| Name | Title | Shop | Location | Phone | Persona | Ownership | DRP | Tier |
|---|
What this list is for. Ten discovery calls with people inside US independent collision (body) shops, to test three numbers before building here: 1 — leakage (what the shop loses per month to insurer short pays and never-billed operations), 2 — pricing (flat monthly, per file, or a share of recovery), 3 — DRP fear (how much fear of losing insurer referral programs stops them fighting). Every criterion below exists to put the person who owns those numbers on the phone.
1. Personas — who inside the shop
The ranking rule: the owner buys, the estimator uses, the office manager sees the proof.
| Persona | Who | Titles | What they feel / why they matter | Target mix |
|---|---|---|---|---|
| P1 | Owner / operator — the buyer | Owner, Co-Owner, President, Founder, CEO, General Manager, Operator, Principal, Partner, “Owner / Estimator”, Managing | Often a former tech or estimator who bought or started the shop; frequently second-generation. Makes the relationship calls: which DRPs to keep, how hard to push an adjuster. Owns all three numbers, especially pricing (2) and DRP fear (3). | ≥70% of shop rows |
| P2 | Estimator — the user | Estimator, Lead / Senior / Collision Estimator, Estimating Manager, Estimatics Manager, Supplement Writer / Specialist, Blueprint Specialist / Blueprinter, Repair Planner, Insurance Coordinator, Production Estimator | Feels time: six cited rebuttals is 2–3 hours. Knows the P-pages and the DEG, and eats small short pays because fighting isn't worth it — the best source for leakage (1). Risk: may hear “automation” as “replacing my job”. | ≤20% of shop rows |
| P3 | Office / AR — the proof seat | Office Manager, Business Manager, Accounts Receivable, Bookkeeper, Controller, Administrator, Receptionist | Reconciles billed vs paid and finds short pays weeks after delivery; can quote real leakage numbers off the ledger. | ≤10% of shop rows |
| TrackB | Channel & insight — not a buyer | 20-group facilitators and consultants, outsourced estimating / supplement services, ADAS calibration and scan sublet vendors, independent appraisers, state association executive directors, trade-press bylines | Sees many shops at once and is how the market is reached. ADAS/scan sublet vendors invoice shops and watch carriers deny calibrations across dozens of them — highest insight value per call. | separate list, 10–15 |
How persona was actually assigned
Not by trusting the source. Title text is re-read on every row and overrides the source label, in this order: owner / co-owner / president / founder / principal / general manager / GM / operator / managing / partner → P1; estimator / estimat- / supplement / blueprint → P2; office manager / business manager / accounts receivable / bookkeeper / controller / administrator / receptionist → P3. Rows already marked TrackB keep it. Anything that matched none of those was dropped rather than guessed — which is why there are no “other” rows in the file. First match wins, so “Owner / Estimator” is P1 and “General Manager / Owner” is P1.
Anti-personas — excluded on sight
- Carrier staff, independent-adjusting-firm staff, consolidator employees. “Appraiser” and “adjuster” titles are usually insurer-side; one was included only where the employer was verifiably a shop.
- Consolidator-owned shops: Caliber Collision, Gerber / Boyd Group, Crash Champions, Classic Collision, Joe Hudson's, Service King, Abra, Quality Collision Group, Sterling Autobody, and other PE-backed MSOs — checked against acquisition news in the last 24 months.
- Not collision: mechanical-only, glass-only, PDR-only, detailing, restoration- or custom-only, RV, heavy truck.
- Fewer than ~15 repair orders a month; more than 10 locations.
- Vendors and consultants posing as shop staff — valid only on Track B.
- People building or selling a competing tool (Rote, VaraFix, Shop Sheet, EstimateScrubber …) — competitive intel, not contacts.
- Ex-industry people with no current shop role (last shop role more than 12 months back).
