Free tool
A launch plan that works backwards
Choose a launch type, how many weeks you have, and your team size, and this builds an ordered checklist counting backwards from launch day. It includes the week after launch, which is where most of the value sits and where almost every launch plan stops paying attention.
Last reviewed 27 August 2026
Result
Tasks before launch day
23
7 more in the week after
Rough effort
80 hours
About 7.4 hours a week each across 2 people
Counting back from launch day
| When | Task | Why |
|---|---|---|
| 4 weeks out | Get the product to a state where a stranger can complete one useful thing unaided. | You said this is not true yet, and no amount of launch preparation compensates for it. |
| 4 weeks out | Write the one sentence you want repeated back to you, and cut every other claim. | Launches fail on positioning far more often than on tactics. |
| 4 weeks out | Put five strangers through the product without you in the room and note where each one stops. | The stopping points are the launch plan. Everything else is decoration. |
| 4 weeks out | List every place your buyers already gather and join those places now. | Accounts with no history get filtered or ignored on launch day. |
| 2 weeks out | Build a list of forty named people you can message personally on launch day. | With no audience, forty individual messages outperform any broadcast you can buy. |
| 2 weeks out | Fix the two biggest stopping points from the stranger tests. | Nothing else on this list matters if the product confuses people. |
| 2 weeks out | Write the launch post, then cut it by a third. | The cut version is always the better one and it is never the first draft. |
| 2 weeks out | Build a path a stranger can complete in under two minutes without signing up. | Launch-day traffic does not wait for an onboarding sequence. |
| 2 weeks out | Set up analytics with one working conversion event, and tag every outbound link with UTMs. | Untagged launch traffic tells you a number and nothing about where it came from. |
| 2 weeks out | Tell three people who will genuinely find this useful what it is, and ask for an honest opinion. | This is a request for a reaction, not a request for votes. |
| 2 weeks out | Write down what a good day looks like as a number you can check by 6pm. | Without it you will judge the day on how the comments felt. |
| Final week | Prepare answers to the five hardest questions, including price, privacy, and why not just use the obvious alternative. | These arrive in the first hour, every time. |
| Final week | Check every link, on mobile, logged out, on a network that is not your office. | The most common launch-day failure is a broken path nobody tested cold. |
| Final week | Block the calendar for launch day. Post, then answer for four hours. | A thread you abandon converts worse than no thread. |
| Final week | Tell existing users what is coming, without asking them for anything. | People who already use it are the only audience that owes you attention. |
| Final week | Put one obvious action on the page and remove the second one. | Two calls to action on launch day means neither gets taken. |
| The day before | Freeze the product. Nothing ships tomorrow except copy fixes. | A deploy on launch morning is how a good day becomes an incident. |
| The day before | Sleep. The first four hours of replies decide the day and you cannot do them tired. | This is a real item, not a joke item. |
| Launch day | Post at the start of your working day, then stay in the thread for four hours. | Early responsiveness is the only launch-day variable fully in your control. |
| Launch day | Reply to every comment in your own words, including the harsh ones. | The hostile comment answered well is the most persuasive thing on the page. |
| Launch day | Ask nobody for upvotes, including in private group chats. | On Hacker News that is a stated penalty or ban, and other surfaces have their own rules you should read. |
| Launch day | Post once in each place you have standing, with text written for that place. | Reddit's Data API terms describe identical or substantially similar content across subreddits as spamming activity. |
| Launch day | Log every objection verbatim in one document. | This is the most valuable output of the day and it is the one people forget to keep. |
After launch, where most plans stop
- The day after: Reply individually to everyone who signed up, within 24 hours. Nothing you automate later will convert as well as this does.
- The day after: Write the follow-up: what happened, what broke, what you changed. That post routinely outperforms the launch post itself.
- The week after: Interview five signups who never activated and ask what they expected. The gap between what they expected and what they found is your positioning problem, stated out loud.
- The week after: Ship the two most requested fixes and tell the people who asked for them. Being told your feedback shipped is the cheapest loyalty available.
- The week after: Turn the launch post into a permanent page and point internal links at it. Launch threads sink. Pages on your own site do not.
- The week after: Pick the one channel that actually produced signups and commit to eight weeks of it. The launch was a spike. The channel is the business.
- The week after: Record the numbers again at day 7 and day 30. Launch-day traffic is the least informative number you will collect all month.
Dates are relative because the page does not know today's date. Count back 4 weeks from your launch day and put these in a calendar, because a checklist without dates is a wish list.
How this works
What the numbers mean.
- 01Tasks are stored with a phase rather than a date, then the phases are stretched across the weeks you have. That is why a two-week runway and a ten-week runway produce the same tasks in the same order at different densities.
- 02Launch type filters the list. Product Hunt items stay generic and process-focused because their rules change, while the Hacker News items quote the published guidelines directly rather than paraphrasing them.
- 03Two of your answers add work rather than removing it. If a stranger cannot use the product unaided, or you have no audience, the checklist inserts the item that fixes it, because those are the two things that decide the day.
Assumptions and limits
- Hour estimates are our operating figures for a small team. They are there to make the total visible, not to be accurate to your situation.
- Every Hacker News item comes from the published guidelines and FAQ. Every Product Hunt item is deliberately generic, because we will not restate rules we have not verified today.
- The week after launch is included on purpose. It contains most of the durable value and it is the part that gets skipped when launch day goes well.
Questions about this tool
Does my product qualify for Show HN?
Only if there is something people can actually try. The guidelines exclude blog posts, sign-up pages, newsletters, lists, and landing pages, which rules out a large share of what founders want to post. If your launch is a waiting list, submit it as an ordinary story or wait until there is something to show.
Can I ask friends to upvote?
Not on Hacker News. The FAQ answers this directly: asking people to upvote is not allowed, and submissions, accounts, and sites that break the rule are penalised or banned. Other surfaces publish their own rules, so read them at the source rather than assuming they match.
How many weeks do I really need?
Enough to test with strangers and fix what they trip on, which is usually two to four weeks of real work. The preparation that matters is product and positioning, not assets. Teams that spend six weeks on launch collateral and none on the stopping points get a busy day and no customers.
Why does the checklist keep going after launch day?
Because launch day is a spike and the week after is where it either converts or evaporates. Individually replying to signups, interviewing the people who never activated, and committing to one channel are worth more than anything you can do on the day, and they are the first things dropped when the day feels good.
Can I run Product Hunt and Show HN together?
You can, and it usually halves both. Each one wants you present in the thread for hours, in a different register, answering different questions. Pick the surface where your buyer already reads, do that one properly, and keep the other for a later release.
This tool is free and there is nothing to sign up for. If you would rather have the work done than calculate it, that is what Zway does.