How to test a website on different browsers

A cross-browser testing workflow for small teams: pick browsers from real traffic, write plain-English checks, and keep evidence for every pass or fail.

cross-browser testingbrowser testing

Pick the browsers your analytics show people actually use. Test the flows that create an account or take a payment in each of those browsers, and keep the screen, the console, and the network log for every pass and every fail. A short list you run is the matrix.

Pick the browsers from your traffic

A browser matrix is the short list of browser, operating system, and device combinations you agree to test. Build it from your analytics. A chart of worldwide browser share is someone else's traffic.

Open the analytics tool you already use. Read browser, operating system, and device for the pages that create an account or take a payment: sign-up, login, checkout, billing. A browser with no traffic on those pages can wait.

Three engines sit under most of the names in that table.

  • Chromium. Chrome and Edge.
  • WebKit. Safari.
  • Gecko. Firefox.

The name is not the whole environment. Desktop Chrome and Chrome on a phone are different windows. Safari on macOS and Safari on iPhone are different windows. Chrome on iOS is not a stand-in for the Chrome you have open on a laptop. Read the browser and the operating system on the same row.

Keep the matrix small enough that someone runs it.

WhenWhat goes on the list
Every releaseThe desktop browser with the most traffic on sign-up and checkout
Every releaseThe mobile browser with the most traffic on those same flows
Layout, form, cookie, or media workThe next engine your analytics still show
A customer names a browserThat browser and operating system, before you add anything else

Review the list when the traffic mix changes.

What breaks across browsers

These are the differences worth a look. They are not a full compatibility catalog.

  • Layout. Flex, grid, sticky bars, and viewport height do not land in the same place in every engine. A submit button can sit under a fixed footer at one width.
  • Form controls. Date inputs, file inputs, and selects use the engine's own UI.
  • Cookies and storage. Safari restricts third-party cookies and can partition storage that Chrome still accepts. A login that depends on a third-party cookie can succeed in Chrome and fail in Safari.
  • JavaScript. An API can exist in Chromium and be missing or partial in WebKit or Gecko. The console on the browser that failed is the evidence.
  • Media. Autoplay rules differ. A video that starts itself in one browser can stay paused in another.

For a failure that only happens in Safari, open the Web Inspector console on that page. The steps are in how to open the Safari console.

Manual checks and scripted engines

A manual pass is enough when the matrix is two or three browsers and the change is small. Open the flow, finish it, and look at the layout. Manual passes get skipped once the same flows are supposed to run on every deploy. The second browser is the one people skip.

Scripted coverage across engines is code you keep:

  • Playwright drives Chromium, Firefox, and WebKit from one API. Playwright's WebKit build is not Apple's Safari on a phone.
  • Cypress is a JavaScript test runner you commit. Its cy.prompt command is documented for Chromium-based browsers only.
  • Selenium drives a browser you configure. Selenium Grid runs those scripts on remote machines, including other browsers, versions, and operating systems.

A device cloud is how you open a real Safari or a real phone when you do not have that device. Renting a device is a different job from a Chromium check.

Write the check in plain English

Monito runs the check in desktop Chromium. The default window is 1440×900. The hosted MCP check tool runs one public URL and an expectation in Chromium. A saved scenario and a CI run use that same browser. Firefox, Safari, WebKit, and a physical phone need a tool that launches those engines, or a person with the device.

The check is the outcome, written in plain English. A sign-up example:

Open the sign-up page. Create an account with a new email and a password the form accepts. The next screen is the account home, the new email is visible, and no error is shown.

The run ends Passed, Failed, or Error. Passed means the checks in the scenario were observed. Failed means the app did not do what you wrote. Error means the check could not verify the scenario. An error is not a pass. If Monito caused the error, the run is not charged. A test that finds a bug in the app still uses credits.

The run keeps screenshots, console messages, and network requests as method, URL, and status. It includes a bug report when the check finds one. A run with no login also records a video. A run that signs in keeps masked screenshots instead of a recording. The evidence is from Monito's run. It is not a recording of a visitor.

Start from the check command or the MCP server.

Run it when you deploy

Save the check if you want the same expectation again. Run it when the flow changes. On Pro or Team, CI can run it after a deploy.

  • GitHub Actions can wait for the verdict and link the evidence on the pull request. Creating that binding requires Pro or Team.
  • Vercel preview deployments can start the same checks. The native webhook does not run production deploys.
  • Netlify Deploy Previews can start the same checks from GitHub Actions or a deploySucceeded function. The function calls the deploy webhook. It does not hold the deploy open.
  • GitLab CI can call the CLI with the deployed URL. Our team has not run that example in GitLab. The API key it uses requires Pro or Team.

The Chromium checks can gate a deploy. Safari and Firefox still need a run in those engines when the change can affect layout, forms, storage, or media. That wider pass does not belong on every pull request. Run it on a cadence you will actually read.

Checklist

  • Read browser, operating system, and device analytics for sign-up, login, and checkout.
  • Write the matrix: browser, environment, and why it is included.
  • Keep desktop and mobile as separate rows.
  • Write the money flows in plain English: who the user is, what they do, and what must be true afterward.
  • Run each flow in each browser on the matrix you claim to support.
  • For the Chromium row, a Monito check records screenshots, console output, and network requests from that run.
  • For Safari and Firefox, use a browser or a runner that launches that engine.
  • On a failure, read the screen, the console, and the network log before changing the product.
  • Put the Chromium checks on GitHub Actions, a Vercel preview, or GitLab CI when that deploy should wait for them. A Netlify Deploy Preview can run the same checks. The GitHub Actions job waits. The deploy webhook does not.
  • Remove checks nobody reads.

FAQ

Do I need to test on every browser?

No. Test the browsers that show up in your analytics on the flows that create an account or take a payment. A short list you actually run is more useful than a long list you skip.

Is Chrome testing enough?

Chrome covers the Chromium engine, and Edge uses that engine too. Chrome does not cover Safari or Firefox. If those browsers appear in your analytics, a Chrome-only pass misses them. A Monito check runs in Chromium, so the pass is a Chromium result.

How do I test Safari without a Mac?

Monito does not run Safari. Use a Mac, or a device cloud that provides Safari or an iOS device. Playwright can launch its own WebKit build, and that build is not Apple's Safari. With Safari open, use the Web Inspector console.

How often should I run cross-browser tests?

Run the matrix when a change touches layout, forms, sign-in, or checkout. Also run the flows that must keep working on a schedule, because a dependency update can break a browser the change did not name. The whole matrix does not have to run on every pull request.

Plans

Current plans are on the pricing page.

A standard test is 1 credit, an Advanced test is 10, and Discovery is 20. Pro is $79/month for 700 credits. Team is $199/month for 1,800 credits. Managed is $750/month, limited to 10 teams, and includes the Team allowance. 50 free test credits every month. Your balance refills to 50 on the 1st of each month (UTC), counting credits reserved by running tests. Unused credits are not added again; purchased credits are kept. Your signup credits count as the first month. Free includes one daily schedule. Verify your email to receive monthly refills. The onboarding wizard does not charge its first run. Pricing.

Related posts

Compare Monito with other tools
All Posts