2. Ownership class — the four buckets
Exactly one value per row. It classifies the shop business, not the person.
| Value | Means | Gate it had to pass |
|---|---|---|
family_owned | Family-owned, family-operated, multi-generation, or the founding family is named. | Requires a quote containing family / generation / grandfather / father / son / daughter / brother / sister / husband / wife / “founded by” / “since 19xx”. |
independent_shop | Single-location or small owner-operated independent with a named owner and no chain or franchise branding. | Requires at least one citation on the row (site, directory, LinkedIn). No quote, no label. |
company | Part of a chain, multi-shop operator, dealership group, franchised banner (CARSTAR, Fix Auto, Maaco, ProColor), or PE-backed group. | Requires a quote containing chain / multi-shop / group / dealer / franchise / banner / network / “part of” / acquired / locations / MSO. |
unknown | Nothing public and quotable either way. | This is the honest default. 93 rows landed here and were left here. |
Validation runs on every row after assembly: a family_owned or company label whose evidence quote does not contain the language above is downgraded to unknown, and an independent_shop with no citation at all is downgraded too. That gate moved 30 rows from independent_shop and 19 from company into unknown — the labels that survive are the ones with something to point at.
Why
company rows are in the file. The brief says drop consolidator-owned accounts, and genuine consolidators were dropped. The 18 surviving company rows are dealership-group body shops and franchise banners — deprioritised, labelled, and sorted to the bottom of the table rather than deleted, because you asked for the classification.3. Tier — who to call first
| Tier | Rule as implemented | Count |
|---|---|---|
| A | Persona P1 and at least one of: a pain quote with URL, an email confirmed verbatim on the shop's own site, or a board/president role recorded in the hook. | 134 |
| B | P1 / P2 / TrackB and at least one of: pain quote, email, LinkedIn. Everything reachable that isn't A. | 190 |
| C | No persona, no channel, or nothing citable. Zero rows ended up here — they were dropped instead. | 0 |
4. Every other column
| Column | Values / meaning | How it was established |
|---|---|---|
drp_status | non_drp · left_drp · 1_2_drps · 3plus_drps · unknown | Direct repair program = an insurer referral contract. non_drp is set only when the shop's own words say so (“we are not a DRP”, “we don't have a contract with insurance companies”, “we work for you, not the insurer”); carrier logos and “preferred shop for ___” indicate the opposite. 22 non-DRP, 3 left, 8 with one or two, 282 unknown — most shops publish nothing either way. |
drp_evidence | URL + verbatim quote | The sentence that justified drp_status. Empty when unknown. |
estimating_platform | CCC · Mitchell · Audatex · unknown | From job posts naming the platform, site copy, or repair-status links. Unknown unless stated — roughly 55% of the market is on CCC ONE, and shops on Mitchell/Audatex are the underserved ones. |
oem_certifications | Free text, e.g. Honda ProFirst, Ford, GM, Subaru, Tesla, I-CAR Gold Class | Site badges and OEM locator listings. A fit signal, not a pain signal. |
adas_calibration | in_house · sublet · none · unknown | Services page, calibration-equipment posts, ADAS tech job ads. |
est_ros_per_month | Integer, or unknown (225 rows numeric) | A model estimate, not a measurement. Derived from published employee count, bays, locations, review volume or stated throughput. The core band for this campaign is 40–80 repair orders a month (≈$2.3M–$4.6M a year); 20–40 and 80–150 are adjacent. |
est_ros_basis | Short clause | Names the evidence behind the RO estimate, e.g. “12 employees on the team page”. Read this before quoting the number. |
est_annual_revenue_usd, employee_count, locations_count, year_founded | Numbers where published, else unknown | Team pages, LinkedIn company pages, directories. year_founded doubles as family/independent evidence (“since 1973”). |
email_verified | yes · no · n/a | yes = we fetched the shop's own domain (homepage plus up to five common contact/about/team paths) and found that exact address in the HTML. no = the address was published by the enricher but not on the pages we fetched — treat those 62 rows as bounce risk. n/a = no email on the row. |
