Who Died on This Road?

Single-source supplier for custom Bluetooth chips. When the supplier failed QA, Jawbone had zero alternatives. $930M couldn't fix a supply chain SPOF.

Bus Factor applies to suppliers, not just people. One vendor = one point of failure.

Built $30M automated warehouses in 10 cities before proving a single neighborhood could generate enough orders to cover delivery cost. Unit economics were negative at every scale — bigger warehouses just lost money faster.

Prove unit economics in one location before you build infrastructure for fifty. Scale doesn't fix negative margins; it amplifies them.

Built for 'in-between moments' that don't exist. Assumed people want 10-minute episodes on phones. People already have TikTok, Instagram, YouTube. No one asked for this format.

Your 'innovation' might be solving a problem no one has. Validate the job-to-be-done before the product.

Sold $10/month unlimited movie tickets while paying theaters full price ($12-15 per ticket). Claimed data monetization would cover losses. Average subscriber saw 2.4 movies/month — each one cost MoviePass ~$20 in losses. Lost $40M/month at peak.

A business model where you lose more money on every customer you add is not a growth story. It's a countdown. Data play is not a substitute for viable unit economics.

Cleaning service with 1099 contractors. Good cleaners left (platform took too much). Bad cleaners stayed. Quality collapsed. CAC spiraled. The delivery chain had no controllable execution node.

If you don't control the quality of your core deliverable, your business is a reputation time bomb.

$400 Wi-Fi-connected juicer that squeezed proprietary $7 packets. Bloomberg showed you could squeeze the packets by hand faster than the machine. The value prop wasn't juice — it was DRM for produce.

If your hardware adds no value over the manual alternative, you don't have a tech company. You have a packaging company pretending.

Sold a 'bezel-less phone' to tech enthusiasts. Enthusiasts care about ecosystem, not bezels. iPhone users had iMessage, AirPods, Apple Watch. A single hardware feature can't overcome ecosystem lock-in.

If you're competing against an ecosystem with a feature, you've already lost. Compete on a different axis or don't compete.

Promised 200+ tests from a finger prick. Physics says you can't. Instead of saying 'this is hard, let's start with 5 tests,' they faked the data. The tech was sci-fi, not engineering.

If your core technology requires violating known physics, it's not a startup — it's fraud. Score your tech honestly at the feasibility stage.

Claimed AI-powered automated bookkeeping. Forbes investigation revealed most 'AI' work was done by human accountants in the Philippines working overnight shifts. Customers churned when books were wrong. The AI story was a staffing arbitrage with a UI.

If your 'AI' requires a human army behind the curtain, you don't have a tech company. You have a staffing agency. The margin profile will never converge to software.

Biofuel from wood chips promised 67 gallons per ton. Lab achieved 40 under ideal conditions. Commercial plant achieved <10. The physics scaled down, not up. Every investor presentation showed the lab number, not the plant number.

Lab results are not production results. If the yield curve declines with scale, you don't have a scaling problem — you have a physics problem.

One-click checkout as a service. Required 1,000+ employees burning $10M/month. Revenue was <$600K/year. The core product (checkout button) added marginal value over Shopify/Stripe native checkout. Zero defensibility, zero path to positive unit economics.

Headcount doesn't equal progress. If your product is a feature, not a platform, no amount of funding turns it into one.

Legal tech startup founded by a non-lawyer. Built software to 'disrupt' law firms, then discovered lawyers bill by the hour and hate efficiency tools that reduce billable hours. Customers were the very people incentivized against the product.

Understand who profits from the current system before you try to disrupt it. If your value prop threatens your customer's income, they're not your customer.

Crowdsourced invention platform. Community voted on what to build. Produced 400+ products in 5 years. Exactly zero became profitable. The crowd loves clever ideas but doesn't do market sizing, COGS analysis, or distribution planning.

Democratizing decisions doesn't democratize accountability. If nobody owns the P&L, nobody makes hard calls. Voting is not validation.

P2P used car marketplace that inspected, photographed, and guaranteed every car. Spent $3,000 per car on reconditioning. Average transaction generated ~$800 in revenue. Every car sold was a ~$2,200 loss. 'We'll fix margins with scale' — they didn't.

Marketplace businesses need network effects to reduce cost-per-transaction. If costs scale linearly with volume, you don't have a marketplace — you have a low-margin service business.

Beautiful design. Better discovery than Spotify. Better social features. But no free tier. Spotify's freemium funnel converted 26% to paid; Rdio couldn't get users to try it. A superior product that lost because distribution beats UX.

The best product doesn't win. The product that's easiest to try wins. If your competitor has a free tier and you don't, your UX advantage is invisible.

Virtual assistant service. Grew from 0 to 400 employees in 4 years. Ran out of cash overnight when a financing round fell through. Had 72 hours of runway and no contingency plan. Employees found out via text message.

Growth without a cash buffer is a suicide pact. If 72 hours of delay kills your company, your operational planning was already dead.