How I Brought My First Product to Market - Idea to Small Business
TL;DR
A mechanical engineering grad with no business experience designed, manufactured, and shipped a portable heated massager for TMJ disorder — going from a notes-app idea to $50K/month in revenue with ~$25K in savings, no investors, and no prior PCB or business experience. The video is a full end-to-end account of every major decision, mistake, and lesson along the way. ---
Key Concepts
TMJ Pen
tap to reveal ↩
A portable electromechanical device combining hot-pack-level heat and vibrating massage pressure, designed specifically for TMJ disorder sufferers — a market with virtually no direct competition
Easy mode product selection
tap to reveal ↩
Choosing a product that is small, low-MOQ, shippable cheaply, self-manufacturable, and in an underserved niche with a personal backstory
Block diagram → component list → schematic → PCB layout
tap to reveal ↩
The step-by-step process for designing a PCB from scratch with no prior experience
LCSC + EasyEDA + JLCPCB trio
tap to reveal ↩
A vertically integrated ecosystem for component sourcing, PCB design, and manufacturing — boards for under $5
Crowbar circuit
tap to reveal ↩
A hardware safety mechanism that shorts the battery across a polyfuse when a comparator detects overtemperature, cutting power without permanently damaging the board
General wellness guidelines (FDA)
tap to reveal ↩
FDA framework allowing products to reference specific conditions (e.g., TMJ disorder) without triggering medical device registration, so long as no treatment/cure claims are made
Meta ad strategy
tap to reveal ↩
Dumping all available video content into a single broad ad set and letting Meta's AI optimize, rather than manually segmenting funnels
MVP-first philosophy
tap to reveal ↩
Ship an imperfect product to real customers as fast as possible to get feedback, generate revenue, and iterate — avoid over-engineering before validation
Notes
§Career Background & How the Business Started
- Studied mechanical engineering but forced himself into CS and EE classes to understand full product design
- Applied widely to product design roles; rejected everywhere without interviews
- Joined a 5-person ADU startup in Austin as an intern; received full-time offer of $70K as "Junior Engineer" with 0.25% equity
- Decided the title, pay, and upside didn't justify the opportunity cost vs. going independent
- Moved back to Philadelphia; took a part-time job assembling and packing orders for a friend's product business at $40/hr
- Gave him free workspace and time to prototype his own product on the side
- Key reframe: didn't "quit his job" cold — found a low-risk bridge role that subsidized his time to build
§Idea Selection — Why the TMJ Pen
- Kept a running notes-app list of product ideas; the TMJ pen won out for structural reasons, not just personal ones
- Personal: his TMJ disorder was worsening; jaw locked closed most mornings and eventually permanently
- Market gap: nothing existed that combined massaging pressure with hot-pack-level heat simultaneously
- Checked all "easy mode" boxes:
- Great founder-story for marketing
- Small form factor → cheap 3D printing, low shipping cost
- Electromechanical → low MOQs from suppliers
- Literal painkiller in an underserved niche with near-zero competition
- Fully self-manufacturable from an apartment
§PCB Design From Scratch
- Had EE coursework but had never designed a PCB; used YouTube to bridge the gap ("YouTube-induced confidence")
- Step 1 — Functional requirements: list everything the device must do
- Adjustable heat, adjustable vibration, indicator LEDs, battery-powered, rechargeable
- Step 2 — Block diagram: map each function to a circuit block, then find an IC that performs each block
- Examples: TP4056 for Li-ion charging; MOSFET transistor circuit for motor driving; boost converter to step 3.7V battery up to 7.2V for motor and heater
- Step 3 — Component selection: use datasheets religiously — they specify exactly which external components to use and how to wire them
- Step 4 — Schematic → Layout: place and connect components in EasyEDA; connect pins per schematic
- Software stack: EasyEDA (free, browser-based) + LCSC (component database/seller) + JLCPCB (manufacturer)
- Find part on LCSC → copy part code → search in EasyEDA → symbol and footprint auto-populate
- One-click export to JLCPCB for manufacturing; boards under $5
§Prototyping the Electronics — Key Iterations & Failures
- V1: forgot to add ISP programming pins; had to learn 6-pin ISP programming, Microchip Studio, and binary export from Arduino IDE
- V1 first power-on: smoke — inductor and boost converter couldn't handle the load; wrong component selection
