Most launch-day disasters are not exotic. They are ordinary things nobody checked: a contact form that sends to the wrong address, a staging “noindex” tag that went live, a broken checkout on one browser, an expired SSL certificate, a missing 404 page. A pre-launch QA checklist turns “I think it is fine” into “I checked, and here is the evidence”. This guide gives a practical checklist for websites and web apps, organised by area and by timeline, with commands and tools you can use right away.
It collects points from other guides in this series: the Responsive Web Design Testing Checklist, On-Page SEO Checklist, Mobile-First Indexing Checklist, Core Web Vitals on Mobile and Touch Targets and Mobile Forms. Use this page as the master list and follow the links for detail.
How to use this checklist
- Copy it into your issue tracker or a spreadsheet, and assign each item to a person.
- Record the evidence: a screenshot, a test result or a link. “Done” without evidence is a guess.
- Prioritise: blockers (must fix before launch), important (fix within days) and nice-to-have.
- Test on a staging environment that matches production as closely as possible, then repeat the critical checks on production right after launch.
1. Functionality
- Walk through every key user journey from start to finish: sign-up, login, search, checkout or enquiry, logout.
- Click every navigation link, footer link and call-to-action. Check for broken links (a crawler or link checker helps).
- Test forms with valid data, empty data, invalid data, very long text and special characters.
- Confirm that form submissions arrive where they should, with the right content, and that auto-reply emails work.
- Check error states: wrong password, expired session, network failure, duplicate submission.
- Test payments in the provider’s test mode, including success, failure and cancellation, then do one small live payment after launch if appropriate.
- Verify roles and permissions: ordinary users must not see admin pages or other users’ data.
- Check that third-party integrations (maps, chat, analytics, payment, email) work with production keys.
2. Browsers and devices
- Test the latest versions of the browsers your visitors use. Check your analytics, or if you have none, cover Chrome, Safari, Firefox and Edge.
- Test at least one low-end Android phone, one recent iPhone and one tablet if you can. Emulators miss real-world font rendering, keyboards and performance.
- Rotate between portrait and landscape.
- Check dark mode if you support it, and increased system font size.
- Use Dev Browser to load the site on your own phone at phone, tablet and desktop widths and catch layout problems quickly. Install it from Google Play.
3. Responsive layout
- Check widths around 360, 390, 768, 1,024 and 1,440 px, plus anything unusual in your analytics.
- No horizontal scrolling. If it occurs, find the wide element, as shown in How to Read Page Source and Inspect CSS.
- Text is readable without zooming, and line lengths are comfortable.
- Tap targets are large enough and spaced out (24 px minimum, 44 to 48 px for key controls).
- The layout survives the on-screen keyboard and mobile browser toolbars. See CSS Viewport Units and the 100vh Problem.
- Images are sharp but not oversized. See Responsive Image Optimisation.
4. Performance
Test on a throttled mobile connection, not only on fast office Wi-Fi. Google’s published “good” thresholds for Core Web Vitals, measured at the 75th percentile of real visits, are LCP at or below 2.5 seconds, INP at or below 200 milliseconds and CLS at or below 0.1. Lab tools such as Lighthouse give early warning, while real-user data gives the truth after launch.
- Run Lighthouse or PageSpeed Insights on your key templates (home, listing, detail, checkout), on mobile settings.
- Compress and resize images, and load the main image first.
- Enable compression (gzip or Brotli) and caching for static files.
- Minimise render-blocking scripts and large third-party tags.
- If you expect a traffic spike at launch, do a basic load test and check database and server capacity.
5. SEO and indexing
- Remove staging blocks. No site-wide
noindexmeta tag, noDisallow: /in robots.txt, no password protection or IP restriction left over from staging. This is the most common launch mistake. - Unique titles and meta descriptions, one h1 per page, canonical tags, correct hreflang if multilingual.
- 301 redirects from old URLs to new ones if you are replacing an existing site, tested with a list of the old top URLs.
- XML sitemap generated, referenced in robots.txt and ready to submit in Search Console.
- A custom 404 page that returns a real 404 status, not 200.
- Structured data validated, Open Graph tags present, favicon in place.
# Check status, redirects and headers for key URLs
curl -sI https://example.com/ | head -n 20
curl -sI http://example.com/ | head -n 5 # should redirect to https
curl -s https://example.com/robots.txt
curl -s https://example.com/ | grep -i -E "noindex|canonical"
6. Accessibility
- Everything works with the keyboard alone, with a visible focus outline and a logical tab order.
- Images have meaningful alt text, and decorative images have empty alt.
- Form fields have visible labels, and errors are announced and linked to the fields.
- Colour contrast meets WCAG guidelines (4.5:1 for normal text is the common target).
- Headings, landmarks and link text make sense when read out of context.
- Run an automated checker, then try a screen reader on at least the main journey. Automated tools find only part of the problems.
