Technical SEO checklist before you launch (or relaunch)

I’ve lost count of how many “SEO emergencies” were just a staging leftover. Pretty homepage. Press email already sent. Then someone notices noindex, or Disallow: /, or a certificate that dies mid-campaign.

This is the list I want in the release PR the day before DNS flips — and again about an hour after. Every step is something you can check without buying a fake “authority” score.

1. Can crawlers even fetch the site?

Open the production robots file: https://yourdomain.com/robots.txt.

Google’s own intro is blunt about what robots.txt is not: it’s for managing crawl traffic, not a reliable way to hide pages from Search. From Google’s robots.txt guide:

Don’t use a robots.txt file as a means to hide your web pages… from Google Search results.

So for launch, the question is simpler: did we accidentally ban everyone?

• Is User-agent: * still sitting on Disallow: / from preview?
• Any path rules that made sense on staging and make no sense now?
• If you’ve got a WAF, has “bot fighting” started soft-403’ing Googlebot while Chrome still looks fine?

Blocking one AI trainer on purpose is a product call. Shipping production with a full-site disallow is usually copy-paste.

Quick pass: Robots.txt Checker.

2. Indexation matches what you meant

Crawlable ≠ indexable. People mash these together constantly.

Google’s noindex docs say they have to crawl the page to see the rule:

We have to crawl your page in order to see <meta> tags and HTTP headers.

And if robots.txt blocks the URL, they can’t see the tag — same doc, same trap.

Before launch, don’t only check the homepage. Hit:

1. Home + one money page (pricing, product, flagship article)
2. A thank-you / confirmation URL
3. Account, cart, or on-site search if those are public
4. One ugly parameterized URL marketing will share anyway

Money pages: not noindex. Thank-you / cart junk: usually noindex on purpose.

The relaunch classic: CMS keeps preview noindex after go-live. I’ve seen it survive two “final” QA passes because everyone tested logged-in preview URLs.

Check both meta and X-Robots-Tag: Noindex Checker. Longer write-up: robots vs noindex vs canonical.

3. One host, HTTPS, certificate days

Pick www or apex. Not both as equals. HTTP should land on HTTPS in one hop if you can help it.

Google prefers HTTPS when signals aren’t a mess — see their notes on canonicalization and HTTPS (invalid certs and HTTPS→HTTP redirects are called out as ways to screw that up). HTTPS as a ranking signal goes back to their 2014 announcement; the operational point in 2026 is still: don’t ship a flaky cert.

Don’t trust “padlock in my browser.” Check days remaining. Let’s Encrypt renewals die quietly when DNS challenges break after a provider swap. CDNs sometimes keep serving the old leaf after origin renewed.

HSTS comes after HTTPS is boring. Not during the cutover chaos.

Tools: SSL Days Checker, Redirect Checker, Security Headers Checker.

4. Sitemap that isn’t lying

Google’s sitemap overview is refreshingly modest:

If your site’s pages are properly linked, Google can usually discover most of your site… Even so, a sitemap can improve the crawling of larger or more complex sites…

So: discovery helper, not a ranking lever.

Also from that page — you might not need one if the site is small (~500 pages) and well linked. Plenty of brochure launches still benefit from a clean /sitemap.xml plus a robots Sitemap: line, especially when the site is new and barely linked.

Rules of thumb I actually use:

• Real XML, not an HTML soft-404 wearing a sitemap URL
• Only preferred-host, 200, indexable URLs
• No thank-you pages, staging hosts, or facet explosions
• Declare it in robots.txt

Validate: Sitemap Checker. Deeper notes: XML sitemaps that actually help.

5. Titles, descriptions, canonicals on the templates that matter

Day-one copy doesn’t need to win awards. It does need to not be broken.

• Titles that aren’t identical on every template
• Descriptions that aren’t empty sitewide
• Canonicals that don’t still point at staging or the old domain
• Absolute canonical URLs (Google recommends against relying on relative ones long-term)

Spot-check home, one money URL, one article with Meta Tag Checker and Canonical Checker. If social posts go out launch morning, glance at Open Graph so you don’t share a blank card.

6. Status codes that match reality

Soft 404s with a 200 status are launch goblins. Custom error pages should return real 404/410. Old URLs after a redesign should 301/308 to the new ones — not a decorative chain of temporary redirects.

If the redesign reused URLs for totally different intent, that’s a content migration, not a CSS deploy. Put the redirect map in the same PR as the theme.

7. Schema and GEO without the cosplay

Ship Organization / WebSite / real Article or Product markup when the page content matches. Skip FAQ blocks invented only to chase rich results.

For AI-ish discovery, a short honest llms.txt beats a slogan dump. More on that in GEO and llms.txt.

Draft check: GEO Content Checker. Whole-site pass: free audit.

Cutover day (boring on purpose)

1. Freeze URL-structure changes for the window.
2. Ship with 5xx + TLS alerts on.
3. Re-run robots, noindex on money templates, SSL days, sitemap, three meta/canonical spot checks.
4. In Search Console, request indexing for a handful of priority URLs — not the whole sitemap.
5. Watch indexing reports for a week before you declare victory.
6. Then get cute with HSTS preload, aggressive cache, bulk pings, etc.

What I’m not promising

A green checklist does not buy rankings. It stops you from donating the first month to avoidable crawl/index accidents. When those are clean, content and distribution can do their jobs.

If the team wants one shareable pass, run a TheSeoSoul audit and drop the /audit/[domain] link in the launch channel.