Visual regression testing
Visual regression and UI testing
A functional test can pass on a page no user could use. Klarent compares what the screen actually looks like, and only flags the differences that matter.
Beyond functional tests
The bugs functional tests miss
Broken layouts
Overlapping elements, broken layouts and text that runs off the screen, on a page where every button still works.
One broken breakpoint
A CSS regression that breaks one responsive layout breakpoint and nothing else, so nobody notices on their own screen.
Cross-browser rendering
Rendering differences between browsers on the same markup, invisible to a test that only checks the DOM.
Component library updates
A design system or component library update that shifts spacing across the whole product at once.
Styling drift
Wrong colours, missing icons and fonts that failed to load, while the element the test looks for is still there.
Hidden on small screens
A call to action pushed below the fold or behind another element when the responsive layout collapses on mobile viewports.
Too many diffs, too little signal
Why most visual testing gets abandoned
Visual regression testing rarely fails because screenshots are hard to capture. It fails because the real regressions drown in noise.
Pixel noise
Pixel-level comparison flags every anti-aliasing difference as a failure.
Dynamic content
Dates, IDs and personalised blocks fail on every single run.
Baseline upkeep
Keeping every baseline image current becomes a job of its own.
Approve everything
So the team approves every visual diff, and the checks stop meaning anything.
How it works
How visual testing works in Klarent
Step 1 - Capture
Baselines are captured as part of the end-to-end journeys you already run.
Step 2 - Compare
Differences are assessed by significance, not raw pixels, so cosmetic noise is filtered out.
Step 3 - Review
Review each visual diff side by side. Accept an intended change, and it becomes the new baseline image.
Platform capabilities
Automated visual testing without the false positives
Significance-based comparison
Differences are weighed by whether a user would notice them, which cuts the false positives that sink pixel diffs.
Dynamic content handling
Timestamps, rotating carousels and personalised content are filtered out, and regions can be ignored entirely.
Confidence-gated baselines
High-confidence intentional changes update the baseline automatically. Anything less certain goes to a person first.
Multiple baselines
Keep separate baselines for A/B variants and localised interfaces.
Layered onto end-to-end tests
Visual checks run inside your existing end-to-end tests, not as a separate suite to maintain.
Baseline approval history
See who approved each baseline change, and when.
Coverage
What you can check visually
Visual UI testing covers more than the page you land on, from single components to the states users only see when something goes wrong.
Full-page layout
Whole pages and their responsive layout, at every breakpoint.
Components
Individual components, checked for consistency across the product.
Cross-browser rendering
The same page across Chrome, Firefox, Safari and Edge.
State-dependent screens
Empty states, error states and loading states.
How they fit
Visual, functional and cross-browser testing
Automated visual testing closes a different gap from the tests you already run. Each one answers its own question about the same page.
Functional testing
Does it work? Functional tests check behaviour: that clicks, forms and journeys reach the right outcome.
Visual regression testing
Does it look right? Visual checks compare appearance with an approved baseline, catching the layout and styling bugs functional tests pass.
Cross-browser testing
Does it work and look right everywhere? Cross-browser runs repeat both checks across Chrome, Firefox, Safari and Edge.
Workflow
Fits your pipeline
Visual diffs in the pull request
Visual testing automation fits the review flow you already have: diffs are surfaced in the pull request, so changes are reviewed before merge.
Approval workflow and change history
Every accepted or rejected visual change is recorded, so the baseline never changes silently.
Results in Slack or Teams
Flagged visual differences reach QA, developers and designers in the channels they already use, instead of waiting in a dashboard.
Integrations
Fits into your existing stack
Connect to GitHub, Jira, Slack, and more - no reconfiguring your workflow.
Add visual checks to a journey you already test.
Pick an end-to-end journey your team runs today, and see Klarent capture baselines, compare by significance and flag only the differences that matter.
FAQs
Visual regression testing compares what a page looks like now with an approved baseline image, to catch unintended changes in layout, styling or rendering. It finds the bugs a user sees but a functional test passes, such as overlapping elements or text running off the screen. For a deeper look, see how AI improves visual regression testing.
Functional testing checks behaviour: that a button works, a form submits and the right outcome happens. Visual testing checks appearance: that the page looks the way it should. A page can pass every functional assertion and still be unusable, which is the gap visual checks close.
Differences are assessed by significance rather than raw pixels, and cosmetic noise such as timestamps, rotating carousels and personalised content is filtered out before it reaches a reviewer. Regions that change on every run can also be ignored.
Your team does. High-confidence intentional changes update the baseline automatically, while medium- and low-confidence differences go to a person for confirmation before anything changes. Every approval is recorded.
Visual diffs are surfaced in the pull request, so a change is reviewed on its branch before merge. Once it is accepted, it becomes the new baseline, and the approval history shows who accepted it and when.
Yes. Visual checks run across Chrome, Firefox, Safari and Edge, on desktop, tablet and mobile viewports, so rendering differences between browsers and broken responsive breakpoints are both caught.
Yes. Check full-page layouts at every breakpoint, or individual components to keep your design system consistent across the product.
No. Automated visual testing is layered onto the end-to-end tests you already run, so baselines are captured along the same journeys, with no second suite to write or maintain.