7. Security basics
- HTTPS everywhere, with a valid certificate, automatic renewal and redirects from HTTP. No mixed-content warnings.
- Remove debug mode, test accounts, default passwords, sample data and verbose error pages from production.
- Keep secrets (API keys, database passwords) out of front-end code and out of the repository. Use environment variables.
- Use strong, unique passwords and two-factor authentication on hosting, DNS, email and admin accounts.
- Validate and sanitise input on the server, use parameterised queries, and protect forms against abuse with rate limits or a challenge.
- Consider security headers such as
Strict-Transport-Security,X-Content-Type-Optionsand a Content Security Policy, tested carefully so they do not break the site. - Update the CMS, themes, plugins and dependencies to supported versions, and remove unused ones.
- Restrict admin areas and file permissions.
8. Content, legal and trust
- No placeholder text, lorem ipsum, test products or internal notes. Proofread headlines and prices.
- Contact details, address, business name and hours are correct and match your other listings.
- Privacy policy, terms and refund or shipping policy are present and accurate, and a cookie notice appears if you use tracking that requires consent. Ask a professional about your obligations, including India’s data protection law where it applies.
- You have the rights to all images, fonts and text you use.
9. Email, DNS and domains
- DNS records are correct and have been given time to propagate. Lower the TTL before a migration and raise it afterwards.
- Both the bare domain and the
wwwversion work, and one redirects to the other. - Email sent from your domain passes SPF, DKIM and DMARC checks so that it does not land in spam. Send test messages to several providers.
- Auto-renewal and registrar lock are on for the domain, and the contact email is monitored.
10. Analytics, monitoring and backups
- Analytics are installed and recording key events (form submit, purchase, sign-up), with internal traffic filtered out if possible.
- Search Console and any webmaster tools are verified.
- Uptime monitoring and error tracking are active, with alerts going to a person who will see them.
- Automatic backups are running, and you have tested a restore at least once.
- Logging is enabled, with sensitive data excluded from logs.
11. Launch plan and rollback
- Write the launch steps in order, with an owner for each step and the expected time.
- Choose a low-traffic window and avoid launching just before a weekend or holiday when nobody is available.
- Define the rollback plan: how to return to the old version, who decides, and how long that takes.
- Freeze content and code changes shortly before launch, except for blockers.
- Tell support staff and stakeholders what is changing and where to report problems.
Timeline: what to do when
| When | Focus |
|---|---|
| One to two weeks before | Full functional, device, accessibility and performance testing on staging. Fix blockers. |
| Two to three days before | Content freeze, redirect map tested, DNS TTL lowered, backups verified, security review. |
| Day before | Final regression test of critical journeys, confirm the rollback plan and who is on call. |
| Launch day | Deploy, remove staging blocks, test the live site end to end, check robots.txt and noindex, submit the sitemap. |
| First 48 hours | Watch errors, uptime and form deliveries, check Search Console for crawl errors, review real-user speed data and fix quick wins. |
| First two weeks | Review analytics and feedback, fix issues, add anything deferred. |
Common mistakes
- Leaving noindex or a robots.txt block from staging in place.
- Testing only on a developer’s fast laptop and a single browser.
- No redirects from old URLs, losing traffic and links.
- Forgetting email deliverability, so enquiries or order confirmations vanish into spam.
- No backup or tested rollback.
- Launching late on a Friday with nobody watching.
- Treating launch as the finish line, with no monitoring afterwards.
A compact sign-off list
- Key journeys pass on the latest browsers and real phones.
- No layout overflow at common widths.
- Performance is acceptable on a throttled mobile connection.
- Staging blocks are removed, redirects work, the sitemap is ready.
- HTTPS, secrets, updates and admin access are secured.
- Forms, email and payments are verified end to end.
- Analytics, monitoring and backups are live.
- Rollback plan and owners are agreed.
Frequently asked questions
How long should pre-launch testing take?
It depends on the size and risk of the project. For a small site, a day or two of focused testing may be enough. For a large application with payments, plan weeks.
Can automated tests replace manual QA?
They complement it. Automation is good for regressions and repeated checks, while manual testing finds usability, visual and unexpected problems.
What is the single most common launch mistake?
Leaving a staging noindex tag or robots.txt block on the production site, which keeps the site out of search results until someone notices.
Do I need to test again after launch?
Yes. Repeat the critical checks on production, because live settings, DNS, caching and third-party keys often differ from staging.
Conclusion
A launch goes well when the boring things were checked: journeys, devices, speed, indexing, security, email and rollback. Work through the list, keep evidence, launch in a quiet window and watch the site closely for the first two days. Check your layouts on a real phone with Dev Browser before you press the button, and see the rest of our tools on the products page.
Disclaimer: This checklist is general guidance and cannot cover every project. Security, legal and accessibility requirements vary, so involve qualified professionals for regulated or high-risk systems.