Technical SEO before launch: what to check on the live domain
Check server responses, HTML, canonical URLs, redirects, sitemaps and access settings before and after a website goes live.
Article language: English

In this article
Everything looks fine in the browser. A search engine, however, may receive an empty HTML document, conflicting URLs or an instruction not to index the page. Review technical SEO on the actual domain: check the server response, the content, the canonical URL and crawler access. Record each check and result so it can be repeated.
SEO means preparing a website so search engines can find and understand it. This article focuses on its technical side. The review should produce a list of checked conditions and specific corrections. A confirmed publication is not the same as confirmed indexing or a particular search ranking.
Which URLs should you start with?
Collect the home page, service pages, contact page, articles and publicly available documents. Record the expected title and purpose of each URL. Add the old addresses that should lead to new locations. This list gives the review a starting point instead of relying on random clicks through the menu.
Choose one primary domain variant and a consistent way to write paths. Versions with and without www, a trailing slash or a file extension may need to be brought into order. Use the chosen variant in internal links and the sitemap. Keep a list of exceptions if an address is intended to serve a different purpose.
During a rebuild, also record which materials will disappear and why. Not every removed page should lead to the home page. The right successor depends on the content and purpose of the old address.
Does the server return the right HTTP status?
HTTP is a way for browsers and servers to communicate. A response includes a status code, such as 200 for a successfully returned page or 404 for a resource that cannot be found. Check it with a network tool, including for an address that does not exist. A displayed "404" message can still accompany a 200 response.
Google's documentation on HTTP status codes explains how they affect content processing. A 200 response allows the material to proceed to further evaluation, but does not guarantee indexing. Record both the status and the actual response content in the review record.
Follow redirects from the old address through to the destination. Avoid unnecessary chains of several hops. Check whether the final address leads to the material the visitor expected.
Is the content present in the HTML response?
Fetch the response for an article or service page and look for its title, description and main content. Also check the view without running JavaScript. If you receive only an empty application container, determine how the content will be made available to search engines and to people whose scripts have not yet run. For a page with a 3D scene, also review the path without WebGL (in Polish) so the absence of the scene does not remove the content or main action.
An application can prepare HTML at build time or render it on the server. What matters most is the result at the actual address. Testing a file on your computer does not confirm that the hosting service makes it available at the correct path. Open the article directly, refresh the tab and compare the response with what you expect.
Check ordinary links between pages too. Navigation that relies only on script-driven actions can make the path to a piece of content harder to reproduce. Every important article should have an accessible link from the archive or related content. For extensive publications, separately check section links and the return to a reading position (in Polish), so navigation within the article also has a clear destination.
Do the title, description and canonical URL match the address?
Each important page needs a title that describes its own purpose. The description should reflect its content. Move between a service page and an article, and check that the metadata also changes during navigation within the application.
A canonical URL indicates the preferred address for content. Google describes ways to indicate a canonical URL and the need for consistent signals. In a practical review, compare the canonical URL, internal links, redirects and sitemap. They should point to the consistently chosen variant.
Check the sharing preview as well: its title, description, image and image URL. An image that works locally may not be publicly accessible. On an article page, structured data should describe the actual title, author, date and content.
What should you check in the sitemap and access settings?
The sitemap should contain existing public URLs that you want to make available for indexing. Remove drafts, accidental variants and outdated destinations. The sitemap's last modification dates should reflect significant changes to each page, for example changes to its content, structured data or links. Rebuilding the application alone is not a reason to change those dates.

Review robots.txt and indexing directives. Settings intended for a draft environment may remain after publication. Check the home page and a sample of subpages instead of assuming the result is uniform.
How do you review a rebuilt article in practice?
Suppose an article changes its address. First open the old link and record the response and redirect destination. Then fetch the new address without first visiting the home page. Check the content, title, canonical URL and archive link. Finally, compare its entry in the sitemap.
If the new address works only after navigating through the menu, report that difference as a separate issue. If the old address leads to an unrelated page, explain the choice of destination. This short scenario combines several checks into one real visitor journey and gives the implementer a way to reproduce the problem.
How do you distinguish an indexing block from draft protection?
Check page access, robots.txt, the noindex directive and response headers separately. They are different mechanisms. Google explains how noindex works: a crawler must be able to fetch the resource to read the directive. Blocking a fetch with robots.txt is not equivalent to a crawler successfully reading noindex. Do not use those names interchangeably in the review record.
Protect a confidential draft with access controls appropriate to the project. Hiding a link or disabling indexing alone should not be treated as access control. After publication, check that the intended version has the right settings and that test domains still meet the agreed protection requirements. Record the expected behaviour of both environments.
How do you review changes to old addresses without redirecting people to the wrong place?
Prepare a map of each old URL, its current equivalent and the reason for the choice. If the material keeps its content and purpose, first consider keeping the address. Changing an article title does not necessarily mean changing its path. When a page disappears, decide on the appropriate response individually instead of sending every visitor to the general offer.
Google's instructions for URL changes cover mapping, redirects and subsequent monitoring. Set out your own scope using the site migration plan. During review, open old links with the parameters people use and from the materials in which the company shares them. For more context, see the article on planning a website migration without losing URLs (in Polish).
Working example: a guide keeps its address but changes its image and description. The review should confirm the new content at the existing path, a consistent canonical URL and a link from the archive. Automatically sending a reader from the guide to a service offer would not fulfil the same purpose. The offer should be available through a clearly described link in the content.
What should you check in links after expanding a blog?
Choose several old and new articles, then follow links between them in both directions. Record links that lead to nonexistent material, a draft or content different from what the description promises. Both the destination's status and the relevance of the connection matter. A link labelled "list of review materials" should open a list of those materials.
Use the information architecture of a B2B website to plan where new content belongs in navigation and the service offer. A large number of valid addresses does not replace a route by which people can find them. For a series of publications, compare titles and descriptions too, so an accidentally copied template does not present several articles as one. The related topic is explained in the article on information architecture for a B2B website (in Polish).
How do you separate site performance from a single measurement?
Plan separate measurements for loading, responsiveness and layout stability. The official Web Vitals overview organises these aspects. Record the device, conditions and tool used. A lab test and data from real visits cover different things; do not present one run as the experience of every visitor.
The article on site speed and Core Web Vitals (in Polish) develops practical decisions about images, scripts and review. Measure important page types, especially if an article and the home page use different elements. A good result does not replace checks of the form, links or clarity of the offer.
Who returns to the checks after publication?
Assign an owner for the review and make a list of addresses to reopen after deployment. Record publication status and observations about indexing in separate fields. When a review reveals a change that needs a decision, keep the reasoning in the project decision register. That gives a configuration update a specific reason and a person to confirm the result. This is discussed further in the article on keeping a project decision register (in Polish).
When checking addresses, review external links and new tabs (in Polish) too. A label should explain the destination and the behaviour should be predictable. A technically valid link can still lose a reader if its label promises different content.
How do you record the review result?
Prepare a short working table: address, expectation, result, date and evidence. Separate an issue that blocks publication from a later improvement. For each task, name the person who will confirm the fix. After deployment, check the correct address again.
- Direct entry to a subpage and refreshing it work.
- The status matches the actual content.
- The HTML contains the page's own title and main content.
- The canonical URL, links and sitemap are consistent.
- Drafts remain outside public publication.
- The sharing preview uses an accessible image.
This work complements the functional website review. Put the requirements in the brief early. The Aeneia Design offer covers the project and publication. For the related method, consult the article on a functional website acceptance checklist (in Polish). Further guidance is available in the article on preparing a premium website brief (in Polish).












