Market pull: which startups win

Why this matters

After reviewing thousands of startups through our fund Julian Capital and through Deep Checks, we keep coming back to one pattern. The biggest tech companies rarely win because the founders were the smartest people in the room. They win because they caught market pull and rode it.

Market pull is when people reflexively want a product the moment they hear about it, and they'll do whatever it takes to get it. It's the force behind “product-market fit.” When you have it, the market pulls the idea out of your hands as fast as you can talk about it. You don't slog to sell.

Paul Graham put the extreme version well: “If you wonder how someone who seems crazy or incompetent could be the CEO of a successful startup, the answer is that some startups are so viral that they'll grow (for a while at least) almost regardless of who's in charge.” (Paul Graham)

You don't want to rely on being a genius operator. You want to pick an idea where the market is already leaning in.

What creates market pull

Market pull shows up when three things are true about the people you're selling to:

• Comprehension: they intuitively get the value without a long explanation. (This kills the need for heavy sales.)
• Value: they see enough value to justify the labor, price, and risk of switching.
• Enthusiasm: they want it now. Immediacy is what overcomes the friction of switching.

The opposite is market push, where you're constantly convincing people they should want the thing at all. That's swimming upstream, and it usually means the idea is low-ROI, commoditized, or too early.

One important nuance for deeptech: you don't need market pull to exist today if you're confident it will exist soon. A lot of hard-tech is built slightly ahead of its market on purpose. What you can't afford is to be permanently early with no reason for the world to change.

The categories where pull tends to live

You don't have to fit one of these, but each one is a proven shape of demand. The closer your idea maps to one, the better your odds. The ones that matter most for deeptech founders:

1. Unlock a new market or line of business for your customer. You give customers a capability they simply couldn't have before. Varda manufactures pharmaceuticals in orbit that can't be made on Earth. Aalo provides on-site nuclear power for data centers that otherwise couldn't get power. Starlink lets airlines sell in-flight internet. You're not competing for budget. You're creating it.
2. Democratize something previously reserved for the few. Deeptech is uniquely good at order-of-magnitude cost reductions. Waymo gives anyone a private chauffeur. Starlink brings high-speed internet anywhere on Earth. When you take something only the wealthy or well-positioned could access and make it mass-market, the demand is already there waiting.
3. Remove expensive, error-prone labor for a low cost. Robotics, AI, and sensing that automate labor-intensive or dangerous work in labs, factories, and the field. When you remove a lot of work and save real money, adoption becomes a no-brainer, especially where labor shortages or compliance create urgency.
4. Lower cost and increase convenience without lowering quality. If customers already buy the category, an equally good product that's dramatically cheaper is an easy switch. SpaceX made launch orders of magnitude cheaper. Solugen makes industrial chemicals from enzymes instead of hydrocarbons, so they're cheaper and cleaner. One caveat: businesses are more risk-averse than consumers, so you usually need a bigger cost delta to move them.

For the fuller deeptech treatment of these categories, see our Prove-Demand guide. A few additional general-market categories, like marketplaces, viral SaaS, and democratizing investment access, are covered in Julian's Market Pull guide. Deeptech founders most often land in the four above.

How to pressure-test your idea

The fastest gut-check is the 100-customer deposit test: ask at least 100 representative buyers how likely they are to put down a cash deposit, and count only the 9s and 10s. It has its own lesson, so we cover the mechanics there. For hard-tech that's years from shipping, the equivalent signal is a signed LOI. We’ll outline this methodology further in Lesson 2.6.