I want you to try something. Before you write another line of code, before you design another screen, before you spend another pound on a solution: let an AI interview 100 of your potential customers.
Sounds a bit mad, doesn’t it? But we’re not in 2021 anymore. The days of raising a massive seed round on a napkin sketch and a confident smile are over. The ‘year of efficiency’ wasn’t a year; it’s the new weather system. And in this climate, building something nobody wants isn’t just a mistake, it’s malpractice.
I used to think the 2023 tech correction was a one-off event. A painful, but necessary, reset. Turns out, the trend is continuing with force. As I write this in mid-2024, we’ve already seen over 100,000 jobs cut from tech companies this year alone. The market isn’t just rewarding efficiency; it’s brutally pun
ishing companies that scaled on assumptions.
The single biggest threat to your startup isn’t the competition, and it’s not running out of cash. Those are just symptoms. The disease, the number one killer according to years of research by CB Insights, is ‘no market need’. It accounts for 35% of all startup deaths.
You can have the most elegant code, a beautiful UI, and a world-class team. But if you’re solving a problem nobody has, or one they don’t care enough about to pay to fix, you’re already dead. You just haven’t stopped spending money yet.
The Billion-Dollar Graveyard of Assumptions
If you think this is just a problem for scrappy startups in a garage, let me tell you a few stories about what happens when you have unlimited money but zero validation.
Remember Quibi? Of course you don’t. Hollywood giant Jeffrey Katzenberg and ex-HP CEO Meg Whitman raised an eye-watering $1.75 billion *before launch*. The idea was ‘quick bites’ of premium video for your phone. They spent a fortune on A-list stars, assuming people wanted a new, slick, subscription platform for short-form content. They never seriously tested that assumption. They built it all in a silo. Turns out, people were perfectly happy with free, shareable content on YouTube and TikTok. Quibi shut down in six months, a bonfire of cash that could be seen from space.
Or how about Juicero? A $120 million-funded, Wi-Fi-connected juicer so powerful it could supposedly lift two Teslas. It only worked with their proprietary £5 packets of pre-chopped fruit. The entire premise was built on the assumption that people needed a high-tech, over-engineered machine to squeeze a bag. Then a Bloomberg report showed you could just squeeze the bags with your bare hands and get the exact same result. The company collapsed.
These aren’t just funny anecdotes. They are monuments to the arrogance of building without asking. They validated their own vision, not the customer’s pain.
The Old Playbook: Smart, Scrappy Validation
Alright, so how do the smart ones do it? For years, the answer has been the Minimum Viable Product (MVP). But an MVP isn’t your product with fewer features. It’s the smallest possible experiment you can run to prove or disprove a core assumption.
The 100 Challenge, Ship the Truth
The 100 Challenge was born from watching too many founders pour time, resources and money into building startups that, frankly, nobody needed. Testing your big idea on a sample size of one (even if that one is your mum) isn’t problem-fit. Whether it’s the instant a spark of inspiration hits or you’re halfway down the road to changing the world, this is the most crucial step you’ll ever take.
Here are the hall-of-fame examples every founder should know:
* The Zappos Bet: Back in 1999, Nick Swinmurn wanted to know if people would buy shoes online. Instead of leasing a warehouse and buying inventory, he went to local shoe shops, took photos, and put them on a basic website. When an order came in, he went back to the shop, bought the shoes at full price, and posted them. He was losing money on every sale, but he was proving, with real money from real customers, that his core assumption was correct. That’s a ‘Wizard of Oz’ MVP.
* The Dropbox Video: Building the seamless file-syncing tech for Dropbox was incredibly complex. So, founder Drew Houston didn’t build it first. He made a 3-minute video that *showed how it would work*, filled with in-jokes for the tech community on Digg. The waiting list for his beta exploded from 5,000 to 75,000 overnight. The video was the product. It validated demand without a single line of user-facing code.
* The Buffer Landing Page: Joel Gascoigne had an idea for scheduling tweets. Instead of building it, he made a landing page that said what the tool would do. The only button said ‘Plans & Pricing’. When people clicked it, a second page explained they were too early and asked for an email address. The clicks were a measure of intent. The emails were proof. He had validated the problem before writing any code.
These methods are brilliant. They are the foundation of good product strategy. But they are also slow, manual, and rely on you to interpret the signals correctly. In 2024, we can go faster.
The New Playbook: Validate at the Speed of AI
This brings me back to my opening challenge: let an AI interview 100 customers.
The core of validation is customer discovery. We call it the ‘100 Challenge’ – proving beyond doubt that a painful problem exists by talking to a hundred people who might have it. The goal is to find the root cause of their pain, to map their agony, before you even think about a solution.
Traditionally, this is a gruelling process. You hunt people down on LinkedIn, you beg for 30 minutes of their time, you take messy notes, and you spend days trying to spot patterns in a sea of transcripts. It’s essential, but it’s a bottleneck.
Here’s the thing: you can now use AI to do the heavy lifting.
1. Craft Your Problem Hypothesis: Start with your core assumption. “I believe marketing managers at B2B SaaS companies struggle to measure the ROI of their content.”
