PremiumBack

1. One-liner

Wins back the five-figure premium your carrier charged for subs you thought were insured.

2. Trend signal — why now?

Three things moved at once.

Carriers got aggressive in 2026. Heightened scrutiny is now being applied to subcontractor COI tracking specifically, driven by rising construction litigation costs (FO Agency). This isn’t a general “insurance is expensive” complaint — it’s a named, current shift in how auditors treat one specific document.

The mechanic is brutal and mechanical. At audit, the auditor reviews all subcontractor payments and requests a COI for every trade partner used. Where a valid COI is missing, those payments get reclassified as uninsured, the labor portion becomes direct payroll exposure, and the carrier charges the full manual rate for that trade’s classification. If a sub doesn’t carry their own comp, their payroll is treated as your payroll at the highest applicable class code on your policy (Fisher Phillips). Result: five- and six-figure premium adjustments. Carriers can apply surcharges up to 200% of the original estimated premium.

And — this is the whole idea — it’s reversible. From LegalClarity: “If a subcontractor had coverage but you simply didn’t have the certificate on file during the audit, getting it now and presenting it with your dispute can eliminate the added payroll charge entirely.”

That sentence is the product. The money isn’t lost because the sub was uninsured. It’s lost because the GC couldn’t produce paper on audit day. The coverage existed. The proof didn’t. And there’s a 30–60 day window from the audit statement date to fix it — after which the original audit becomes final and the full amount is due.

Everybody in this market sells the wrong product. COI tracking software — bcs, TrackMyVendor, myCOI, SmartCompliance — answers “is this certificate expired?” and has been commoditized to near-zero (bcs: free up to 25 vendors; TrackMyVendor: first 25 vendors free, then $39/mo). Meanwhile premium recovery consultants, who solve the actual money problem, only take clients paying $100,000+/year in premium, at contingency fees as high as 50%. The contractor paying $30K/year in premium and eating a $40K audit back-charge has nobody. Tracking software won’t help — the audit already happened. Recovery consultants won’t take the call.

Provenance:

• Signal 1 (demand): Uninsured/undocumented subcontractors produce five- and six-figure premium adjustments at audit; carriers reclassify sub payments to the highest applicable class code — Fisher Phillips / FO Agency — observed 2026-07-29
• Signal 2 (feasibility): The charge is reversible by retroactively producing the COI within a 30–60 day dispute window; disputes are viable up to ~3 years back — LegalClarity — observed 2026-07-29
• Signal 3 (economic): Premium recovery is an established paid service on contingency, but gated to employers paying $100K+/yr in premium, leaving smaller contractors unserved — Apex Services — observed 2026-07-29 Category: Underserved niche

3. The opportunity

Two adjacent markets both refuse to serve the same customer, for opposite reasons.

COI tracking software is a race to zero. It’s a preventative tool sold to people who haven’t been hurt yet, which is why it prices at $0–39/month and why adoption among small GCs is poor — nobody buys insurance against paperwork. Critically, it is useless at the moment of maximum pain. The audit bill is on the desk; a tracker that would have prevented it twelve months ago is worth nothing today.

Premium recovery consultants solve the real problem but are a human-labor business. A consultant reads your audit worksheet, finds classification errors, chases COIs, and negotiates with the carrier. That’s 10–25 hours of skilled work, which is why the floor is $100K+ in annual premium and the fee is up to 50% of recovery. Below that threshold the math doesn’t work for them.

The gap is the contractor with $2–25M in revenue paying $15K–80K a year in comp and GL premium, who just got a $25K–60K audit back-charge. The recovery work is genuinely valuable to them and genuinely unprofitable for a human consultant to perform. That is exactly the shape AI collapses: the expensive part is reading a messy audit worksheet, cross-referencing it against a year of sub payments in QuickBooks, identifying which specific subs triggered which specific charges, and then chasing exactly those subs for exactly the right documents.

Do that for $500 flat plus 15% of what’s recovered, and you’re 3× cheaper than a consultant who won’t take the client anyway.

