The niche problem

The conventional startup wisdom says: pick a niche. Find a narrow market, go deep, and own it. It's clean advice, and it works, if you already have the connections, domain expertise, and distribution channels to reach that niche. But what happens when you don't? That's the part nobody talks about. The niche advice assumes you already know the territory. If someone tells you to go after the healthcare sector or the logistics industry, the unspoken prerequisite is that you have relationships there, understand the workflows, and can get meetings with decision-makers. Without that, you're standing outside a locked door with a great product in your hands and no way in. I've been thinking about this a lot, and I think the framing is wrong. The real question isn't "should I go niche or go broad?" It's "how do I distribute what I build?"

The distribution problem

Most people frame the niche debate as a product question. Should your product do one thing really well, or should it do many things? But the harder question is always distribution. You can build the most targeted solution in the world, but if you can't reach the people who need it, it doesn't matter. Andrew Chen, a general partner at Andreessen Horowitz, makes the case that startups need dual theories, one about product-market fit and one about distribution. Having a great product insight isn't enough. You need a corresponding insight about how to get that product into people's hands. And for most founders, especially technical ones, that second insight is the harder one to come by. Sam Shank, founder of HotelTonight, puts it even more bluntly: don't start with the product, start with distribution. His company was built around the insight that the App Store was about to become a massive distribution channel. The product, last-minute hotel booking, was designed to fit that channel perfectly. The problem with the niche-first approach is that it often ignores this reality. Going narrow only works if you can actually reach the narrow audience you're targeting.

The wide net with targeted distribution

Here's the alternative I've been experimenting with: build something adaptable, then distribute it in a targeted way. Think about how OpenAI operates. ChatGPT is fundamentally a wide net, a general-purpose AI assistant that can help with almost anything. But OpenAI doesn't market it as "the everything tool." They target specific use cases, specific industries, and specific workflows. They've built integrations with Shopify for commerce, rolled out features for education, and launched enterprise plans tailored to business needs. The platform is broad, but the distribution is focused. This is a different mental model from the typical niche advice. Instead of narrowing the product, you narrow the distribution. You build something flexible enough to serve multiple markets, but you sell it one market at a time, adapting your pitch and positioning to whoever you're talking to. For developer tools specifically, this approach makes a lot of sense. The developer tools market is notoriously difficult to distribute in. GitHub Marketplace has over 10,000 apps, but fewer than 5% gain meaningful traction. Acquisition costs are rising, and standing out requires more than just being technically good. You need a distribution edge.

Why pure niche is harder than it sounds

The advice to niche down assumes a level of access that most founders simply don't have. If I wanted to build something for the carbon emissions sector, where would I even start? I don't have contacts in that world. I don't understand their procurement cycles. I don't know what conferences they attend or what Slack communities they hang out in. Domain expertise is a real advantage, but it's not something you can just decide to have. As Nitin Kumar writes, the right move is to treat domain knowledge like any other scarce resource, deploy it where the cost of being wrong is high, and avoid paying too much for it when it can be learned through customer validation. The niche-first path works beautifully when you're already an insider. A doctor building healthcare software, a logistics manager building supply chain tools, a teacher building education technology. But when you're an outsider? The barriers to entry aren't just technical, they're social and structural.

The adaptable product approach

So here's what I'm actually doing. I'm building AI tools that are designed to be adaptable. The core product isn't specific to any one industry. It can be configured and shaped to fit different use cases, different workflows, different types of users. But the distribution is very targeted. Instead of trying to reach everyone, I find specific people I actually know, understand their problems, and show them how the tool fits their world. The pitch isn't "here's an AI tool that does everything." It's "here's how this solves the specific problem you told me about last week." This sidesteps the biggest problem with the niche approach: you don't need to be an expert in someone's field to sell them something useful. You just need to understand their problem well enough to demonstrate that your adaptable product can solve it. And understanding one person's problem is a lot easier than understanding an entire industry. Paul Graham captures this well: it's better to make a few users love you than to have a lot of ambivalent ones. The "few users" don't have to be in the same niche. They just have to genuinely love what you've built for them.

Distribution as the real moat

The startup graveyard is full of great products with no distribution. Brian Balfour, former VP of Growth at HubSpot, argues that product-channel fit is just as important as product-market fit. Your product needs to be designed for the channels you can actually access. Dropbox didn't win because they had the best file sync technology. They won because they built a referral system directly into their onboarding flow that turned every new user into a distribution channel. The product was good, but the distribution mechanism was brilliant. For a solo founder or a small team without deep industry connections, the adaptable-product-with-targeted-distribution model has a few concrete advantages:

The niche will find you

I think the best version of the niche advice isn't "start narrow." It's "start wherever you can distribute, and let the niche emerge." Build something flexible. Sell it to people you can actually reach. Pay attention to which use cases generate the most energy and pull. Then double down on the niche that chose you, not the one you chose from a whiteboard exercise. The conventional path says: pick your niche, then figure out distribution. I think the better path, especially when you don't have domain expertise or industry connections, is the reverse. Figure out distribution first. The niche will follow.

References