Anatomy of a Venture: From a casual question to a $10K/month plan.

How Old Growth took a one-line observation and developed it, over a series of working sessions, into a shippable consumer software venture — including the moments the original plan died and what replaced it.

The question

It started the way most good ventures start — with a question that didn't sound like a venture.

"Why can't we get the equivalent of a credit report for our health records? The underwriters see it, but the consumer can't order a copy."

Pete brought the observation to Old Growth as a passing thought. Not a business plan. Not a pitch deck. A friction he'd noticed in the way insurance underwriting actually works.

It would have been easy to nod and move on. Most ideas die in that nod. The Old Growth process is built to do the opposite — to take observations seriously, sit with them long enough to understand whether they're describing a real market gap or just a frustration, and then either kill them quickly or develop them rigorously.

This is the story of one of those developments.

First move: map the asymmetry

Before deciding whether there was a business, Pete wanted to know whether the observation was correct.

It was. The Medical Information Bureau (MIB Group), Milliman IntelliScript, and ExamOne ScriptCheck are nationwide consumer reporting agencies under the Fair Credit Reporting Act. Insurers pull them on every individually-underwritten life, disability, long-term care, and critical-illness application. Consumers have a federal right to request the same files — but the process is paper-form, weeks-long, and the reports come back opaque.

Errors in those files cause real harm. Pete's research surfaced specific cases: a misplaced ICD code can move a consumer from Standard rating to Table 4 underwriting — a difference of $10,000 to $25,000 in premium over a 20-year term policy. The MIB itself paid $2.425M in a 2025 class-action settlement for failing to disclose its data sources properly. Milliman IntelliScript has its own active class action over accuracy.

The asymmetry was real. The harm was real. The question was whether a startup could earn the right to fix it.

The first plan (and its hidden flaw)

Pete drafted an initial operator's plan — "OpenFile" — built around the obvious wedge: see what life insurance underwriters will see about you, before you apply, while you can still fix it.

A $39 software kit. Consumer pulls their own files. AI parses them. Generates dispute letters they file themselves. The legal posture was deliberately narrow: consumer-only data flow, no third-party furnishing, no insurance license required. Operating breakeven projected for month 9.

The economics looked clean enough to take seriously. Twelve-week MVP. $30-50K bootstrap. $86K Year 1 revenue at the realistic scenario.

We could have stopped there. Many advisory engagements do.

The walkthrough that killed v1

The Old Growth process doesn't trust plans that haven't been customer-tested. Before recommending a build, Pete role-played as customer #001 — walking through the product flow as if he'd just landed on the homepage.

Five minutes in, the plan broke.

The MIB consumer file is only populated if you've previously applied for individually-underwritten insurance in the last seven years. Milliman IntelliScript only contains data if an insurer has previously pulled your prescription history with your authorization. First-time applicants — roughly half of the target customer in the original plan — have empty files.

The whole pitch — "see what underwriters will see" — delivered nothing useful to a first-time applicant. Their MIB file would be empty. Their IntelliScript file would be empty. The $39 product would feel like a refund waiting to happen.

This was the moment most casual ideas die quietly. Refile didn't.

The kill criterion that fired was one Pete had explicitly written into the original go/no-go memo: "if the answer to 'what do I show them?' is mostly empty fields, it's not a product." Pre-committing to that test before the walkthrough was what made the failure useful instead of devastating.

The plan needed to change, not the founder's resolve.

The pivot: narrow the wedge

Most ventures, at this point, get widened. Pete went the other direction.

If the product didn't work for first-time applicants, it had to work for people who'd already been through underwriting — people whose files were guaranteed to have content worth showing. That meant:

• 60% of the target customer: adults declined or rated up on a life insurance application within the last 12 months
• 25%: existing policyholders shopping replacement coverage
• 15%: lapse re-applicants

Smaller market. Vastly higher motivation. The previously-declined applicant has a concrete, painful, expensive problem they're already trying to solve. The first-time applicant has a vague intent to maybe get insurance someday.

The positioning shifted from a generic transparency pitch to something concrete:

Got declined for life insurance? Find out exactly why, fix it, and apply to the right carrier next time.

Price doubled to $89 base. Blended ARPU projected at $120 once add-ons attached. A BGA partnership channel — referring pre-vetted declined applicants to impaired-risk brokerages — pushed the all-in per-user revenue toward $200-250.

The v2 plan looked stronger than v1 on every dimension that mattered. We could have shipped that too.