2. Automate the Outreach & Interviews: Use AI tools to identify and contact hundreds of potential interviewees. Instead of a 30-minute Zoom, you can design an interactive AI-driven interview that people can complete in 10 minutes. It can ask open-ended questions, probe deeper on interesting answers, and adapt its line of questioning on the fly.
3. Synthesise the Truth: Feed all 100 (or 500) of these interview transcripts into a modern language model. Ask it: “What are the top 5 most frequently mentioned pain points? What words do they use to describe this problem? Are there any surprising frustrations I missed? Group the responses by company size.”
You’ve just done months of customer research in a matter of days. You haven’t just validated a problem; you’ve got the exact language your customers use, the edge cases you never considered, and hard data on which segment feels the pain most acutely.
This isn’t about replacing the founder’s intuition. It’s about giving that intuition superpowers. It’s about shipping the truth, not just your opinion.
A Warning: Validation Is Not a Silver Bullet
Now, this is the crucial part. Even with perfect validation, you’re not safe. The rules of the game have become more complex. The events of the last year have given us some brutal new lessons.
Lesson 1: A validated product isn’t a validated business.
Look at Google. In early 2024, they laid off core engineers from the Flutter and Dart teams. Flutter is a brilliant, widely adopted, and beloved product. It was validated. But in a climate where Google is laser-focused on AI and Cloud revenue, its connection to the core business was deemed too indirect.
The lesson: You must validate that your product makes money, or directly supports the thing that does.
Lesson 2: A validated product isn’t immune to strategy shifts.
The most painful example is Microsoft shutting down Tango Gameworks. A year earlier, Tango released *Hi-Fi Rush*, a game that won BAFTAs and was a critical and commercial hit. It was a perfectly validated product. But it wasn’t a billion-dollar blockbuster like *Call of Duty*. When Xbox strategy shifted, this beloved, validated hit was on the chopping block.
The lesson: You must constantly validate your product’s place within the company’s grand strategy.
Lesson 3: A validated business model isn’t immune to hype cycles.
Stability AI built Stable Diffusion, the foundational model that changed the world. The tech was more than validated. They raised $100 million at a billion-dollar valuation. But the compute costs were astronomical, and turning an open-source model into a profitable enterprise business proved fiendishly difficult. The hype validated the tech, but nobody had validated a sustainable business model. #
The lesson: Validate the commercial model with the same rigour you validate the technology.
Lesson 4: Your most validated asset isn’t immune to leadership whims.
In a move that baffled the industry, Tesla fired its entire 500-person Supercharger team. This network is arguably Tesla’s single greatest competitive advantage—a product so validated that other car giants were paying to use it. A single, top-down decision made for short-term cost-cutting dismantled a decade-long strategic moat overnight.
The lesson: Your product’s survival is ultimately tethered to rational, stable leadership.
Build What Matters
Look, the message here isn’t to be timid. It’s to be smart.
Building a company is an act of faith, but it doesn’t have to be a blind one. The tools are now available to replace blind faith with hard evidence, faster and cheaper than ever before.
Don’t be the founder who builds a beautiful, expensive solution to a problem nobody has. Don’t be the team that builds a beloved product that doesn’t fit the strategy.
Use the new playbook. Use AI to understand your customers at a scale and speed that was impossible two years ago. Find the pain. Get the evidence. Validate the problem, validate the business model, and validate the strategic fit.
Kill your bad ideas before they kill your company. Then, and only then, go and build something they can’t live without.
## Sources & References
- [The Big Picture: This Isn’t Over](https://layoffs.fyi/)
- [The Top Reason Startups Die (And How to Avoid It)](https://www.cbinsights.com/research/startup-failure-reasons-top/)
- [Quibi’s Billion-Dollar Field of Dreams](https://www.wsj.com/articles/quibi-is-shutting-down-what-went-wrong-11603316935)
- [The Cautionary Tale of Juicero](https://www.bloomberg.com/news/features/2017-04-19/silicon-valley-s-400-juicer-may-be-feeling-the-squeeze)
- [When the Product Fails the Physical World: Fisker’s Collapse](https://techcrunch.com/2024/04/29/fisker-issues-another-layoff-notice-to-employees-as-troubles-mount/)
- [The Zappos Bet: Selling Shoes Without a Warehouse](https://www.inc.com/magazine/20060901/hidi-how-i-did-it-nick-swinmurn.html)
Arthur is the AI native startup operating system I’m building in public — not hype, but a system that turns input into structured execution and tracks founder progress. If you want to follow or use it, it’s open for early access.
→ Access Arthur
Enjoy and hope to see you soon.
❤️ & ✌🏼
Thank you for reading. If you liked it, share it with your friends, colleagues and everyone interested in the startup Investor ecosystem.
If you've got suggestions, an article, research, your tech stack, or a job listing you want featured, just let me know! I'm keen to include it in the upcoming edition.
Please let me know what you think of it, love a feedback loop 🙏🏼
🛑 Get a different job.
Subscribe below and follow me on LinkedIn or Twitter to never miss an update.
Spot on advice: validation saves billions. Quibi raised $1.75B pre-launch without ever testing if anyone wanted 10 min commute videos in a TikTok world.
A deep dive on the validation traps and how to avoid them 👇
https://ventureleap.substack.com/p/why-quibi-failed-and-the-three-questions