Do things that don't scale to win your first 100 users

Most new products do not grow on their own. Someone has to go and find the first users, one at a time, and make them happy enough to stay. “Doing things that don’t scale” means accepting that early growth is manual work: personal outreach, setting people up yourself, and doing parts of the service by hand. It fits any product before it has a reliable acquisition channel, and it is close to mandatory for marketplaces, which start with nothing on either side.

Why it works

At the start, your numbers are tiny, so each user has an outsized effect. Ten people who love the product give you feedback, testimonials and word of mouth. A thousand people who signed up from a launch post and never came back give you nothing. Manual recruiting lets you choose who those first users are, so you can focus on the people with the sharpest version of the problem.

It also teaches you what to build. When you set up an account by hand, you see every confusing field and every missing feature. When you do a task for the customer behind the scenes, you learn exactly which parts are hard. That knowledge makes your later automation far better than anything you would have designed up front.

How to do it

1. Pick a narrow starting group. Choose one city, one job title or one community where you can reach people directly. A marketplace might start with dog walkers in a single neighbourhood. A SaaS tool might start with bookkeepers who use one accounting package.
2. Build a list of 50 named people. Use your network, niche forums, LinkedIn, local groups and industry events. Write down each person’s name, how you found them and the problem you think they have.
3. Reach out one by one. Send a short, personal message that mentions their situation. Ask for a 15-minute call or offer to set them up. Skip mass email at this stage.
4. Set people up yourself. When someone agrees to try the product, do the setup with them on the spot. Import their data, create their first project or write their first listing. Do not send a link and hope.
5. Do the hard part by hand. If a feature is not built yet, do it manually behind the scenes. Match buyers and sellers yourself, write the report yourself, or fix the data yourself. Be honest if a user asks how it works.
6. Stay close after signup. Check in after the first week. Send a personal thank-you note. Fix their bugs the same day. Your first users should feel they made a great choice.
7. Log every repeated task. Keep a simple sheet of what you do by hand, how often and how long it takes. This becomes your automation roadmap.
8. Automate one bottleneck at a time. Once a task is eating real hours each week and you know exactly how it should work, build the product version of it.

Example

Paul Graham’s essay on this idea describes how the Airbnb founders went door to door in New York to recruit hosts and help them improve their listings, including taking photos of the apartments themselves. The same essay describes Stripe’s founders setting up new users on their own laptops on the spot instead of sending a sign-up link. Neither move would work at large scale, and neither needed to. Each one got the product into the hands of people who would actually use it.

A hypothetical B2B example: a scheduling tool for physiotherapy clinics starts with five clinics in one city. The founder visits each one, migrates their calendar from spreadsheets by hand and sits with the receptionist for an hour. She notices every clinic asks for SMS reminders, so that becomes the first feature built after launch.

Metrics to track

• Outreach reply rate. The share of personal messages that get a response. Rising replies mean your targeting and message are getting sharper.
• Activation of hand-recruited users. The share of manually onboarded users who reach your key action in their first week. This should be clearly higher than self-serve signups.
• Retention of the first cohorts. How many of your first 50 users are still active after one and three months. Stable retention is the signal to start scaling.
• Referrals from early users. The number of new users who arrive because an early user told them. Growth here shows you have made people genuinely happy.
• Hours per user. Time spent per new user each week. Watch it fall as you automate the most repeated tasks.

When it backfires

• Never stopping. Manual work is a starting tactic. If you still hand-build every account after two years, you have a services business with software costs.
• Hiding a broken product. Doing the work by hand can mask the fact that the automated version will never be good enough. Test the real product with a few users early.
• Recruiting the wrong people. Personal outreach to friends who say yes out of kindness gives you false signals. Target strangers with the real problem.
• Promising what you cannot keep. If early users expect same-day personal service forever, they will feel let down when you scale. Be clear about what is special to the early days.

Questions people ask

How long should I keep doing things by hand?

Until the manual work is the thing slowing growth down. If you can still answer every new user personally and growth is steady, keep going. Automate the step that eats the most hours first.

Isn't this just sales?

Partly. It also covers onboarding, support and even doing the product's job by hand. The point is to learn exactly what users need while you win them, so you build the right thing later.

Does this work for marketplaces with two sides?

Yes, and it matters most there. Recruit the harder side by hand first, usually supply, in one small area or niche. A marketplace with a few great listings in one city beats a thin one everywhere.

Sources

• Do Things that Don't Scale (Paul Graham)