Your app is empty. Now what?
Five things people actually do, and what each one really costs.
The cold start problem gets described as a growth problem, which is why the advice is always about growth. But on the day it bites you it is not a growth problem at all — it is a first-impression problem. Someone lands on a product that works perfectly, sees nothing happening, and leaves in four seconds. They will not tell you why. You will read it as “no product-market fit” and start changing the wrong things.
Here is the actual menu, including the options nobody writes down.
1. Be the whole community yourself
The classic answer, and the one that genuinely works: post everything, reply to everyone, keep it up for longer than feels reasonable. Reddit did it. Plenty of forums did it.
What it costs: months of your attention at the exact moment you need that attention for the product. It also has a ceiling — one person cannot produce the impression of many people disagreeing, because you only have one set of opinions and one way of writing. Readers feel this before they can name it.
2. Recruit a founding cohort by hand
Twenty people who owe you a favour, dragged in one DM at a time. This is the highest quality option and the one most likely to produce real retention, because those people are actually interested.
What it costs: your network, which is finite and non-renewable. And it is slow in a way that does not match a launch date. It is also fragile: a founding cohort that arrives to an empty room mostly leaves again, so you often need something in the room first — which is the problem you were trying to solve.
3. Fake it and hope
Fabricated accounts presented as real users. Purchased reviews. Screenshots of numbers that never happened. It is common enough that an entire fraud-detection industry exists to catch it.
What it costs: more than people expect. Platform bans, consumer-protection exposure, and the specific damage of a user who finds out — because they always find out, and the story they tell afterwards is not about your product. If you are showing fabricated traction to an investor or an acquirer, that is a different category of problem with a different category of consequence.
This is the option this product exists to make unnecessary, which is why our own Terms prohibit using it that way.
4. Hide the emptiness
Waitlists, invite-only, “coming soon”. Turn the empty room into an exclusive one and let scarcity do the work.
What it costs: it postpones the problem rather than solving it. The day you open the doors the room is still empty, and now the people walking in had their expectations raised first. It works when the scarcity is real and the demand already exists; it works badly when you are using it to avoid the question.
5. Seed it, and say that you seeded it
Put activity in the room, label it as what it is, and remove it as real people arrive. Restaurants seat early diners by the window. Markets have stallholders who show up before the customers. Nobody thinks this is fraud, because nobody is pretending the seat-filler is a satisfied regular.
What it costs: the discipline to keep the label on. The moment you take it off, you are doing option three and you have all of its problems. That is the whole distinction, and it is not a fine line — it is the difference between a demo and a deception.
How to seed without lying
Three rules make the difference, and they are worth applying whether or not you use this product:
- Label every synthetic account, visibly, in the product. Not in a footnote, not in the Terms — on the profile, where a user reading a comment can see it.
- Keep it out of your numbers. Synthetic activity is set dressing, not traction. If it reaches a metric you report to anyone, you have crossed the line.
- Make it removable. Seeding you cannot undo is a permanent claim about your history. Log every write so the whole thing can come back out.
Dead Internet enforces the first and third in code: the AI disclosure is written by the same function that writes the member and cannot be switched off, and every row is recorded so one click removes all of it. The second one is on you — and it is in the Terms because it matters more than the other two.
The part that actually helps
Seeding buys you the thing you cannot otherwise buy: a product that looks worth joining during the weeks when it is not yet worth joining. It does not create demand, retention or product-market fit, and anyone selling it as such is selling option three with better branding. What it does is stop you losing the visitors you already earned to a blank screen — and give you an honest few weeks to find out whether the thing works.
The manifesto is the longer version of this argument. The seed data pages cover what it looks like in your specific database.