Your first 100 users: which channels to try, in what order

The first hundred users are worth more as a source of information than as a number. You need to learn who wants the product, which words make them try it, and where they get stuck. A channel that lets you talk to each person does that. A channel that sends anonymous visits does not.

Paul Graham’s essay “Do Things that Don’t Scale” makes the same case for startups: “You can’t wait for users to come to you. You have to go out and get them.” He also explains why small numbers shouldn’t put you off, with plain arithmetic: “If you have 100 users, you need to get 10 more next week to grow 10% a week.” Ten people is a list of names, and you can work through a list by hand.

So the order below runs from most contact with each user to least, and from no money to some. It is a default. Change it when your product gives you a reason to.

Begin with former colleagues, clients, friends in the trade, people who replied when you first described the idea, and anyone who has complained in public about the problem you solve. Write to them one at a time.

• Bad: “Hey, I just launched my app, would love your support!” sent to two hundred people.
• Better: “You mentioned last year that chasing late invoices took you a day a month. I built something for that. Could I set it up for you on a fifteen-minute call?”

Offer to do the setup yourself. Graham describes how Stripe’s founders did this: when someone agreed to try the product, they asked for the laptop and installed it on the spot, a habit Y Combinator came to call a “Collison installation” (source).

Cost: an hour or more per user, counting the message, the call and the follow-up. Working if: people reply, some use the product a second time without being reminded, and a few introduce you to someone else. Stop when: you have run out of people you can honestly say you know, or the conversations have stopped telling you anything new.

Next come the forums, subreddits, Slack and Discord groups, and niche newsletters your users read. This only works if you take part before you promote. Read the rules on self-promotion, answer questions for a few weeks, and mention your product where it actually answers what was asked.

Rules differ. Hacker News, for example, reserves Show HN for “something you’ve made that other people can play with”, rules out landing pages and sign-up pages, and asks you not to get friends to upvote or comment (Show HN guidelines). The guides on launching on Reddit and writing a Show HN post go into each.

Cost: a few hours a week, for weeks, most of it spent answering other people. Working if: people ask follow-up questions, sign-ups arrive from the thread, and your name starts to be recognised. Stop when: several honest, on-topic contributions in one community have produced no clicks and no conversation. Move to another community before you give up on the channel.

A launch board gives you a day or a week of visibility and a page that stays up afterwards. IndieTide is one of these: a weekly board where each product keeps a permanent page linking to the maker’s site. Treat any board as a short burst. Graham’s line on launches applies: “All you need from a launch is some initial core of users” (source).

Look at who comments on recent launches before you choose. Where the visitors are mostly other makers, a tool for makers will do better than software for dentists. Pick a handful using the checks in judging launch directories and where to launch, and prepare with the launch checklist.

Cost: a day or two of preparation that you can reuse, then an hour per listing and a day of answering comments. Working if: your analytics show referrals from the listing that turn into sign-ups. Stop when: you have listed on the few boards that have a real audience. Submitting to fifty more is usually time taken from the channels above.

Search is slow, so start the groundwork in week one and expect nothing from it for a while. Google says of changes to a site that some “might take effect in a few hours, others could take several months”, and that there is no guarantee any particular site will be added to its index (SEO Starter Guide). Nobody can promise you a ranking either; Google’s own advice is to walk away from anyone who does (source).

The early work is small:

1. Make sure the site can be indexed and set up Search Console. See getting indexed by Google and Bing.
2. Write one page for each specific problem your product solves, using the words your first users used in conversation.
3. If people compare you with a known product, write an honest comparison. See alternatives pages.
4. Consider a small free tool that does one part of the job without an account.

Cost: half a day to a day per good page, and several days for a tool. Working if: Search Console shows impressions for the queries you aimed at, then clicks. Stop when: a page has been indexed for months and has no impressions for its topic. Rewrite it for a narrower query or drop it. Don’t stop the channel because month one was empty.

Partnerships and integrations

An integration with a product your users already use, a listing in its app marketplace, or a swap with a newsletter in your niche. These come later because the other side will want to see a working product with real users, and an integration costs development time.

Cost: days to weeks of building, plus the back-and-forth. Working if: sign-ups arrive through the integration and those users stay. Stop when: a partner goes quiet after two follow-ups, or an integration has been live for a few months and nobody has connected it.

Ads

Ads are last because they are the one channel here where you pay for the visits. If visitors from free channels don’t sign up, there is little reason to expect paid visitors to. Before spending, you should know from the earlier channels which wording gets people to try the product, what share of sign-ups become paying users, and roughly what one paying user is worth to you.

Cost: a budget you can afford to lose entirely, plus time learning the ad platform. Working if: the cost of getting one paying user is below what that user pays you. Stop when: the test budget is spent and that sum doesn’t work. Fix the budget before you start.

Working seven channels at once means each gets an hour a week, and none gets enough to show a result. Graham recommends starting narrow for a similar reason, comparing it to “keeping a fire contained at first to get it really hot before adding more logs” (source). Run one direct channel and one slow one, for example personal outreach alongside search pages.

1. Write down the test before you start. The channel, the hours per week, the end date and what you will count. For example: “Twenty personal messages a week for four weeks. Count replies, calls and people who used it twice.”
2. Count something close to real use. Sign-ups and repeat use tell you more than votes, likes or visits.
3. Tag your links. Use UTM parameters, and ask new users how they found you.
4. Judge at the end date. One bad week is noise. A full run with nothing to show is an answer.

If you are starting from zero followers, launching with no audience covers the first weeks in more detail, and what to do after a quiet launch covers the case where the first attempt lands flat.