A beautiful launch.
A checkable foundation.

1. Check the public response.

Open the canonical page without a signed-in account. It should return useful HTML containing the main explanation and ordinary links. Check the HTTP status, the page-level robots directive and any X-Robots-Tag response header together.

For theoretic.si, product pages are explicitly public. Account forms are excluded from indexing, and private workspaces require a session. Missing pages return a genuine 404. Crawl restrictions never replace account access control.

2. Give each page one clear address.

Use a self-referencing canonical for each distinct page. Resolve alternate trailing-slash addresses consistently, and keep tracking parameters from becoming separate preferred pages. Include useful public canonical URLs in the sitemap and link to them from the site.

Our sitemap is built from the same explicit public-page list as the server routes. Adding a page means deliberately adding its route; private workspace URLs are not discovered or collected into that list.

3. Check the image behind the logo markup.

Our October 10 audit found that the newly introduced brand images inherited a noindex response header. They loaded correctly in a browser, but that header conflicted with the requirements for an organization logo in Google’s structured data.

We added the public brand files to the explicit indexing allowlist. The logo now has its own accessible image URL and is referenced consistently in the organization metadata. Checking that an image “looks right” is different from checking its response headers.

Logo requirements: Google Organization structured data ↗.

4. Cache assets without caching private pages.

The homepage previously requested six separate stylesheets, and its static assets were served with no-store. We combined the public CSS into a minified file whose filename includes a content fingerprint.

A changed bundle receives a new URL. That lets the browser keep the old file without confusing it with a newer release. Other eligible public assets use validators so an unchanged file can receive a small “not modified” response. Account HTML and private workspace responses remain excluded from storage in shared caches.

This is a delivery improvement, not a claimed Core Web Vitals score. Measure loading, responsiveness and stability separately, on realistic devices and connections. Field data may be unavailable until a site has enough real traffic.

5. Verify discovery after release.

Check that Google can fetch the sitemap and that the important URLs are represented in Search Console. Use URL Inspection to distinguish a crawl problem, an indexing restriction and a page still awaiting indexing.

A successful sitemap submission is not confirmation that every page is indexed. An indexed page is not confirmation that it ranks for a valuable query. Record those outcomes separately, and review them after reporting has caught up.

For a new or substantially improved page, request indexing through the normal Search Console workflow when appropriate. Repeated requests do not make crawling faster. Google’s recrawl guidance ↗ explains the process.

We will report search outcomes when there is enough evidence to support them. This example documents the technical changes to our own website; it does not claim a traffic or ranking increase.