Key takeaways
- Treat launch as a go/no-go decision where every signal has a repeatable pass/fail test, not a to-do list judged by feel.
- Run five technical checks before flipping DNS: HTTPS, mobile Lighthouse score of 70+, unique meta tags, working 404 handling, and verified analytics events.
- Show your homepage to three strangers for five seconds each; three correct answers on what it does, who it's for, and what to click is the pass bar.
- Verify AI crawlability pre-launch by pasting homepage copy into a chatbot and confirming text lives in the HTML, since day-one crawler snapshots stick.
- At zero users, only fix blockers to first visit or signup; verify one distribution channel and draft your launch post before DNS flips.
Launching a website is a go/no-go decision, not a to-do list: you verify a short set of binary signals in the 48 hours before DNS flip, and each one either passes or blocks launch. This checklist gives you the exact test for each signal, so two founders running it reach the same verdict on the same site.
What does 'ready to launch' actually mean for a site with zero users?
Launch-ready means every go/no-go signal is verified, not felt. "The site looks good" is not a signal. "Lighthouse performance score of 90+, tested from a cold browser" is, because two people running the same check get the same answer. That is the bar for the rest of this guide: each pre-launch task gets a pass/fail method you can run in the 48 hours before you flip the DNS.
Why generic checklists leave the go/no-go decision to luck
Most launch lists say "make it fast" or "polish your copy" and stop there. They never define the threshold, so a solo builder shipping on Lovable or Cursor defaults to vibes and launches when anxiety runs out. Google's own tooling shows the alternative: PageSpeed Insights / Lighthouse documentation defines exact scoring thresholds for performance, so two founders testing the same site reach the same verdict.
The same logic applies to everything else: copy, conversion paths, AI search visibility. If a check cannot be repeated by a stranger and produce the same result, it is not a check. We applied this standard to our own relaunch process in data-driven marketing: why your relaunch is a replay.
Which technical signals must pass before you flip the DNS?
Run these five checks in order. Any fail means no-go, because each one silently kills traffic you worked to earn.
| Signal | Pass condition | How to verify |
|---|---|---|
| HTTPS | No mixed-content warnings | Open the live URL in an incognito window; check the padlock icon shows no warnings |
| Mobile performance | Lighthouse Performance score of 70+ on mobile | Run PageSpeed Insights on the deployed URL, not localhost |
| Meta tags | Unique title and description on every page | View source and search for <title> and name="description" |
| 404 handling | Unknown URLs return your 404 page, not a blank screen | Visit yourdomain.com/thispagedoesnotexist |
| Analytics | Pageviews and one custom event fire | Open your analytics dashboard in a second tab and watch yourself arrive |
The 10-minute mobile performance test
Open your site on your actual phone over cellular, not wifi. If the first screen takes longer than three seconds to show anything useful, fix images and fonts before launch. Lovable and Bolt exports often ship unoptimized hero images, so compress anything over 200 KB.
The analytics and event smoke test
Visit your site, click your main call-to-action, and confirm both the pageview and the click event appear in your dashboard within a minute. If your signup button fires no event, you will launch blind. A pre-launch checklist like our 14-day guide pairs these checks with the distribution work that follows.
Does your homepage pass the 5-second stranger test?
Show your homepage to three people who have never seen the product for five seconds, then close the tab. Ask each one: what does this do, who is it for, and what would you click next? Three correct answers on all three questions is a pass. Two out of three correct is a fail on message clarity, not a near-pass.
How to run the test with three people who've never seen the product
Recruit in a Discord you're active in, a build-in-public thread, or a friend who fits your target user but hasn't seen the site. Don't brief them beyond "look at this for five seconds." If they can't name your audience or your next action, the fix is usually a clearer headline, not a redesign. Keep your hero line to one sentence and one visible call to action.
Before launch, also confirm search engines can see that copy: Google's documentation on how search works explains why blocked pages or thin content can't rank. For the follow-on motion after launch, the 90-day AI launch system covers turning clarity into distribution.
Is your site readable by AI search before anyone visits?
ChatGPT, Perplexity, and Claude cite sites they can crawl and parse, so AI search visibility is a launch check, not a post-launch project. Run two tests before you flip the DNS.
The copy-paste test: ask an AI chatbot about your category
Paste your homepage copy into a chatbot with web access disabled and ask: "What does this company do, and who is it for?" If the answer is vague, your entity description is vague. Your homepage should name your category, your audience, and your differentiation in plain sentences a model can lift verbatim. "The AI marketing co-founder for solo founders and vibe coders" passes. A clever tagline with no nouns does not. Then repeat the test on your about page and one product page.
Structured data checks that take 15 minutes
No code required. View your page source and confirm a title tag, a meta description, and one H1 that states what you do. Use Google's Rich Results Test (search.google.com/test/rich-results) to confirm the page renders and parses cleanly. If you shipped on Lovable or Bolt, check that your text lives in the HTML, not behind a client-side render that crawlers may miss. Fix anything that fails here before launch, because AI crawlers index your empty site on day one and that first snapshot sticks. We cover why this belongs in the launch window rather than a later marketing sprint at https://heycatch.ai/blog/ai-search-visibility-is-a-launch-problem-not-a-marketing-one.
