The first customers usually do not arrive through a clean funnel. They arrive through some unfair little door: an existing user list, a founder email, a search page nobody else bothered to write, a launch day decision made while the site is still warm.
That can make founders uncomfortable. We want growth to look like a system from the start. But early growth is often closer to hand-to-hand work. The system comes later, after you have learned which parts deserve to be repeated.
Warm starts beat fair starts
Buffer Analyze is a good example because it was not pretending to be universal at the beginning. Tom Redman started with existing Buffer Publish customers, ranked the people most likely to be a fit, and emailed them personally. The first personal batch converted around 40% into trials. Later batch emails to groups of 25 were less effective, around 25%.
There is a lesson hiding in that drop. Personal attention was not a cute detail. It was part of the channel. The prospect could feel that the product was meant for them, or at least that a real person had thought about whether it might be.
Make early users feel seen
VEED had a different problem. It had usage, but not revenue. After a YC rejection, the team shipped a simple paid upgrade around watermark removal and got its first 20 paid users. In six weeks it reached more than 200 paid users and £1K MRR.
What I like is the human stuff around the paywall. They took photos holding the names of their first 100 paid users and replied to some emails with personal videos. That is not a retention strategy you can scale to a million users. Good. It was not supposed to be. At that stage, the job is to make the first believers feel the team is awake.
Use search as the slow machine
Kapwing shows the other side of the coin. While early users need care, the product also needs a way to compound. Kapwing made small tutorials for jobs people were already searching for. The production standard was not television. It was useful screencasts, shipped quickly, with the product link near the top.
Later, Kapwing found that old pages with strong conversion but weak rankings were worth refreshing first. That sounds obvious after the fact, which is usually the sign of a good tactic. Do not polish the page that gets traffic but no users. Start with the page that already turns visitors into users, then give it a better chance to be found.
Do not waste the launch
The gpt-engineer Product Hunt story is a useful warning. A waitlist can protect a product, but it can also smother a launch. Anton Osika wrote that the team turned the day around by removing the waitlist once the infrastructure could handle it, posting on Hacker News, and making the launch post shorter. The result was 500K+ impressions, 16K signups, and 850 paying users.
The move was not simply "remove friction." It was more specific than that. Let people use the thing while they are still paying attention. Then their comments, reviews, and second posts are based on experience instead of goodwill.
Where this applies
For SaaS, start with the warmest existing segment and invite them manually. For creator tools, make the first paid users feel visible. For AI products, do not hide the product behind a waitlist once the demo is stable enough. For consumer apps, use search and tutorial demand to catch people at the moment they already want to do something.
The trap is trying to make the first version look fair. Fair channels are crowded. Early traction usually comes from the edges where you have some small advantage: trust, speed, taste, timing, or the patience to write the page nobody else wanted to write.