Zway.ai

Playbook

The Stealth Launch Sequence

Work backwards from launch day. Spend the twelve weeks before it assembling an audience and a list of people who owe you a reply, run the day itself as a staggered sequence rather than a broadcast, then spend the week after answering every single response. The week after is the part most launches skip.

Last reviewed 27 August 2026

Outcome
A launch day with an audience already assembled, and a week after it that turns attention into conversations
Timeframe
12 weeks plus launch week

The steps

Run it in this order.

  1. 01

    Fix the date and write the working-backwards list

    Put a real launch date in the calendar twelve weeks out and put the whole plan on one page beneath it. A launch without a fixed date slides, and a launch that slides never assembles an audience, because nobody can be invited to a day that does not exist yet.

  2. 02

    Write the one sentence you want strangers to repeat

    Before any asset exists, write the single sentence someone would use to describe you to a colleague. Say it out loud to five people who do not know the product, then ask them to repeat it back the next day. If they cannot, rewrite it. Everything downstream is built on that sentence.

  3. 03

    Build a named list of 100 people, not a follower count

    Weeks twelve to eight. Open a spreadsheet and add real names with a real reason each person would care, sourced from your inbox, former colleagues, people who have publicly complained about the problem, and members of the communities you plan to launch in. One hundred people who will open a message beats ten thousand anonymous followers.

  4. 04

    Start earning standing in two communities now

    Weeks ten to two. Pick the specific subreddits, Slack groups, or Discords where your buyers already are and participate with no product mention at all. Standing accrues by the week and cannot be bought on launch day, so this runs quietly in the background the whole time. Two communities done properly, never fifteen.

  5. 05

    Publish four pieces early so search has time to find them

    Weeks eight to three. Answer the questions your buyers type before they know your category exists. Search indexing and AI assistant coverage both move on timescales you do not control, so anything published during launch week arrives too late to help launch week. Early publication is the only version of this that works.

  6. 06

    Freeze every asset two weeks out

    By week two the landing page, demo recording, screenshots, and the exact text of every launch post should be finished and reviewed. Nothing gets edited in the final fourteen days. The reason is practical. Launch week attention has to go to answering people, and you cannot do that while rewriting a headline.

  7. 07

    Message your hundred people individually the day before

    Not a bulk send, not a newsletter blast. One message each, written to that person, referencing why you thought of them specifically. This is a full day of work and it is the highest-yield day in the entire twelve weeks, because a launch with a hundred people already primed behaves completely differently from one without.

  8. 08

    Run launch day as a staggered sequence

    Publish in order rather than all at once. Start with the hardest channel while you are fresh and alert, then release the rest across the day so each post gets your full attention during its first hour. A post nobody is watching in hour one is a post that is already over.

  9. 09

    Answer everything for seven days straight

    Launch week is a conversation job, not a publishing job. Reply to every comment, question, and signup within the hour where you can, and offer a call to anyone whose question sounds like a buying question. Most launches stop the moment the posts are live, which is exactly why most launches produce a spike and nothing after it.

  10. 10

    Audit in week two and commit to one channel

    Write down which channel produced actual conversations rather than clicks, and which of your hundred people replied. Then commit to the single strongest channel for the next ninety days. The launch is really a data-gathering exercise for that decision, and the decision is worth more than the launch was.

A launch is not an announcement

A launch is the one day you get to spend an audience you assembled earlier. If no audience exists on the day, the announcement lands in an empty room, and no amount of copywriting fixes that.

This reframe changes what the twelve weeks before are for. They are not for finishing the product, which you are doing anyway. They are for building the list, the standing, and the search presence that launch day spends. Teams that treat the pre-launch period as pure engineering time arrive at the date with a finished product and nobody to show it to.

The twelve weeks, in order

Weeks outThe workWhat exists at the end
12 to 9Date fixed, sentence written, list startedA one-page plan and about 40 named people
8 to 5List to 100, communities entered, first pieces publishedStanding in two communities, three or four published pages
4 to 3Assets built, beta users interviewed, quotes collectedPage, recording, screenshots, real user language
2 to 1Freeze. Individual messages drafted but not sentNothing left to write on the day
Launch dayStaggered publishing, live responsesA list of people who replied
Week afterConversations, calls, fixesPipeline, and one channel decision

The only row people reliably execute is the launch day row. The rows above and below it are where the outcome is actually decided.

Launch day is an operations day

Nothing should be written on launch day. If you are drafting copy at nine in the morning, you have already lost the hours that mattered.

What you are doing instead is watching. Each post has roughly an hour where the people who see it first decide whether it goes anywhere, and your presence in that hour is the variable you control. Answer the first three comments properly, including the sceptical ones. Concede anything fair immediately. Threads are read by far more people than ever comment in them, and a founder visibly answering hard questions reads better than a flawless post nobody is tending.

Keep a single document open and log every person who engages, with a note. That document is the actual asset the day produces.

The week after, which nobody plans

This is the part that separates launches that mattered from launches that did not.

Attention decays fast, but the people it delivered do not disappear. In the seven days after, work the log. Reply to everyone. Send a short personal note to anyone who signed up and did not come back. Offer a call to anyone who asked a question that sounded like they were evaluating rather than browsing. Fix the one thing that three or more people got stuck on, and tell the people who reported it that you fixed it.

None of this scales, and it is not supposed to. It is the highest-conversion work available to a company at this stage, and it is available for about a week before the moment closes.

What this sequence will not do

It will not produce a hockey stick. A launch run this way typically produces a few hundred visitors, a few dozen conversations, and a small number of people who genuinely want the product. That is a good outcome and it is what to plan for.

It will also not work if compressed. The two components that need real time are community standing and search presence, and neither can be accelerated by working harder in the final fortnight. If you have four weeks rather than twelve, run the list building and the launch day sequence properly, skip the community and search work, and accept a smaller day.

Playbook questions

Is staying in stealth worth it at all?
Usually less than founders think. Stealth protects an idea from competitors, which is rarely the actual risk, and it costs you twelve weeks of audience building, which usually is. If you do stay quiet about the product, you can still build in public about the problem, and that costs you nothing while buying most of the benefit.
What if I have no audience and nobody to put on the list?
Then the list is built from strangers, which takes longer but works. Find people publicly discussing the problem you solve, in forums, in issue trackers, in job posts, and talk to them about the problem with no pitch attached. Thirty people who have replied to you once is a real launch list.
Does launch day traffic actually matter?
Only as a means to conversations. A launch that produces ten thousand visitors and no replies is worse than one that produces four hundred visitors and thirty conversations. Judge the day on how many people you ended up talking to, because that is the only output the following month can be built from.
Can I launch on several sites on the same day?
You can, but stagger them by hours rather than firing at once, and read each site rules first. Hacker News guidelines exclude blog posts, sign-up pages, newsletters, and lists from Show HN, and asking people to upvote is penalised. Getting one channel right beats getting four channels half right.
What happens if the launch flops?
Almost all of them look like flops on the day, because a launch is a spike and a spike decays. The thing that determines whether it mattered is what you do in the following week. If you keep talking to everyone who showed up, a quiet launch still produces the first customers.

Or have us run it.

This playbook is yours to use. If you would rather not run it yourself, that is the entire point of Zway.

Book a demo

30 minutes. You leave with the plan either way.