- Subsequent revisions added:
- Correct 6-pin spring-loaded ISP connector
- Battery protection circuitry (eliminated need for pre-protected batteries)
- Voltage reading capability
- ESD protection
- Temperature control problem: PWM-only control meant too-long heat-up time and inability to maintain temperature against skin
- Wanted clever solution: measure heater resistance change via shunt resistor to infer temperature
- Failed — heating element is nichrome, which barely changes resistance with temperature
- Solution: just used a thermistor (resistance drops as temperature rises; voltage-divider readout to microcontroller)
§Safety Engineering — Redundancy for Overtemperature
- Failure modes that could cause tip to overheat were the critical risk (device contacts face)
- Layered safety stack:
- Comparator monitors thermistor voltage; when threshold exceeded, output flips to 3.3V
- 3.3V triggers a triac, which shorts battery to ground
- Polyfuse blows first, cutting all power — then self-resets after current stops
- Result: no single component failure can cause overheating
§Mechanical Design Iterations
- ~190 prototypes over ~2 years; consistent hexagonal pen-form factor throughout
- Early mistake: over-engineered vibration decoupling (motor mechanically isolated from casing) — scrapped for MVP simplicity
- Key design priorities: minimize part count, ease of assembly, ease of manufacturing
- Component retention evolved:
- Snap-fit PCB insertion from front (vs. top) → easier assembly, less component damage risk
- Zip-tie channel embedded inside casing wall → dual-purpose cable management + motor retention (no extra part)
- Eventually: screws retained the cap, tip, and thermistor simultaneously
- Button design went through ~5 iterations; settled on two-color printed button cover (planned graphic panel membrane was deferred as MVP sacrifice)
- Casings switched from horizontal to vertical print orientation → ~50 units per print bed
§Vibration Noise — A Painful Debugging Story
- Noise sources were interconnected: loose battery/PCB rattling against walls, warped print walls, cap fit
- Tried: flipping motor orientation, screwing in motor, gluing motor, foam tape — none fully solved it
- Breakthrough appeared to happen by accident — then discovered the "fix" was actually a prototype motor with a shorter shaft
- Production motors had longer shafts → larger eccentric moment → more noise
- Switched to shorter-shaft motors for all production units
- Residual electrical noise discovered much later: underpowered filter capacitor + PWM frequency too low (audible whine)
§Business Formation & Patent Decision
- Formed LLC and business bank account early; keeps personal/business finances separate for liability protection
- Decided against pursuing a patent; reasoning:
- A patent only grants the right to sue — requires thousands more in legal fees to enforce
- Broad patents are expensive (tens of thousands); narrow patents are easy to design around
- Chinese competitors are practically unsable anyway
- Worst case without patent: fail, lose no extra money, learn a lot
- Worst case with patent: fail, lose tens of thousands at the worst possible time
- Theragun example: had patents, still got copied everywhere, survived by being the premium brand
- Exception cases where patents may make sense: Amazon-primary businesses, investor/licensing plays, genuinely novel deep tech
§Getting First Customers
- Started by posting YouTube videos to TMJ subreddit and engineering subreddits
- Created Instagram account (MyTMJ Relief); initial strategy: post TMJ resources (blog posts, exercise tutorials), then follow-spam commenters on large TMJ accounts
- Bought 200 followers to avoid looking like a ghost account
- Went viral on TMJ subreddit by posting complaint-style content with subtle Instagram links
- DM'd followers individually to solicit pre-orders; first order was a $15 deposit
- Pivoted content strategy from TMJ tips to behind-the-scenes engineering/entrepreneurship journey
- First viral video: ~10,000 new followers
- Trade-off: audience shifted to engineers/entrepreneurs, not TMJ patients
§Crowdfunding Campaign (Indiegogo)
- Launched when ~10K followers and working prototype existed; raised $13K total ($7K in first two days)
- Bootstrapped the campaign: no paid email lists, no consultants
- Tactics: personal DMs to every TMJ Instagram account, text blast to 4 years of contacts, LinkedIn post, Reddit post (moderator-approved), two YouTube videos, influencer shares from TMJ accounts
- Positioned as a patient-built product for an underserved market — made genuine relationship-building easier