4. Target market

• Primary customer: Owner or office manager at a US general contractor or specialty trade firm, $2–25M annual revenue, 10–60 employees, who subcontracts significant work and just received an audit statement with a back-charge. Concentrated in residential/light-commercial GCs, roofing, concrete, framing, mechanical, and site work — the trades that lean hardest on 1099 crews and small subs.
• Why they buy: The bill is real, it’s due now, and it’s often larger than the quarter’s profit. From the research: audit recalculations amount to “tens of thousands of dollars for a mid-size business,” and “some contractors receive five-figure audit bills.” The owner’s alternatives are (a) pay it, (b) fight it themselves with no idea how, or (c) call a recovery firm that won’t return the call. Every one of those is worse than $500.
• Rough TAM reasoning: Roughly 12,981 active machine shops exist in NAICS 332710 as a size comparison; the US construction contractor population is far larger — hundreds of thousands of firms in the $2–25M band across GC and specialty trades. I don’t need a precise number: every one of them gets audited annually, and a meaningful slice gets a back-charge every year. If even 3,000 contractors a year buy one engagement, that’s a business.
• Why now for them: Carriers tightened COI scrutiny in 2026 specifically. Contractors who passed audits for a decade are getting hit for the first time this year — and a first-time back-charge is exactly when someone goes looking for help, because they have no internal playbook for it.

5. Product sketch (MVP)

• Upload the audit worksheet. Drop in the carrier’s audit statement and worksheet PDF. The product parses it and shows, in plain English, which line items were reclassified, at what class code and rate, and what each one cost you.
• Connect the books. Pull the year’s subcontractor payments from QuickBooks/Xero (or a CSV) and match every reclassified dollar to the specific sub who triggered it.
• The recovery list. A ranked list: “These 7 subs account for $38,400 of your $47,000 back-charge. Five of them almost certainly had coverage.”
• Automated COI chase. One click sends each identified sub — and their agent — a templated request for the certificate and endorsements covering the exact policy period in question, with automated follow-up. This is the labor that makes consultants expensive.
• Endorsement verification, not just certificate collection. Checks whether the returned document actually carries what’s needed — the certificate proves the policy exists, the endorsement decides who pays. A cert that says “additional insured” in the description box with no attached endorsement has no legal weight.
• The dispute packet. Generates a carrier-ready dispute letter with the recovered COIs indexed against the specific audit line items they refute, formatted to the carrier’s process.
• Deadline clock. Tracks the 30–60 day dispute window from the audit statement date and escalates as it closes, since missing it makes the original audit final.
• Next-year prevention mode. After the dispute resolves, flips into lightweight monitoring for the current policy year — collecting COIs as subs are onboarded, so next year’s audit is clean. This is where the subscription revenue lives.

6. AI angle — what’s load-bearing

Remove the AI and this is a consulting firm, which is precisely the business that already exists and already refuses this customer.

Three places AI does real work:

1. Reading the audit worksheet. Carrier audit documents are inconsistent, dense, and semi-structured — class codes, rates per $100 of payroll or per $1,000 of sub cost, exposure bases, credits. Parsing these reliably across dozens of carrier formats is the single most expensive human step, and it’s a document-understanding problem that’s only recently become tractable.
2. Reconciling worksheet to ledger. Matching “SUBCONTRACTED LABOR — 5403 CARPENTRY — $412,000” against 60 vendor names in a QuickBooks export where the sub is entered three different ways is fuzzy entity resolution across messy real-world data. Rules break; this doesn’t.
3. Judging endorsement sufficiency. Reading a returned COI and its endorsement pages to determine whether it actually establishes coverage for that period, that entity, and that scope — CG 20 10 for ongoing operations, CG 20 37 for completed operations, primary and noncontributory wording, waiver of subrogation. This is the judgment call that makes a recovery specialist worth their fee.

That’s the whole margin. AI turns a 15-hour engagement into a 30-minute one, which is what lets you serve a customer at $500 whom a consultant can’t touch below $100K of premium.

