The most common website testing mistakes are testing only on the developer’s own setup, waiting until release week, treating an automated accessibility score as proof, overlooking real mobile conditions, and reducing performance to one stopwatch number. Avoid them by defining the environments and user tasks your site must support, checking small changes early, and combining automated checks with manual and human evaluation.
1. Testing only on your own browser and device
A page working on your laptop does not establish that it works for visitors using different browsers, operating systems, screen sizes, hardware, or assistive technologies. Universal coverage is impractical, so the goal is not to test every possible combination. It is to agree on a support range and test representative environments within it.
Build a target environment matrix
Work with the site owner or product team to identify the browsers, operating systems, device classes, and assistive-technology paths that matter to the audience. Record what is in scope, what is not, and which combinations are priorities. Include common desktop and mobile environments rather than assuming one browser represents them all.
- Coverage: Which supported browsers, operating systems, screen sizes, and assistive-technology paths are represented?
- Fidelity: Is a check running on physical hardware, an emulator, or a virtual machine? Choose based on what the test needs to reveal.
- Core functionality: Confirm that important tasks remain accessible even when the appearance or behavior is not identical across environments.
MDN’s cross-browser testing guidance emphasizes that a developer’s own machine is not a stand-in for the audience. Real devices can reveal issues that are easy to miss in desktop simulation; emulators and virtual machines can extend coverage when physical hardware is unavailable. Neither a single phone nor one browser covers the full target range.
2. Saving all testing for the release crunch
When checks happen only at the end, regressions are harder to isolate and there is less time to fix them. Test each small implementation phase before committing it, then broaden coverage as the feature matures.
Use a staged testing rhythm
- While implementing: Check each small part in a couple of stable desktop browsers. Catch obvious layout and behavior issues close to the change that caused them.
- As the feature takes shape: Add a keyboard-only pass or basic screen-reader navigation check, plus a mobile platform in the project’s target range.
- Before release: Expand to the agreed browser and device matrix and run the relevant accessibility, task, and performance checks.
This sequence is a starting point, not a substitute for the project’s support requirements. A change that affects an unusual or high-priority environment may warrant checking that environment earlier.
3. Treating an automated accessibility score as proof
Automated accessibility tools can flag common issues, but an automated result alone cannot establish that a site conforms to an accessibility standard or is usable. W3C describes WCAG success criteria as testable, while noting that some assessments require human testers for part or all of the evaluation. Conformance checks and usability testing answer related but different questions.
Combine tool checks with manual review
- Use automated checks as one input. MDN lists Lighthouse accessibility audits, axe, and WAVE as examples of tools; their inclusion does not mean any one tool can establish conformance.
- Check that the page’s logical source order still makes sense with CSS disabled.
- Review text and background contrast, and make sure color is not the only way information or status is conveyed.
- Operate the site without a mouse to check keyboard navigation.
- Run task-based usability testing, and include people with disabilities in usability test groups where possible.
A passing scan can coexist with a confusing task flow or a problem that requires human judgment. Record both technical findings and what participants can actually accomplish.
4. Ignoring real mobile conditions
A responsive layout viewed in a desktop window is not the same as checking the site on a mobile environment in its support matrix. Browser behavior, hardware, and input conditions can differ, so include Android or iOS testing when those platforms are in scope.
Choose the right level of device fidelity
Use physical devices where possible for checks that depend on real hardware or interaction. Emulators and virtual machines are useful ways to extend coverage when physical devices are unavailable, but treat them as a different test environment rather than proof of behavior on every real device. A physical Android phone can be a useful aid; it is not a replacement for the target matrix.
5. Treating speed as one stopwatch number
Performance includes whether a page loads, responds promptly to interaction, and scrolls or animates smoothly. A single load-time reading does not cover all three. Media, JavaScript, HTML, CSS, and rendering choices can all affect performance.
Measure the experience you intend to improve
- Check page loading, interaction responsiveness, and smoothness as distinct dimensions.
- Record the environment and conditions for each measurement so results can be interpreted and compared meaningfully.
- Use more than one run before drawing conclusions; do not describe a site as fast based on one unqualified test or metric.
- When a result is poor, connect it to the user experience and the part of the page or interaction being measured before choosing a fix.
MDN treats measurement as part of the development workflow. A useful performance check says what was measured and under what conditions, rather than presenting a number without context.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Clear out junk files and repair common Windows errors3Fix the driver behind crashes, sound loss and screen glitches6. Making vague accessibility claims or leaving the standard unspecified
A test plan should name the accessibility standard and conformance target it is evaluating. “Accessible” without a defined basis is too vague to guide implementation or review.
Rank #4
State the standard and the project’s obligations
W3C identifies WCAG 2.2 as a Recommendation and advises using the latest WCAG version when developing or updating policies. WCAG 2.2 adds nine success criteria beyond WCAG 2.1. The appropriate conformance target and any legal or contractual obligations depend on the project and jurisdiction; no WCAG level automatically settles every obligation.
W3C’s WCAG 2.2 overview records an update on 12 December 2024. Confirm the applicable standard and requirements when setting the project’s test plan rather than relying on a generic claim of compliance.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.7. Treating screenshots as a substitute for functional testing
Screenshots can help compare rendered pages across supported viewports, but a visual capture does not establish that forms, navigation, keyboard access, assistive-technology flows, or other interactions work. Use visual checks as one part of a broader test plan, alongside functional, accessibility, and performance evaluation.
Best Value
Capture representative states
For visual review, specify the URL, viewport, and page state you want to compare. Dynamic content, consent dialogs, and other overlays can affect what appears in a capture, so consider whether the test should preserve those elements or intentionally remove them. Keep the screenshot comparison tied to an expected visual result; do not treat a matching image as proof of usability or conformance.
Or skip the browser setup
For screenshot capture as one part of your visual checks, ScreenshotNeo can return an image or PDF from one GET request. Its capture process can accept cookie/consent banners and remove 60+ known consent platforms, newsletter popups, and chat widgets; each step can be turned off. Bot checks/CAPTCHAs, blank pages, timeouts, failed loads, and cache hits are not billed, and response headers identify the page verdict and billing status. Its MCP server provides take_screenshot, get_page_info, and capture_pdf for AI agents using Claude, Cursor, or another MCP client. A free plan includes 1,000 screenshots a month with no card; paid plans start at $5 for 3,000.
Learn about ScreenshotNeo. Example using cURL (replace the URL with the page under test); see the ScreenshotNeo documentation for options:
curl -G "https://api.screenshotneo.com/v1/shot" -d access_key=YOUR_API_KEY --data-urlencode url=https://stripe.com -o shot.webpSign up for 1,000 free screenshots a month, with no card.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Turn the fixes into a practical test plan
- Agree on the supported browser, operating-system, device, and assistive-technology range with the site owner or product team.
- For each change, identify the user task and the environments most likely to expose a regression.
- Check small implementation steps early, then expand toward the full agreed range before release.
- Combine automated accessibility checks with manual review and task-based usability evaluation.
- Measure loading, interaction responsiveness, and smoothness in recorded conditions.
- Name the accessibility standard and conformance target, and account separately for project-specific legal or contractual requirements.
Keep the plan proportionate to the site, but make its scope explicit. That makes gaps visible before release instead of leaving “tested” to mean only that the developer’s own browser looked right.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