Which conversion paths must be tested in the 48 hours before launch?
The full-funnel dry run with a test card
Open an incognito window and walk the exact path a stranger would: land, click signup, verify email, reach the app. If you take payments, run a real card in test mode (Stripe gives you test card numbers for this) and confirm the receipt email arrives. Then break things on purpose: submit an empty form, use a dead email address, abandon the payment halfway. A pass means every failure state shows a human-readable message and a way back, not a blank screen. For a deeper sequence of what to check post-launch, see our HN launch breakdown.
What to do if a signal fails at hour 40
Fix only blockers: anything that stops signup, payment, or first use. A misconfigured SSL certificate counts, because browsers will scare users off before they see your page; Let's Encrypt documents the fix at their HTTPS setup docs. A typo in your footer does not. Ship the fix, re-run the failing test once, and log the result as pass or no-go. Everything else waits until after you have real users.
Do you have a distribution signal lined up before going live?
A site with zero visitors on day one is a launch failure you can predict. Before flipping DNS, verify one place you can actually post to: pick two or three communities where your target users hang out, read their self-promo rules, and confirm your post survives AutoMod. If you cannot name the subreddit or forum and quote its rule about promotional content, that channel is not lined up yet.
The 90/10 rule check for each community
Most founder-friendly communities enforce a 90/10 rule: nine genuine contributions for every one self-promotional post. Before launch, spend time answering questions, commenting on other builders' threads, and building a comment history in each account. A brand-new account whose first post is "I built this" gets removed in minutes. Track this as a pass/fail: do you have real activity in this community from an account older than a week?
Drafting the launch post before launch day
Write your build-in-public post now, not on launch morning. Lead with the problem you hit, what you shipped, and one specific lesson, then link the site. A draft you can reread tomorrow catches the "we're excited to announce" tone that gets posts ignored. This is also where a tool like HeyCatch fits: it plans which channels to post in and drafts the posts, so launch day is execution rather than scrambling. Either way, the check is the same: one verified channel, one drafted post, done before DNS flips.
What should you deliberately NOT fix before launch?
At zero users, most polish is procrastination. A missing blog, a rough logo, only three testimonials, no pricing page A/B test: all ship-anyway. None of them stops a stranger from understanding what you built or signing up.
True no-gos are narrower: a signup flow that fails, a homepage that fails the 5-second stranger test, or a domain that AI crawlers can't fetch. If those pass, launch.
No-go vs. ship-anywhere: a decision table
Run every open item through one filter: does this block a stranger's first visit or first signup? If no, it ships. Rewrite the copy after ten real sessions tell you what's confusing, not before. The same logic applies to distribution: don't build a content engine pre-launch, just show up where users already are and follow the 90/10 rule Reddit's own self-promotion guidance describes (Reddit self-promotion guidelines). One useful launch-week post beats a month of unscheduled "content strategy."
How do you run the 48-hour go/no-go review?
Put every signal from the previous sections on one scorecard. Each row gets a pass, a fail, or a skip. The rule is simple: any fail blocks launch. Skips are allowed only for signals you have deliberately deferred (the "do not fix" list), and you must name them in advance.
The signal scorecard
List each check with its verification method and owner (you, since you are solo). DNS and SSL pass, the 5-second stranger test passes, AI crawlers can fetch your pages, your signup flow works end to end, and one distribution post is drafted. Count fails. One fail means fix it before flipping DNS. Zero fails means go.
What to do in the first 24 hours after a go
Ship the distribution post, watch analytics for real visitor behavior, and log every bug instead of fixing it mid-launch. Fix only what blocks a stranger from signing up.
FAQ
How long does a pre-launch checklist take for a simple landing page?
The verification fits comfortably inside 48 hours, and most of it runs in a few focused hours. The five technical checks take under an hour, the 5-second stranger test needs a day of scheduling, and the AI crawlability checks take about 15 minutes per page. Distribution prep, including community activity and a drafted post, is the longest item.
What is the most commonly missed step when launching a website?
Analytics verification. Founders install the snippet but never confirm a pageview and a custom event actually fire, so they launch blind. The second most missed step is AI crawlability: if your text renders client-side, crawlers index an empty snapshot on day one and that first impression sticks. Both checks take minutes before the DNS flip.
Should I launch my website before I have the product finished?
Yes, if the site passes the go/no-go signals. Launch-readiness is about the site, not feature completeness. A signup flow that works, a homepage that passes the 5-second stranger test, and pages AI crawlers can fetch are the bar. A missing blog, rough logo, or thin testimonials are ship-anyway items, not blockers at zero users.
Can I fix SEO issues after launching a website?
Some, but not the first snapshot. AI crawlers and search engines index your empty or broken site on day one, and that initial version lingers. Title tags, meta descriptions, and one clear H1 take minutes to verify pre-launch, so there is no reason to defer them. Copy refinement based on real visitor behavior is the post-launch work.
Do I need analytics set up before launch day?
Yes, and verified, not just installed. The pass condition is concrete: open your dashboard in a second tab, visit the site, click your main call-to-action, and watch both the pageview and the click event appear within a minute. If your signup button fires no event, you cannot see what real visitors do in your first 24 hours.