Blog / Technical SEO /

A practical review of response headers for site owners

A response header can affect caching, security, redirects or how a browser treats a page. Yet most site owners only look at headers when something breaks, and then it’s a bit late for a calm look.

A review doesn’t mean changing every value. First, understand which headers the site sends, which layer adds them and what they’re meant to control. Then check the response from the visitor’s side, and involve the right technical owner before touching production settings.

Capture the real response

Check representative URLs across the main site, subdomains and error states. Record the status code and headers for each, and note whether a CDN, proxy, application or web server may be changing them.

A command-line request and a browser request can differ by method or cache path. Use a repeatable method and explain its limits. Don’t publish sensitive infrastructure detail in a public audit.

Understand intent before changes

Review headers relevant to redirects, caching, content type, security and robots directives. Confirm the expected behaviour with the platform or security documentation.

A header that looks odd may be required by an application, and removing it can break something. Compare the response with the owner’s policy and test on staging where possible. A missing header isn’t the same as a demonstrated user or security problem.

Verify the result

If a change is approved, record the previous value, the owner and the rollback route. Test pages and assets after deployment, including redirects and cached responses.

Recheck from more than one path when a CDN is involved. I’ve spent enough time with DNS, SSL and Cloudflare settings to distrust a header that I’ve only seen from one place. Keep the review about evidence and risk, not a checklist score.

Keep the work reviewable

Start with one question the current setup can answer, such as what the homepage returns from two locations.

Record the date, method and location of each request. A header seen through a cache can be a snapshot of the cache.

Compare one ordinary page with one exception, like an error page or a static asset.

Know which layer sets which header. Without that, you may “fix” the wrong layer and see no change.

If two tools report different headers, check method, timing and scope before choosing one.

Keep confirmed facts, likely explanations and proposed changes in separate lines.

Test any rule on a representative URL on staging first and inspect the live response afterwards.

Name the person who can change the CDN, the server and the application. When evidence is incomplete, say what is missing and choose the next check.

Record each change with a reason and an owner, and revisit the same URL after deployment.

Write down, in a line each, what every important header is for and who owns it. Next time something looks strange, you’re comparing against a record, not somebody’s memory.

A good handover names the person who can change the CDN, the person who can change the server and the person who can change the application. A broad “tech team” is hard to ask at short notice.

Keep the maintenance note brief but specific: the owner, the source and the event that should trigger another review, such as a hosting move or a security update.

When you report a finding, put its limitation next to the recommendation, so a colleague sees what’s confirmed, what’s uncertain and why the proposed step is proportionate.

Keep a short list of assumptions still to confirm, such as whether staging mirrors production headers, with a name against each.

Pay particular attention to caching. A page that is cached for too long can show stale content after an update, and one that is never cached can slow the site down. Neither shows up as an error, so you only see them by reading the headers.

Remember that a redirect is also a response. Check that the status code is the one you intend and that the destination header points where the redirect register says it should.

Share the findings in plain language with the owner. A list of header names means little to a shop owner, but “your homepage is cached for a day” does.

Which header would you investigate first if the site behaved differently across environments?