← All links
Theme
Using automatic theme
Back to top

Article notes

Do Things that Don't Scale

By Paul Graham

Read the original

Listen

Audio version

A direct reading of the notes, with clickable timestamps throughout the article.

Total length: 6:20
6:20 remaining

In one sentence

In the early life of a startup, the founder’s job is not merely to build a scalable product but to create momentum through labor-intensive actions that would be impractical at scale. Manual work is often the fastest route to finding users, learning what they need, and discovering what should eventually be automated.

Overview

Graham challenges the “better mousetrap” assumption that a good product will automatically attract users. Startups are initially fragile and usually need a deliberate push. Founders should recruit users personally, serve them unusually well, focus on a narrow group, and do manually whatever is required to make the product useful. As demand grows, these practices can gradually be systematized. The essay contrasts this approach with the seductive but usually ineffective “big launch” or major partnership.

Core ideas

Founders create initial momentum

A startup is more like an engine that needs cranking than a projectile that only needs launching. Early growth generally comes from founder effort, not passive exposure or product quality alone.

Recruit users manually

The first users are usually found through direct outreach, personal introductions, sales, and hands-on onboarding. Early numbers may look insignificant, but a sustainable weekly growth rate compounds dramatically.

Treat early users exceptionally well

Small companies can offer a level of attentiveness that large companies cannot: personal help, rapid responses, customized service, and thoughtful gestures. This both retains users and reveals what the product should become.

Ship before perfection

An initial product model is almost always incomplete because users’ needs become clearer through actual use. In low-risk domains, launching once there is genuine utility and learning directly is usually better than polishing indefinitely.

Start with a contained fire

A narrow market can reach critical mass more easily than a broad one. Facebook’s early focus on Harvard students made the product feel socially complete; similar concentration can help marketplaces and other startups gain traction.

Use manual work as product research

Founders can assemble hardware themselves, act as consultants for one customer, operate software on a customer’s behalf, or perform supposedly automated tasks by hand. This creates fast feedback and teaches them what automation should eventually handle.

Prefer a real foothold to an impressive appearance

A single user with an urgent problem can be enough to begin shaping a valuable product. Early traction matters more than looking large, professional, or strategically important.

Be skeptical of big launches and partnerships

A launch can provide an initial user base but rarely creates lasting growth by itself. Large-company partnerships are often slow and distracting. Strategies that promise growth without substantial founder effort deserve suspicion.

Practical takeaways

Caveats and counterpoints

Questions worth revisiting

Return to this when…

Return to this essay when evaluating an early startup idea, feeling discouraged by small initial numbers, debating whether to automate too soon, or tempted by a flashy launch, broad positioning, or prestigious partnership. Its central diagnostic is: what concrete, labor-intensive action can the founders take now to make the product useful to real people?

References

  1. Original Do Things that Don't Scale