Quibi, Humane AI Pin, and Juicero all raised significant funding but ultimately flopped. I think many of their biggest risks could have been tested earlier with relatively simple experiments.
- Quibi
- we believe: people are willing to pay $5-8/month for short, high-quality shows that can only be watched on their phones.
- to verify, we will: create a landing page with sample trailers and subscription options, and market it to as many target users as possible.
- measure: the percentage of visitors who signed up for a paid subscription
- we’re right if: at least 5% signed up. This would help validate whether people actually want another paid streaming service before spending millions on content production.
- Humane AI Pin
- we believe: people are willing to pay $699 upfront and $24 per month for a screenless device that replaces some of their daily phone activities.
- to verify, we will: give 20 participants a simple wearable microphone connected to a human operator for two weeks, simulating the Humane AI Pin experience.
- and measure: how frequently participants use the device.
- we’re right if: at least 50% use the device consistently. This would help validate whether people actually will develop a habit of using the product, not just from excitement.
- Juicero
- we believe: customers are willing to pay $400 for a juicer because it produces meaningfully better juice than squeezing by hand.
- to verify, we will: conduct a blind taste test with 10 participants comparing machine-pressed and hand-squeezed juice, and ask them how much more they will pay for the better tasted one.
- and measure: how many participants prefer the machine-pressed juice and would be willing to purchase the machine.
- we’re right if: at least 80% of participants prefer the machine-pressed version and is willing to pay extra for it. This would help validate whether the expensive hardware actually offered enough additional value.
I think the common lesson across all three cases is that building an impressive product doesn’t necessarily mean customers will pay for it. Before investing heavily in development, it’s important to test the riskiest assumptions about actual customer behavior, ideally through the cheap and representative experiments.