7. Localization angle (if any)

N/A — this is a US-only play. The opportunity is created by a specific national market structure: state-regulated workers’ comp class codes, NCCI-style experience rating, carrier audit and dispute procedures, and standard ISO endorsement forms (CG 20 10 / CG 20 37). Those specifics are the product’s knowledge. The same pain exists in other markets but the mechanism differs enough that it’s a separate product, not a translation. Start in 3–5 states with the highest contractor density and clearest dispute procedures, then widen.

8. Business model — path to $1M–$5M ARR

The trick here is that the entry product is transactional and the retained product is subscription.

• Pricing — Recovery engagement: $500 flat to run the analysis and produce the dispute packet, plus 15% of premium actually recovered. On a $40K back-charge with $28K recovered, that’s $500 + $4,200 = $4,700. A recovery consultant at 50% would charge $14,000 — and wouldn’t take the client.
• Pricing — Prevention subscription: $149/month for ongoing COI collection, endorsement verification, and audit-readiness for the current policy year. Priced deliberately above the $0–39 commodity trackers because it’s sold to someone who has already been burned and now knows what the paperwork is worth.
• ACV: $2,300 blended. A customer who enters via a recovery engagement ($3,500 average) and converts to prevention ($1,788/yr) is worth meaningfully more in year one, less in steady state.
• Rough math to $1M ARR: 400 prevention subscribers × $149 × 12 = $715K, plus ~120 recovery engagements/year × $2,500 average = $300K. Total ~$1.0M. That’s 400 retained contractors — a small fraction of one state’s contractor population.
• Rough math to $5M ARR: ~2,000 prevention subscribers plus ~600 recovery engagements annually. Requires two things to be true: national coverage of carrier audit formats, and an agent/broker referral channel producing steady inbound (see §9). Realistic in year 3–4, not year 2.
• Expansion path: Per-sub pricing tiers as contractors grow; add GL audit alongside workers’ comp; sell the prevention product to the insurance agents as a book-wide service — an agency with 200 contractor clients is a single sale worth $30K+/year and solves distribution permanently.

9. Go-to-market wedge — first 100 customers

The unlock is that the pain has a date on it. Audits happen on policy anniversary, and most policies renew Jan 1 or Jul 1. That makes the customer findable and the timing predictable — unlike almost every other SMB product.

• Independent insurance agents are the channel, and their incentive is aligned. When a contractor gets slammed with a $40K back-charge, the agent gets an angry call and risks losing the account at renewal — but the agent can’t fix it themselves. Handing the client a tool that claws money back makes the agent the hero and saves the account. Scrape the Big “I” (Independent Insurance Agents & Brokers of America) state association member directories for agencies with construction practices, then contact the 300 with visible contractor specialization. Offer a revenue share on recoveries. Ten productive agents, each with 40+ contractor clients of whom a handful get back-charged annually, is the first 100 customers.
• Time the outreach to the audit calendar. Cold outreach in Feb–Mar and Aug–Sep, 6–10 weeks after the two big renewal dates, when audit statements have landed and the dispute clock is running. The subject line writes itself and is specific enough to get opened: “Your audit reclassified subs you probably had certificates for.” Generic SaaS outreach gets ignored; a dated, specific, expensive problem does not.
• Content aimed at the panic search. A contractor who just got the bill searches exactly one thing: “why did my workers comp audit go up so much” / “dispute workers comp audit subcontractor.” Publish a state-by-state dispute-deadline reference and a free audit-worksheet explainer that takes an uploaded PDF and shows what each reclassified line cost. That free tool is the top of the funnel and produces the qualified lead at the exact moment of intent.
• Trade associations and peer groups. State-level ABC (Associated Builders and Contractors) and NAHB chapters run contractor peer groups where an owner who recovered $28K will say so out loud. One case study with a real number does more than any ad. Sponsor two chapter meetings, present the mechanic, collect the room.
• Bookkeepers who serve contractors. Construction-specialized bookkeepers see the audit bill land in the client’s inbox and are already in QuickBooks. Twenty of them, each with 15 contractor clients, is a durable referral base.