The pressure test

Old Growth's discipline at this stage is to run parallel reviews before committing capital. Three independent streams in 90 minutes:

Legal review

A focused regulatory critique of the v2 features against US insurance, FCRA, FTC § 5, and state UDAP regimes. The findings:

• The marketing claim "find out exactly why you were declined" was unsubstantiated under FTC § 5 — you produce a probabilistic inference, not a fact
• Florida statute § 626.112(8) caps unlicensed referral compensation at the lesser of $50 or 5% of commission — the $100-150 BGA referral fee was illegal in Florida
• The carrier-matching feature crossed a soft regulatory line in New York and Florida around solicitation
• And the combination of carrier-matching plus referral fees was, in the reviewer's words, "the pattern regulators look for"

Competition analysis

No self-service consumer-pays product existed for MIB or IntelliScript disputes — the whitespace was real. But two FCRA plaintiff law firms (Francis Mailman Soumilas and Consumer Attorneys PLLC) were already productized around the exact wedge, taking contingency-fee cases. A cluster of condition-specialist brokers owned the SEO real estate for "declined for life insurance" queries. The threats weren't direct competitors; they were partner-or-incumbent dynamics that needed addressing.

Brand and domain

The placeholder name "OpenFile" was generic. Twenty-five candidates were scored against memorability, brandability, fit to the use case, and trademark cleanliness, with .com availability checked authoritatively against the Verisign registry. The recommendation: Refile. A real English verb, action-oriented, on-strategy. RefileHQ.com and Refile.co were both available.

The honest strip-down

This is where most planning processes flinch. The legal review surfaced real issues. The competition analysis surfaced real complexity. The instinct is to bolt on more features, more disclaimers, more partner conversations.

Pete went the other direction again.

He pushed back: "Even FCRA litigation referrals require legwork beyond what we can build now and launch for revenue."

He was right. Plaintiff law firm referrals are structurally cleaner than insurance broker referrals — bar rules are clearer, per-case economics are higher — but they still require relationship building, intake protocols, and bar-rule fee-sharing review. Anything partner-dependent puts the launch on someone else's calendar.

If the goal was cash flow as fast and cleanly as possible, the right move was to strip every partner dependency out of the plan. No BGA referrals. No FCRA litigation referrals. No carrier-matching feature requiring legal opinion letters in FL and NY. No state-by-state compliance work.

Just the consumer, the files, the dispute kit. Software margins. Day-one payback.

The plan you'd actually ship

The third draft — v3 — is the version that ships.

The product: A web app that walks a recently-declined life insurance applicant through pulling their MIB, IntelliScript, and MVR files. AI parses what comes back. Generates a decline-rationale worksheet (carefully worded to stay defensible under FTC § 5) and a templated FCRA § 1681i dispute packet that the consumer files themselves.

The brand: Refile. RefileHQ.com and Refile.co reserved. ITU trademark filing planned within 30 days of name commit.

The economics

The startup cost

The revenue ramp

Operating breakeven at month 5. Founder full-time at month 9. Year 1 cumulative direct revenue: ~$98K. No partners required. No referral channels. No carrier-matching infrastructure.

Just software that does one job, well, for people who have a concrete problem and $89.

What this process looks like

The Refile case study is what an Old Growth engagement looks like at its best: a real observation, taken seriously, developed through three rounds of structured pressure, and stripped of every assumption that couldn't survive scrutiny.

The pattern, repeated across engagements:

1. Start with the observation, not the answer. Most ideas get over-defined before they've been understood. The first job is to map the actual market structure — who's there, what's broken, where the asymmetries live.
2. Pre-commit to kill criteria. Write them down before you fall in love. Make them specific. "Hallucination rate >5%" and "refund rate >15%" and "no BGA willing to pay $100/lead in writing." When one fires, you stop, regardless of how attached you've become.
3. Stress-test by playing the user. Walk through the product as customer #001. Most plans break inside five minutes when you do this. The faster they break, the cheaper they break.
4. Pressure-test in parallel. Legal, competitive, and brand reviews run concurrently, not sequentially. The interactions between findings are where the load-bearing assumptions live. The carrier-matching feature looked fine alone; combined with referral fees it looked like an enforcement-action diagram.
5. Be willing to strip rather than expand. When the pressure test surfaces real risks, the brave move is usually to remove features, not add disclaimers. v3 is shorter than v1 was. It's also the version that actually ships.
6. Ship something concrete. A $10K/month software business is not a unicorn. It's a real cash-flow venture, controllable by the founder, defensible on its own legal posture, and a credible Year 2 launching pad to bigger products. We build for the version that earns first.

What's next for Refile

The next 30 days for Refile are deliberately not about code:

1. Stand up a pre-order landing page at $89 — validate willingness-to-pay before the build budget commits
2. Engage FCRA-savvy insurance regulatory counsel for the non-CRA scoping memo
3. Commit to the Refile brand and grab the domains
4. Publish the first five SEO articles, optimized for decline-intent queries

If the pre-orders come in, Refile gets built. If they don't, $5,000 of testing has saved a $45,000 build budget. Either outcome is a win.