In this essay, Paul Graham tells why startups founders should go for unscalable things at first.

RECRUIT — Recruiting users manually is the most common one. When at Y Combinator, Stripes founders were used to get people try their beta on the spot: “Give me your laptop”. As absolute numbers are small, measure progress by weekly growth rate.

FRAGILE — To locate the most promising vein of users, go for a untargeted launch, observe which kind of users seem most enthusiastic, and seek out more like them.

DELIGHT — As startups can provide a level of service no company can, try hard to make your initial users happy.

EXPERIENCE — It's not the product that should be insanely great, but the experience of being your user. Even with an early, incomplete, buggy product, make the difference with attentiveness. It's the best way to start a very positive feedback loop.

FIRE — Sometimes, focusing on a deliberately narrow market is the right thing to do. It's always worth asking if there's a subset of the market in which you can get a critical mass of users quickly.

MERAKI — Hardware startups should assemble their own hardware. Pebble did.

CONSULT — B2B startups could pick up a single user and act as consultant building a specific product. Fit its needs perfectly, and you'll usually find you've made something other users want too. But, to retain your freedom, make sure not to be paid.

MANUAL — Some startups could be entirely manual at first. If you can, do it as long as you can.

BIG — Launches and partnerships usually don't work as a way to get growth started.

VECTOR — Try thinking of startup ideas as pairs of what you're going to build, plus the unscalable thing(s) you're going to do initially to get the company going.

Source: http://paulgraham.com/ds.html