10. Build complexity — justification

Medium. The heavy lifting is document understanding across inconsistent carrier audit formats plus fuzzy matching against accounting exports — real engineering, but off-the-shelf multimodal models plus QuickBooks/Xero APIs cover it. No custom model training, no proprietary dataset needed at the start. The genuine work is the domain encoding: class codes, endorsement forms, per-state dispute procedures, and per-carrier dispute processes. That’s research and structured content, not novel technology.

Estimate 12–16 weeks to a v1 for a pair, and I’d narrow it further for the MVP: two or three states, workers’ comp only, the five largest contractor carriers’ audit formats. Manual fallback for anything unparsed — a founder reading a weird worksheet by hand for the first 50 customers is fine and is how you learn the domain.

11. Gating checklist

Gate	Pass?	Note
Legal in target market	✅	Assembling documentation and drafting a dispute on the customer’s behalf is administrative support, not regulated activity. Must avoid presenting as licensed insurance advice or public adjusting — customer files their own dispute; the product prepares the packet. Worth one legal opinion per state before launch.
Ethical — no harm / dark patterns	✅	The product recovers charges for coverage that genuinely existed. It does not help anyone conceal uninsured labor — where a sub truly had no coverage, the charge is legitimate and the product says so.
Market exists (evidence above)	✅	Recovery consultants are an established paid category; the only question is serving below their $100K floor.
1–5 person team can build this	✅	Two people, 12–16 weeks to a narrow v1.
Launchable with <$50K / ₹40L	✅	Inference, API costs, and legal review. Well under $50K.

All five pass.

12. Feasibility score

Axis	Weight	Score	Notes
Problem intensity	20	17/20	Hair-on-fire by definition — an unexpected five-figure bill with a 30–60 day fuse. Loses points only because it’s episodic (annual) rather than daily, so the subscription half of the model is a softer sell than the recovery half.
Demand evidence	15	12/15	Strong structural evidence: an existing paid consulting category, documented carrier behavior, a documented reversal mechanism. Docked because I could not source verbatim contractor complaints — forum sources (Practical Machinist–style construction equivalents) were inaccessible during research, and I won’t invent quotes. This is the axis to validate first.
Build feasibility	15	11/15	Doable in 12–16 weeks by a pair, but carrier-format variance and accounting-export messiness are real engineering drag. Not a 6-week solo build.
Distribution clarity	15	12/15	The agent channel is specific, incentive-aligned, and enumerable; the audit calendar makes timing predictable. Docked because agent partnerships take time to land and the direct-outreach conversion rate is unproven.
Revenue mechanics	15	12/15	Pricing is anchored against a real, known alternative (consultants at up to 50%) and the value is a literal dollar figure on the customer’s desk. Docked because the transactional-to-subscription conversion rate is the load-bearing assumption for ARR and is currently a guess.
Time to first revenue	10	8/10	Contingency structure means you can sell before the product is finished — manually service the first ten customers and get paid. First revenue plausible in 6–8 weeks.
Defensibility	10	4/10	Weakest axis, honestly. Execution moat only. The domain encoding (carrier formats, class codes, state procedures) compounds and the agent relationships stick, but a funded competitor could replicate this in six months. No data moat at the start.
Total	100	76/100

13. Qualitative modifiers

Founder-fit tags

sales-heavy · domain-expertise-required

This is not a build-it-and-they-come product. It needs someone who will sit in ABC chapter meetings and work insurance agents, and it needs real fluency in comp audit mechanics — either a founder from the insurance/construction world or a co-founder who is. A pure technical team will build the parser and never find the customers.

Key assumptions to validate (3–5)

1. Assumption: A meaningful share of audit back-charges are genuinely reversible — i.e. the subs did have coverage and the GC just lacked the paper. How to test: Get 15 real audit worksheets from contractors (via 3 friendly agents) and manually work them. Measure what percentage of reclassified dollars can be documented away. If it’s under 30%, the value proposition collapses.
2. Assumption: Contractors will pay a contingency fee rather than fight the audit themselves or just pay it. How to test: Offer the service manually to 20 contractors mid-dispute-window. Count how many sign. Ten signatures is a green light.
3. Assumption: Recovery customers convert to the $149/mo prevention subscription at 30%+. How to test: Can’t test pre-launch — track it from customer one. This is the ARR assumption; if conversion is under 15%, the business is a services firm with lumpy revenue, not SaaS.
4. Assumption: Insurance agents will refer rather than see this as a threat to their relationship. How to test: Pitch 20 independent agents with construction books. If fewer than 5 want in, distribution needs rethinking.
5. Assumption: Carrier audit worksheets are parseable at acceptable accuracy across the top 5–8 contractor carriers. How to test: Collect 30 real worksheets and measure extraction accuracy before writing production code.

