The launch checklist
Everything to do before you tell anyone, in the order that stops you doing it twice. Includes the UK bits nobody else mentions.
Launching badly is recoverable. Launching with a broken signup, no analytics and a homepage that says "Create Next App" is a waste of the one day people are actually looking.
Work through this the week before, not the morning of.
The week before
Analytics installed and firing. Check it records your own visit. Half of all analytics setups are silently broken and the owner finds out a fortnight later.
Every page has a real title and meta description. Search for Create Next App in your repo. It's still in there.
Open Graph image. One image, 1200x630, with the product name on it. Every link anyone shares uses this. Test it at opengraph.xyz rather than trusting it.
It works on a real phone. Your actual phone, on mobile data, not the browser's responsive mode. Different thing entirely.
Signup works from a browser you've never used. Private window, different browser, ideally a different device. You have a logged-in session and cached everything, so you're the worst possible tester of your own signup.
Password reset and magic links arrive. Check spam. If you're on Supabase's built-in mailer, they will be in spam. Point it at Resend.
Someone else uses it in front of you. 10 minutes, no help from you, no explaining. Watch where they stop. This is worth more than everything else on this list.
The legal bit (UK)
Boring, quick, and the part that gets skipped.
Privacy policy if you collect anything at all, including analytics. UK GDPR applies from your first visitor, not from your first thousand.
Cookie consent only if you use cookies that aren't strictly necessary. Plausible and Fathom don't set cookies, so you can skip the banner entirely, which is a decent reason to use them.
Terms if you take money or host user content.
A trading name and contact address on the site if you're selling. Sole trader is fine, you don't need a limited company. You do need to be findable.
Taking payments? Read up on VAT before the first sale, not after. If you're selling digital goods to consumers in the EU there's no threshold, it applies immediately, and a merchant of record handles it for you. There's a longer piece on that.
The day before
Set a spending cap on every API key. Especially anything AI. A post that does well can cost you real money in an afternoon, and the alert arrives after the bill.
Check the free tier limits you're closest to. Database connections, email sends, function invocations. Know which one breaks first.
Deploy and test the production URL. Not preview. Production, on the real domain, with the real environment variables.
Write the posts. Don't write your Product Hunt description at 7am while the launch is live. Have the tagline, the description, the screenshots and the first comment ready in a text file.
Decide what success looks like. Pick a number. 50 signups, 20 conversations, 5 people who come back on day 2. Without one you'll read the same result as a triumph or a disaster depending on your mood.
Launch day
Post where your users are before you post where founders are. The founder communities give you a traffic spike and very few real users. The niche community gives you 20 users who care.
Then stay at the keyboard. Reply to everything for the first 6 hours. The difference between a launch that converts and one that doesn't is usually whether the builder was around to answer.
The day after
Email everyone who signed up. Personally, one at a time, referencing something specific. At 50 users that's an hour and it will do more for retention than any feature.
Then look at the one number you picked. If you hit it, keep going. If you missed it badly, go and find out why before you build anything else.
What doesn't matter
Perfect copy. A logo. Dark mode. An onboarding tour. A roadmap page. The exact launch date.
All of it can wait. Nobody has ever failed because their launch tweet was posted on a Tuesday.