How to find and exploit the "glitch" that can make any business take off Summary — My First Million

Jeff Harmon, founder of Harmon Brothers ad agency and Angel Studios, shares how he repeatedly found “glitches” — overlooked attention arbitrage opportunities — to build billion-dollar brands (OraBrush, PooPourri, Squatty Potty, Purple Mattress) and a top-10 movie studio (Angel Studios, $500M+ revenue) with no industry background.

In his final university class, Dr. Bob asked students how to sell his tongue cleaner (OraBrush); the class said only 8% would buy online, so partner with P&G or J&J.
Harmon countered: 8% of the U.S. is ~30 million people — enough to clear the garage; Dr. Bob hired him on the spot.
First move: bought a Facebook page called “Kisses” (1.3M followers) for $5K, renamed it OraBrush, spammed bad-breath/kissing content — Facebook changed rules, sales failed.
Google ads cost $3–5 per $1 revenue; the breakthrough came from YouTube’s new ad platform (cost <$0.01/view), giving a profitable “beachhead” to optimize from.
Harmon combined door-to-door sales skills (potatoes, LDS mission in Ireland) with a viral video format inspired by iJustine’s jump cuts and a Howcast “spoon test” for bad breath.
Produced a $500 video (“How to Tell If You Have Bad Breath”), offered free brush + paid shipping (shipping = COGS), scaled to ~$10M exit.

User Method: solve a problem you personally have; most breakout companies (Airbnb, OraBrush) started this way — Dr. Bob invented OraBrush because his missionaries in the Philippines had terrible breath.
Attention arbitrage: look where attention is flowing but competitors aren’t spending; example: agentic shopping via Grok/Tesla bots — early movers can plug into APIs before incumbents adapt.
Incumbents’ innovator’s dilemma: established players allocate 1–10% to new channels; a newcomer can allocate 100% and redesign the entire experience for that channel (e.g., Angel Studios using AI tools Hollywood rejects as “low brow”).
Conviction on small signals: millions downloading clean “Game of Thrones” torrents proved demand for filtered content, even though a random sample would show near-zero awareness.

Lock stakeholders in a cabin for 2–3 days with creatives, attorneys, decision-makers — no one leaves until a production-ready script is signed; avoids months of email neutering.
Hire comedians, not marketers: easier to teach a comedian marketing structure than a marketer humor; comedians provide left-field premises that still serve the message.
Structure first, jokes second: problem → solution → 3 proof elements → credibility → CTA; every joke must advance the message or it’s cut.
Pre-retreat prep: 3–4 writer-comedians receive product weeks early, use it, bring rough scripts; Day 1 = script reads; pick winner by evening; Days 2–3 = cannibalize best bits into lead script.
Still works for complex offers: used for Angel’s “Runner” campaign (Alan Ritchson, Owen Wilson, ~100M views) to explain the 2-tickets/month membership model; but theatrical launches (monthly $10–50M startups) require additional tactics.

Origin: wanted to watch “Game of Thrones” without sex/nudity/rape scenes; discovered millions illegally downloading clean torrents — built a legal version (VidAngel).
The Chosen pilot as 19-min YouTube ad: uploaded full pilot as skippable ad; “if you watch to the end, crowdfund us” — 3% completion yielded 17,000 investors, $11M raised on $800K ad spend (~8% cost of raise, comparable to Wall Street fees).
Live-stream telethon + social proof: final 48 hours, Dallas Jenkins marathon stream; real-time investor pop-ups (“$100 from Idaho”) created momentum.
The Chosen outcome: 200M+ viewers worldwide, 7 seasons, top-5 global series; Angel released first 3 seasons, now on Amazon.
Guild model: 3M members pay $12–20/mo → vote on every film (veto power), get 2 theatrical tickets/mo, give feedback on rough cuts; filmmakers get 50% of profits, Angel reinvests 50% → $300M paid to filmmakers to date.
Hit rate: Angel averages higher box office per film than any independent studio (incl. A24), most theatrical locations, top-10 U.S. studio 3 years running; “base hits business with occasional home runs” (Sound of Freedom: $14M production, $40M marketing, $250M global box office).

Typical indie film: wealthy entrepreneurs self-fund (~1,000/year); most never find distribution; Lionsgate only produces ~35%, acquires the rest.
Angel’s edge: Guild de-risks acquisition — members vote, attend, provide marketing flywheel; Angel takes only profit split (no 12.5–30% distribution fee off gross).
Membership math: average member attends 3–4 films/year (breakage covers ticket costs); marketing + member acquisition are main costs (Angel lost $50M EBITDA last year while scaling).
Hollywood’s brutality: retired exec who worked Vegas mob casinos and Hollywood said “my best friend in Hollywood was worse than my worst enemy in Vegas” — knife fight culture; Angel’s mission: rewrite rules via values-based content (anti-nihilism, PG-13 action/westerns/faith 2×/year).

Kevin James / Matt Taylor campaign: Kevin created Instagram/TikTok art-teacher character (Bob Ross style); team lived with him on Long Island producing content.
Face-hashing warm-up: Kevin reposted old clips to “heat up” his face in TikTok/IG/YouTube algorithms; fresh Matt Taylor channel instantly went viral because algos recognized Kevin’s face + trending status.
Clipping at scale: similar to Tate/Clavvicular armies — thousands of cuts to find viral winners; algorithm rewards volume + recognizable faces.
Super Bowl stunt: bought 12 seats, put Kevin in tux with flowers in center seat, surrounded by influencers; zero branding, pure mystery → Megatron appearance, hundreds of millions of organic views.

Personal vibe coder: dedicated engineer takes Harmon’s late-night prototypes to MVP; once product-market fit proven, hands to core product team for scaling.
Nightly agent runs: writes plan for Claude, launches agents, leaves laptop running (even monitors from Tesla via hotspot); agents hit permission prompts unless “bypass everything” mode.
Harness system: brand-guide + safety harnesses auto-check every model output (Claude, GPT, Grok); when Harmon breaks something, eng team builds new harness.
Cool build: A/B trailer testing tool — sends randomized trailer links, collects completion/engagement data, identifies winners fast.
LTV/attrition modeling: clean data + AI = instant cohort analysis; Harmon believes AI 10× productivity → hire more people, not fewer (multiplying two numbers: don’t shrink headcount).

Family first: nothing in work compares to marrying early, having kids sooner and more than you think; that’s where real success lives.
Abundance over litigation: mentor Steve Oskoui said unless >$5M, lawsuits drain 18–36 months of innovation energy; brother Neil opened Bible to “if sued for cloak, give coat also” — let go of a few million owed, focused on next thing, built big agency.
Golden goose mindset: you’ll lay more eggs; don’t treat any single idea as your only shot — abundance mentality unlocks more opportunities.