Technical SEO: The Checklist to Run Before You Launch

Technical SEO disasters are rarely subtle. They are a staging robots.txt shipped to production, or a redesign launched without redirect mapping. Both are entirely preventable, and both routinely cost sites the majority of their organic traffic.

Indexation — get this wrong and nothing else matters

Check these before anything else, because every other optimisation is worthless if the site cannot be indexed.

• The production robots.txt is the production version. A staging file containing Disallow: / reaching production is the single most expensive one-line mistake in web development. Verify the live file by fetching it, not by trusting the deployment.
• No stray noindex tags. Crawl the built site and list every page carrying a robots meta tag. A staging-wide noindex frequently survives into production on a subset of templates.
• The CMS “discourage search engines” setting is off. WordPress in particular ships this checkbox, and it silently overrides everything else.
• No HTTP authentication left on. Password-protected staging behaviour carried into production blocks crawlers as effectively as it blocks users.

Do this check on launch day and again 48 hours later, because deployments after launch have a habit of restoring staging configuration.

One URL per page, declared

Duplicate-URL problems dilute ranking signals across variants of the same page. Four decisions settle it:

1. Pick a canonical hostname — www or non-www — and 301 the other to it. Both resolving with a 200 is the default failure.
2. Force HTTPS. An http variant returning 200 doubles every URL on the site.
3. Decide on trailing slashes and redirect the other form consistently. /about and /about/ both returning 200 is the same duplication in a less obvious costume.
4. Emit a self-referencing canonical on every page, using the absolute URL in the form chosen above. Confirm that parameterised URLs — tracking tags, filters, pagination — canonicalise correctly rather than being ignored.

The quickest test: request every variant of your homepage and confirm exactly one returns 200 and the rest 301 to it.

Redirect mapping — the redesign killer

If URLs change during a redesign, every old URL with traffic or links needs a 301 to its closest equivalent. This is the work that gets compressed when a launch date slips, and it is the reason redesigns lose traffic.

• Export every URL from Search Console, analytics and the old sitemap, then crawl the old site to catch pages missing from all three.
• Map to the closest specific equivalent. Bulk-redirecting everything to the homepage is treated as a soft 404 and preserves nothing.
• Use 301, not 302. Temporary redirects do not consolidate signals.
• Avoid chains. Old URL to intermediate to final wastes crawl budget and loses value at each hop; point directly to the destination.
• Test the mapping against staging before launch, not after.

Performance and rendering

Two checks account for most launch-time performance regressions:

• The LCP element is server-rendered and preloaded. Identify what it actually is — usually a hero image or headline — and make sure it is not waiting on JavaScript or a lazy-loading attribute. Never lazy-load the LCP image.
• Explicit dimensions on every image and embed. Missing width and height attributes are the dominant cause of layout shift.

If the site renders client-side, verify what Googlebot sees using the URL Inspection tool’s rendered HTML — not the view-source output. Content that only exists after a client-side fetch may be indexed late, partially, or not at all.

Structured data and the crawl basics

• Validate every schema type in the Rich Results Test. Invalid JSON-LD is ignored entirely rather than partially, so a single syntax error removes all eligibility on that page.
• Do not mark up content that is not visible on the page. It is a guidelines violation and a manual-action risk.
• Ship an accurate XML sitemap containing only canonical, indexable, 200-status URLs — and reference it from robots.txt.
• Give every page a unique title and meta description. Templated pages sharing one title are among the most common and cheapest issues to fix.
• Check internal linking depth. Anything more than three clicks from the homepage is crawled less often; orphan pages may not be crawled at all.

The launch-day sequence

In order, on the live site, the hour it goes public:

1. Fetch robots.txt and read it.
2. Crawl the site and filter for noindex, 404 and 5xx responses.
3. Test twenty of the highest-traffic old URLs and confirm each 301s to the right place.
4. Confirm one homepage variant returns 200 and the others redirect to it.
5. Submit the sitemap in Search Console and request indexing on the key templates.

Then check again in 48 hours. A meaningful share of launch SEO failures are introduced by the first deployment after launch, not by the launch itself.