Skip to content
Dev Ring Technologies
Let’s talk
Back to blog

Browser cache: check that a website update is visible

A correction appears for you, but someone else still sees the old text or the wrong image. Before asking for every cache to be cleared, prepare a reproducible comparison. This guide helps you check a public update and give the website team useful observations. It does not describe Dev Ring’s configuration or promise a universal propagation time.

1. Define the expected evidence before reloading

In our fictional example, a page replaces “Appointments on Tuesday” with “Appointments on Thursday” and receives a new illustration. Record the exact address, language, approved text and expected image. Add the announced publication time and, if your team provides one, the version reference. A page opening successfully does not prove that this correction is live.

Choose a simple journey too: open the page directly, reach it from its listing, then return after another page. Keep the browser, device, time and visible result for each check. Check public content only; use an intended test environment for actions that send messages or change data.

2. Separate the browser from other possible copies

The HTTP cache can store a response in the browser or in a shared cache, such as a delivery intermediary. The site may also have its own caching mechanisms. Clearing a local copy therefore does not necessarily remove the copy responding upstream. Ask the team which layers actually exist on this website.

First compare a returning visit with a fresh test profile, without added extensions, at the same address and in the same language. A difference is a clue, not proof that the browser alone is responsible. Keep both observations before any intervention and avoid deleting data from your everyday profile.

Source : IETF — RFC 9111: HTTP Caching · Source : Mozilla MDN — HTTP caching

3. Use reloading as a limited diagnostic

In Chrome, the team can open the developer tools’ Network panel, select Disable cache and reload. Browser caching is disabled while the tools remain open. Record whether the expected text and image appear, then clear the checkbox and repeat a normal visit. A result obtained only with this option is insufficient to accept the release.

This check concerns the browser cache; it does not guarantee bypassing every cache on the site. If an offline application or service worker is present, have the team check its storage too. To share evidence, prefer a targeted screenshot: a full network export can contain sensitive information.

Source : Chrome DevTools — Network features reference · Source : Mozilla MDN — HTTP caching

4. Read the clues without confusing directives

Ask the team to record the resource address and its response headers, especially Cache-Control and, if present, ETag. The unqualified no-cache response directive allows storage but requires validation before reuse. no-store requests that the response not be stored. These names therefore describe different rules; do not copy a blanket configuration to solve one case.

ETag is a validator for a representation. In a conditional request, a 304 response can indicate that the stored copy may be reused; by itself, it does not mean the expected new text has been published. Always relate technical clues to the visible content and announced version. A freshness lifetime is not a universal publication commitment.

Source : IETF — RFC 9111: HTTP Caching · Source : Mozilla MDN — Cache-Control · Source : Mozilla MDN — ETag

5. Check the page and its files together

Current text with an old image is still incomplete in our example. Open the image actually loaded and record its address, then compare it with the planned delivery. MDN describes changing resource URLs when their content changes as a way to allow longer caching. The team can thereby manage versioned files while keeping a strategy suited to the page.

Do not rename public pages yourself or add an arbitrary parameter to every link. Ask which file should change and how the page should reference it. If a targeted purge or correction is needed, have the website owner responsible perform it, then repeat the returning and fresh visits. A successful purge does not replace those observations.

Source : Mozilla MDN — HTTP caching

6. Accept the result and keep its limits

Repeat the initial journey with normal settings, then in another published language and on a device relevant to the change. Compare text, image, links and back navigation. If visitors receive different results, keep their times and conditions separate; do not infer a general fix from a single screen.

Complete the worksheet for each affected page. An actionable report identifies the expected result, observation and sequence that reproduces the difference. After intervention, retain the newly tested version and cases still unchecked. This acceptance check measures neither a performance gain nor a search-ranking result.

Website update verification worksheet

Website update verification worksheet
CheckResult to observeEvidence to keep
ReferenceApproved text and imageAddress, language and version
Returning visitUpdate visible without special settingsTime and targeted screenshot
Fresh profileSame expected contentEnvironment and comparison
DiagnosticCache effect separated from final resultOption used and recorded clues
Related filesConsistent page and imageAddresses actually loaded
After correctionReproducible journeyVersion and remaining limits

Frequently asked questions

Should every visitor be asked to clear their cache?

Keep this as a possible, controlled diagnostic. First check the versions being served and the update strategy with the team. Acceptance must also succeed under normal returning-visit conditions.

Does a 304 response mean publication failed?

No. It can be part of validating a stored copy. Compare the reused content with the approved change; the status alone does not settle the question.

Is a new private window enough evidence?

It provides a comparison point, not a guarantee about every cache and visitor. A fresh test profile and a returning visit, documented separately, help reproduce the issue.

Prepare your next update

Choose one page and a specific change. Agree on the expected evidence with the team, then keep a before-and-after publication observation for both types of visit.

Sources and scope

Primary sources and publisher documentation consulted on November 5, 2026. HTTP rules come from the RFC and practical explanations from MDN and Chrome. The scenario and worksheet are editorial suggestions; no client observation, site configuration or performance measurement is claimed.