§FDA Encounter
- On day 3 of Indiegogo, received FDA email claiming marketing of a medical device without approval/clearance
- Had previously researched FDA classification:
- Class 3 (implants, pacemakers): requires FDA Approval — very expensive
- Class 2 (mid-risk): requires 510(k) clearance — hundreds of thousands of dollars
- Class 1 (low-risk): exempt from both, but must register annually (~$10K/year fee) and follow labeling/manufacturing guidelines
- Closest predicate categories: "electric therapeutic massager" (Class 1, exempt) or "powered heating pad" (Class 2, but also 510(k) exempt)
- FDA email incorrectly cited approval/clearance requirements — those don't apply to Class 1 exempt devices
- Key legal escape hatch discovered: FDA General Wellness Guidelines
- Products can reference specific conditions (e.g., TMJ disorder) if marketed as supporting a healthy lifestyle choice that "may help living with" the condition
- Cannot claim to treat or cure — language like "may relieve tension" is acceptable
- Most competing heated massagers use exactly this framing and don't appear in FDA databases
- Wrote a detailed justification response; FDA replied asking only for minor wording changes (adding "may" before wellness claims)
- Net result: no registration required, actual FDA correspondence on file as evidence of compliance
§Scaling with Meta Ads
- Started running Meta ads in December after batch 2 arrived
- Scaled from $20/day → $300/day over ~6 weeks
- Best-performing strategy: one broad ad set with all available video content, full control given to Meta AI
- No manual funnel segmentation
- Best performers included UGC from TMJ-space doctors and videos that didn't explicitly pitch the product
- Hit $25K revenue in December, $50K in January
§Current State & Batch 3 Preparation
- Paused ads and shipping to finalize batch 3 (1,000 units)
- Fixes going into batch 3:
- Fixed underpowered filter capacitor (electrical noise)
- Fixed PWM frequency (audible motor whine)
- FCC certification completed
- Fixed USB-C charging (previously only worked with USB-A to USB-C cables)
- Reduced LED brightness (added resistors)
- Switched to rubber-insulated thermistors (previous enamel-coated wires could short against aluminum tip)
- Side projects during break: writing a free TMJ book as a lead magnet; testing Formlabs Form 4 resin printer for a limited transparent edition
- Goal: $100K/month, then outsource manufacturing/fulfillment/marketing to free time for new product development
Actionable Takeaways
- Pick a product on "easy mode": small, cheap to prototype, low MOQ, niche underserved market, personal story — eliminates nearly all upfront capital risk
- Don't quit cold: find a bridge job that pays the bills and gives you time and space to prototype before going all-in
- Learn PCB design via the EasyEDA + LCSC + JLCPCB stack — start with a functional requirements list, build a block diagram, then find ICs that map to each block
- Read datasheets: they specify every external component, value, and wiring detail — most schematic work is just following the datasheet
- Design safety redundancy in layers: identify failure modes that could harm users and ensure no single component failure can trigger them
- Ship an imperfect MVP: waiting for perfection delays the customer feedback loop and cash flow that fund improvements
- Research FDA general wellness guidelines before building any health-adjacent product — the framing of your marketing claims determines your regulatory burden more than the hardware does
- Skip the patent (usually): redirect that capital toward product improvement, marketing, and customer service — the ability to out-compete is more durable than a piece of paper
- Post every day, then let the algorithm tell you what works: organic social volume feeds both follower growth and Meta ad creative testing
- Give Meta full creative control: dump all content into one broad ad set and let the algorithm optimize rather than guessing at funnel segmentation yourself
Quotes Worth Keeping
“
If the first product you ship is perfect, you're shipping way too late.
“
Find a job that gives you the time and supports your ability to build on the side. Only when you have a minimum viable product you're ready to ship should you go all-in.
“
A patent doesn't shield you from anything. All it does is give you the right to sue someone that allegedly copied your invention.
“
If you can't out-compete a knockoff, maybe just humble yourself and realize you just need to learn how to become a better businessman.
“
Selling my first ever product so well that it starts getting copied is already a huge win.