The site editor (CMS) already has the corrected title—the page name shown in the browser tab—but visitors still receive the old one. Saving the text again may not help: the old response may be in the CDN, application, or browser cache. Compare the HTML from the site server, the public CDN response, and the page in the browser to find the remaining old copy.
Find where the old text remains
A cache stores a response for reuse. The no-cache directive allows it to be stored but requires checking that it is up to date before use. Official documentation.
Record the new title in the editor and the actual old title on the public page. Then compare the HTML generated by the server with the response through the regular public route, using only authorized diagnostic access. If the origin server response is already new but the CDN response is old, saving the record again does not address the cause. If both are old, check the generator and its data source. If the HTML is new but document.title is old, check the client-side code. This will help you locate what needs changing: the HTML generator, CDN cache, or script that changes the title.
Compare the data and responses
| What to compare | What it shows | What to compare it with | What should match |
|---|---|---|---|
| CMS field | The new text is saved | Check the generator | Not just the editor |
| Site server HTML | The rendered page | Compare with the CMS | New values |
| CDN response | The public copy | Check that it is up to date | New HTML content |
| 304 | Reuse of the response | Compare the ETag or modification date | The client’s body is known |
Include guest and authenticated modes, the regular URL, and the diagnostic request provided for this purpose. An administrator may bypass the cache, so a successful view by an administrator does not confirm what external visitors receive. A parameter after ? may create a separate cache entry. A fresh page with a random parameter does not confirm that the regular URL has been updated. If the CMS or CDN varies the response based on a cookie or Vary header, record those differences alongside the result. Save the response body for each mode, not just the HIT or MISS status. The same cache header does not guarantee identical content in two responses.
Get the HTML and headers for the regular URL
Save the ETag, Last-Modified, Cache-Control, Age, and any available infrastructure headers alongside the extracted meta tags. These indicators help with verification, but their absence does not prove that there is no cache. Check which component sets the freshness period for the HTML and what CMS change triggers a cache update. For the title and description, check the source document, not an API request that returns the new text separately. Compare the canonical request URL: a technical variant with a trailing slash or different host may have a different cached response.
The terminal command will save the response for the public URL; pass it to a developer if needed.
curl -sS -D headers.txt -o response.html 'https://example.com/service/'Compare the response age and available cache headers from headers.txt with the canonical and robots tags in response.html. Then repeat the request after clearing the HTML cache for that specific page. The new document should contain the current tags without a bypass parameter in the URL. Replace example.com with your domain; this GET request does not change the site settings. curl help.
Look for the new title and description in response.html. If the server behind the CDN already returns the new value but the public URL returns the old one, update the CDN cache for that page. If the HTML is new but document.title is old, check the script that changes the title.
Check that the description updates
Change the description for one service on a test copy, keeping the previous value. Get the response from the origin server and the public response for the same URL. If the new text is visible only on the origin server, clear the HTML cache for that page using the CDN’s standard command or button, then repeat the regular GET request. Next, save another edit to the record and check automatic updating: manual cache clearing should not be required every time. This test does not establish the state of the search index; it checks the publishing chain. On a live site, make a safe editorial change or use a change that has already been made; do not replace public text with an arbitrary diagnostic marker.
Update the required copies
Check whether the CMS updates the cache for the page itself, the related list, and language versions. Clearing only one copy may leave the old title at a URL with a trailing slash or in another language. Add the necessary addresses to the CMS or CDN update handler. Do not disable the entire cache if the task is to ensure that one representation is up to date. After making the change, check both a cold request and a repeat request. Set the HTML retention period based on how often the editor changes this page. Static bundles and images have different requirements; do not automatically apply the new rule to them just because they are served by the same CDN.
Why 304 does not contain new HTML
For a conditional request, the client may send If-None-Match or If-Modified-Since and receive 304. The 304 response does not contain new HTML: the browser uses its saved page. Check what text is in that copy and whether the ETag or Last-Modified value changed after editing. If the server incorrectly returns the old ETag, the browser may continue showing the old text. no-cache allows storage followed by validation, while no-store prevents the response from being stored. Do not choose a directive based on the everyday meaning of its name. Check a regular full response and the provided validation process, recording both observations separately.
Save the record again and open the page as a guest
The regular URL, without an added query, returns the new meta tags both on the first visit and when opened again. A CMS change updates the cache as expected; manual clearing is not needed after every edit. Check an old search snippet separately, using its crawl date.
Read also: Caching strategies for web applications: from browser to CDN and How to monitor changes to Title, meta robots, and canonical. Site checks are available on the reChecker technical audit page.