Risk flags

1. Regulatory / licensing risk: Depending on how the service is presented, some states may view fee-based representation in an insurance dispute as public adjusting or unlicensed insurance advice. Mitigation is structural — the customer files their own dispute and the product prepares documentation — but this needs a real legal opinion in each launch state, not a guess. This is the risk most likely to force a business-model change.
2. Contingency revenue is lumpy and collection is hard: Getting paid 15% of a recovery requires knowing the recovery happened and the customer paying after they’ve already got their money back. Expect collection friction. May need to shift toward flat fees, which weakens the “no win, no fee” pitch that makes the sale easy.
3. Episodic pain, subscription pricing: The customer feels acute pain once a year. Retaining them at $149/mo for eleven months of quiet is the central business-model risk. The prevention product must produce visible ongoing value (collected COIs, endorsement gaps flagged) or churn will be brutal.
4. Low defensibility: Nothing here is technically hard to copy. The bet is that the market is too small and too unglamorous for a funded competitor to prioritize — which is exactly this portfolio’s thesis, but it’s still the thinnest axis.
5. Carrier relations: At scale, carriers may notice a tool systematically reversing their audit findings and tighten documentation standards or dispute procedures in response. Unlikely to be fatal, but it could raise the bar over time.

14. Structured verdict

Score:                  76/100
Verdict:                GO
Confidence:             Medium
Best-fit builder:       Sales-driven founder with insurance or construction-ops background,
                        paired with one strong technical builder
Time to revenue:        6–8 weeks (manual service delivery before product completion)
Capital to launch:      $15–25K (inference, APIs, multi-state legal review)
Top 3 assumptions to validate first:
  1. Reversibility rate — manually work 15 real audit worksheets, measure the % of
     reclassified dollars that documentation can eliminate. Need >30%.
  2. Willingness to pay — offer the service manually to 20 contractors inside their
     dispute window. Need 10 signed.
  3. Agent channel — pitch 20 independent agents with construction books.
     Need 5+ willing to refer.
Kill criteria:
  - Abandon if manual review of 15 real audit worksheets recovers less than 30% of
    reclassified dollars on average — the core promise is then false.
  - Abandon if fewer than 5 of 20 contractors inside an active dispute window will
    sign a $500 + 15% engagement.
  - Abandon if legal review in 2 of the first 3 target states says the fee structure
    requires an adjuster or producer license.
  - Abandon if recovery-to-subscription conversion sits below 15% after 40 customers —
    pivot to a pure services firm or exit.

15. Next step — 1-week validation sprint

• Day 1–2: Recruit 3 independent insurance agents with construction books (LinkedIn + local Big “I” chapter). Ask each for 5 anonymized client audit worksheets from the last 18 months where a back-charge was applied. Target: 15 worksheets in hand.
• Day 3–4: Manually work all 15. For each reclassified line, determine whether the sub plausibly had coverage and whether documentation could eliminate the charge. Produce a single number: average recoverable percentage of reclassified dollars.
• Day 5: Call 20 contractors currently inside a dispute window (sourced through the same agents). Pitch $500 + 15% of recovery. Count signatures — real ones, with a signed engagement letter, not “sounds interesting.”
• Decide go / no-go on: average recoverable share of reclassified dollars >30% across the 15 worksheets, AND ≥10 of 20 contractors signing an engagement letter. Both must clear. Either one failing kills it.

Both outcomes are falsifiable and both are measurable in five days. If the recoverable share comes in at 12%, this idea is dead and I’ve spent a week instead of six months.