email_source | site-verified · enrichment-unverified | Plain-language version of the same fact. |
pain_signal | URL + verbatim quote (≤30 words) | Public evidence the shop or person has fought insurers, or wishes it could: a specific cut line or carrier named, a lawsuit, an appraisal clause, a DOI complaint, testimony, a trade-press quote, a public non-DRP stance. Present on 68 rows. |
pain_type | lawsuit_or_escalation · trade_press_quote · public_post · non_drp_stance · association_activism · none | The shape of the evidence, so you know whether the opener is “I read your post” or “I saw the filing”. |
hook | One line | The most specific public thing to open with — a dated post, a board seat, a lawsuit, a founding year. Present on 146 rows. |
ownership_evidence | URL + verbatim quote | The basis for ownership_class. Empty means the class is unknown. |
contact_channels | e.g. linkedin, email, phone | Convenience roll-up. Every row in the file has at least one. |
confidence | high · medium · low | high = email and LinkedIn; medium = one of the two; low = neither resolvable (none remain). |
source_urls | Semicolon-separated URLs | Where the row came from — for the slice-built rows this includes the page that named the person. |
5. Evidence rules the whole list obeys
- URL plus verbatim quote, or it didn't happen. Every signal, every ownership class, every DRP status carries the sentence and the link. No quote, no claim.
- Unknown stays unknown. Fields were never filled from inference — which is why a third of the ownership column reads
unknownand 282 of 324 DRP statuses do. - Public business channels only. Shop email, shop phone, business LinkedIn. No personal cells, no home addresses.
- Independence checked before anything else, because consolidators buy independents constantly — that is what removed Nick Orso's Body Shop (Gerber now runs the same address and phone), K&M Collision (bought by Collision Partners, Jul 2026), Painters Collision Centers (sold to Classic Collision, Jan 2024), Cascade Collision and DCR Systems.
- Employer side verified. “Appraiser” and “adjuster” are usually insurer staff, so those rows were kept only where the employer was demonstrably a shop.
- Recency preferred — the last 12–24 months outrank older, except for an escalation such as a lawsuit or testimony.
- One contact per shop, and one row per person. Where two people at the same shop qualified, the higher persona rank wins, then the richer contact record.
- Quotas enforced: P1 ≥70% and P2 ≤20% of shop rows (the P2 tail was cut to the best 60 by contact richness), ≥6 states with no state above 25% (44 states/DC, largest 8.6%), Track B held separately.
6. Where the rows came from — and what was not verified
Five Parallel FindAll people searches produced 385 LinkedIn-identified US shop people; an eligibility filter removed carriers, consolidators, dealers, PDR/glass/mechanical, non-US and non-target roles; five Parallel enrichment passes added shop details, ownership, size and pain evidence; published emails were then confirmed by fetching 266 shop domains directly. Seven subagent slices worked the sources a search index misses — trade press and litigation, state association board rosters, non-DRP shop sites, LinkedIn pain posts, channel/TrackB vendors, and a state sweep — and every slice's findings were recorded on a shared research board with its evidence captured.
What that does not buy you:
- RO-per-month and revenue figures are estimates from public proxies;
est_ros_basistells you which. - 62 emails are unconfirmed by our own fetch and may bounce.
- A model wrote most rows and was checked in samples, not exhaustively: 10 sampled LinkedIn URLs were re-resolved live (8 returned name, company and title), and pain quotes on 3 rows were re-fetched and found verbatim on the source page. Sample-level proof, not line-by-line certification.
- Three worker agents died on model stream errors mid-build; their slices were rebuilt from their transcripts and re-verified, so a small share of their intermediate candidates was lost rather than included.
- Known individual risk flags are already in
hook/confidencefor row-level reading: a franchise banner taken up since the LinkedIn was written, a title sourced from a directory only, an owner with a federal indictment on record, a principal whose current role is a consultancy rather than the shop.