Of course you can build a better search engine than Google

Doing so many projects, we often get asked how we find ideas to work on. While the answer is technically the same as any entrepreneur — we look for places where markets are operating inefficiently or customer needs aren’t being met — we thought it might be interesting to offer a few specific techniques we use repeatedly. We’ll talk about niches today, and if there’s enough interest we’ll cover other approaches in the future.

Often when people try to figure out what problems are worth working on they dismiss areas where there’s a large incumbent doing a decent job, on the theory that it’s going to be hard to do what a company like Google does better than they do. That might be true, but that doesn’t mean the whole sector should be ignored. Inherently, large incumbents can’t serve every niche well, and in a big market even niches can sustain a worthwhile business.

There’s a bunch of reasons that big companies can’t serve niche markets effectively — niches are too small to give them a big enough return on investment, their fixed costs are too high to make niches profitable, their internal incentives don’t reward staff for exploring new niches, etc, etc. Startups, of course, don’t have these problems.

A good niche has a couple of additional requirements though.

Perhaps obviously, it has to be big enough to sustain the sort of business you want to build. On the surface this can be a difficult requirement — small enough that the incumbent can’t serve it effectively, but large enough that it will support a big business — but you have a little flexibility. The niche doesn’t have to be large enough to sustain your business on its own as long as it can also serve as a way to enter adjacent markets. It’s worth remembering that not everyone has to build a large, venture-scale business either.

You must also be able to deliver a competitive product for your audience. It’s no good building a search engine that, say, offers more privacy than Google if your search results don’t actually answer the user’s questions.

So how can you find a niche worth targeting? One trick is to look for popular complaints about the incumbent. Not only does their popularity imply something about market size, the fact that people are willing to complain in public about it implies that they care enough about the issue to change their behaviour.

All this might sound like moving the goalposts — just redefine the question until you can answer yes! — and that’s exactly right. While it might be a bad thing in a discussion though, it’s exactly what you should be doing here. For people who care about privacy, DuckDuckGo is a better search engine than Google. For people looking for usable images, Creative Commons is a better search engine than Google. For people looking for baseball data, Baseball Reference is a better search engine than Google.

Redefine the question until you find something you can do better than the incumbent, and then figure out how big that market is. If it’s big enough — or maybe even if it can give you a beachhead to move into a larger market — you’ve found something worth doing.