavioral health, addiction, detox, and PHP-IOP billing is precisely [[michael-holzum]]'s world at ABA Centers of America, and Exact Billing Solutions is the direct comparable. What still keeps it behind Palm Beach: the $2.2M ask prices the milestone-closing work into the multiple, and its stated 87% SDE margin ($548,378 on $628,801 of revenue with three staff) is unreconciled, so treat that SDE as a question rather than an input.

What Ray's agents absorb in the first 18 months: eligibility verification, prior authorization submission and follow-up, claim scrubbing and submission, denial triage and appeal drafting, payment posting, and provider credentialing — a revenue cycle management shop is close to a pure structured-knowledge-work cost base, which is why it is the cleanest thesis fit even though the price does not work.

Single next diligence step: ask Website Closers directly why it was re-listed at a $400,000 higher price, and get the "$300k+ Annual Revenue" seller bullet reconciled against the stated $628,801.

3. Integrated Medical and Behavioral Health Platform, Tampa — 6/10

Asking $600,000, revenue $1,100,000 in the structured field, SDE $244,200, established 2017, 19 employees, Hillsborough County, WebsiteClosers / Seller Force. BBS 2466935.

It ranks third on cheapest entry, verified liveness, and the same behavioral-health family edge — not on its economics, which are the worst in the top group: minus $39,000 to plus $7,000 a year after a licensed administrator.

What Ray's agents absorb in the first 18 months: eligibility checks across 77-plus payer plans, prior authorization for psychiatry medication management, scheduling and no-show prediction, telehealth coordination, and billing follow-up.

Single next diligence step: one question, before anything else — is trailing-twelve revenue $2.4M or $1.1M? The listing has contradicted itself for 119 days.

Standing lane action: AI-hardware sourcing (no ranked candidate yet)

Per Ruling 2 the dental-lab / CAD-CAM / job-shop / print-shop lane is open, and its channel is direct outreach, not marketplaces: FDLA membership rolls + Sunbiz for labs, trade directories + Sunbiz for job and print shops. The St. Petersburg lab comp (rescored 6/10, dead listing) shows a live find would slot mid-table on score with real upside on axes 1 and 5. First live, verified candidate in this lane gets scored into the next refresh.

Two gates before any of this moves

Gate 1 — the phData outside-work question. The sharpened filter permits healthcare admin services, revenue cycle management, accounting, bookkeeping, and marketing agencies, and bans anything competitive with phData: data analytics consultancies, data engineering shops, business intelligence consultancies, AI agencies. All three recommendations sit on the permitted side. But moonlighting consulting was ruled out by the founder himself, and none of these are passive. Owning an operating business while a Forward Deployed Engineer at a consultancy is a real conflict-of-interest question that has not been asked, and it should be asked before a Letter of Intent (LOI), not after.

Gate 2 — the home build competes for the same cash. Building on the current lot rolls the $414,000 payoff into a construction-to-perm note of roughly $1.5M-$1.7M, taking all-in carry to about $13,000-$14,000 per month against $4,500 today. The household test for that is plus $8,000-$9,000 per month, and the build decision is already gated on six consecutive months at target from October 2026 through March 2027. A $232,000-$247,000 acquisition down payment and a construction down payment still cannot both come out of $285,000-$320,000 of cash. These are one decision, not two.

A third thing worth naming

The four-walls spec says he must not become the operator — the whole thing has to be his lane, with a real upside stake. The step-away model assumes a General Manager precisely to protect that wall, and the corrected bar changes this too. The Palm Beach row straddles the bar with a $126,000-$162,000 loaded General Manager already inside it, so the wall stands on that candidate. It does not stand everywhere: on the other rows the numbers only close by deleting the General Manager, and any version that does is a version where the thesis has failed its own test. Worth knowing before a broker call rather than after an LOI.

Methodology and confidence

Search date: 2026-08-30. Channels: BizBuySell, BizQuest, businessesforsale.com, business2sell, Venturu, DealStream, HealthCareBizSales, Sunbelt, Transworld.

The binding methodological fact: BizBuySell, BizQuest, DealStream, Transworld, businessesforsale detail pages, LoopNet, Truforte, and HealthCareBizSales all return HTTP 403 to programmatic fetch. Only businessesforsale.com search pages and business2sell read cleanly. Everything else required a real browser session. This is unchanged from May and should be treated as a permanent constraint on this workflow, not a transient failure.

Confidence:

• HIGH on the ten browser-verified rows: liveness, asking price, revenue, SDE, and the published SBA terms on the Palm Beach practice.
• MEDIUM on snippet-derived rows (numbers 5, 6, 7, 9) — figures come from search-engine snippets of listing cards, not from reading the listing.
• LOW on reason for sale, employee count, and financing terms for every unverified row. These do not survive on listing cards.
• ESTIMATE, explicitly: the 10.5% SBA rate, the General Manager band, the 3.5x multiple in the required-SDE solve, and the implied ~$900,000 Palm Beach price.

One data-quality warning worth carrying forward. Several Florida senior-care listings repeat near-identical price, revenue, and cash-flow ratios across different cities (Orlando $265,000 / $1,710,000 / $257,400; Daytona Beach $262,500 / $1,724,850 / $257,590), and one Miami business appears twice under two different titles with identical figures ($169,999 / $588,508 / $135,357). This is templated broker posting. Do not count these as distinct targets, and treat any Florida senior-care count as inflated.

Mapping against Ray Data Co

The May shortlist asked "which business can agents rebuild?" This refresh initially asked "which business, after debt service and after paying someone else to run it, replaces the paycheck?" — and the founder's 2026-08-30 ruling corrected that back: paycheck replacement at 1-2 hrs/week is the milestone the agent rebuild is supposed to produce, not the entry filter. The operative question is now the May one, with the step-away number as the scoreboard: which business can agents rebuild fastest toward the milestone?

The finding that survives past any individual listing: once the bar is measured correctly, the Service-as-a-Software acquisition thesis and the step-away goal are no longer mutually exclusive - they are competing for the same cash. A business priced at 3.5x and large enough to clear the corrected bar needs a $232,000-$247,000 down payment, which consumes 73-88% of visible cash; the recommended Palm Beach deal needs closer to $90,000. Either is spendable, neither is spendable alongside a construction down payment. That is a sequencing decision, not a sourcing problem, and no amount of additional marketplace scraping changes it.

This also lands on the same wall as [[../../.claude/projects/-Users-ray/memory/project_income_seat_gap_400k]]: the constraint is demand generation, not capital. An acquisition buys existing demand, which is genuinely the point — but at these prices it buys demand at a multiple that consumes the cash flow it was bought for.

Related

• [[2026-05-03-tampa-target-shortlist]] — the shortlist this refreshes and supersedes
• [[../../04-finance/2026-05-03-path-to-5m-liquid-analysis]] — the financial plan this serves, including the sharpened acquisition filter and the phData competitive ban
• [[../../06-reference/2026-05-03-heyrico-service-as-a-software-shift]] — the L1/L2/L3 framework behind axis one of the fit test
• [[../../06-reference/2026-05-03-icbd-holdings-curative-ai-research]] — the behavioral-health revenue cycle management competitive picture and the Publix RICO risk flag that still applies to candidate 2
• Four-walls commitment spec (memory: user_career_commitment_shape) — the instrument behind the operator warning above; the mission, whole-lane, build-then-teach, and upside-stake test that any acquisition has to clear
• Home build and household cashflow (memory: project_home_build_and_household_cashflow) — gate 2, and the source of the two blanks he has to fill