# Visual Regression Testing for Websites · WPCloudLab > Visual regression testing for WordPress and any other stack: desktop and mobile screenshots, pixel diffs, AI review that tells new content from broken layouts. Canonical: https://wpcloudlab.com/features/visual-regression [Home](/)/Visual regression visual regression · pixel diff · ai review # The site is up. It just looks broken. Most maintenance failures never write a line to an error log. A plugin update nudges the CSS, the hero collapses on mobile, the buy button slides below the fold — and every classic monitor keeps reporting green. Visual regression testing watches the thing your visitors actually see: the rendered page. [Get started](/signup)[how it guards updates →](/features/safe-updates) HTTP 200 means the server answered. It says nothing about the hero that collapsed, the font that never loaded, or the button now living below the fold. 001 / how it worksbaseline → capture → diff → verdict 01 ## Baselines you approved. The reference isn't whatever the site looked like yesterday — it's a version you signed off. Comparing against an approved baseline is what makes a diff mean something. 02 ## Captured desktop and mobile. Every capture runs at desktop and mobile viewports, because layouts don't fail equally. A page can look perfect at 1440px while its navigation is a vertical pile at 390px. 03 ## Pixel-diffed, not eyeballed. Each capture is compared against its baseline pixel by pixel. Humans skim; diffs don't. A section that shifted forty pixels in a footer nobody scrolls to still gets flagged. 04 ## Classified by AI. An AI reads the diff and asks the only question that matters: new content, or broken layout? A fresh blog post is routine. A hero missing its image is not. 05 ## Escalated when it's close. Diffs the AI can't call with confidence go to engineers. Standard plans include AI-assisted visual change review; Extended adds priority review by our team. 06 ## Approved when intentional. Redesigned on purpose? The new look goes through a short approval flow and becomes the new baseline — so tomorrow's diffs compare against your intent, not your history. 002 / where it runsmonitoring · updates · review depth ## As standing monitoring. Visual regression monitoring runs on every plan, Economy included. It catches the drift that happens between updates — a theme edit here, a plugin’s stylesheet there, a font that stopped loading. ## Around every update. The same machinery wraps every update we run — screenshots before and after, diffed and reviewed, with automatic rollback on regression. The full pipeline is on the [Safe Updates page](/features/safe-updates). ## With graded attention. AI handles the clear verdicts on its own. Standard adds AI-assisted review of visual changes; Extended adds priority review by our engineers. The plans sit side by side on the [pricing page](/pricing). 003 / beyond wordpressany rendered page ## If a browser can render it, we can watch it. Captures run from the public URL in a real browser — no agent, no plugin, nothing to install. Next.js apps, static sites and custom stacks get the same treatment as WordPress: approved baselines, desktop-and-mobile captures, pixel diffs, AI review. It runs alongside [uptime monitoring](/features/uptime) as part of the monitoring-only coverage for [non-WordPress sites](/for/non-wordpress). One watches whether the server answers; the other watches what it answers with. 004 / questionspixels, plainly ## Visual testing questions, answered. Which pages do you capture? The ones that matter: typically the home page, key landing and template pages, and for stores the product and checkout flow. The set is agreed when your site is connected and adjusted as the site grows — capturing everything would bury real signal in noise. Do you test mobile layouts too? Yes — every page is captured at both desktop and mobile viewports on every run. Mobile is where layout regressions hide, and it's usually where clients spot them first. How does the AI avoid false alarms on normal content changes? It classifies what changed rather than just measuring how much. A new blog post, a swapped photo or an updated price reads as routine content; a collapsed section, missing element or overlapping text reads as a regression. Diffs it can't call with confidence go to a human — AI-assisted review on Standard, priority review on Extended. What happens when a real regression is found? It's flagged with the before/after evidence and an engineer looks, usually inside the hour. If an update caused it, the update rolls back automatically to its restore point. Either way, your dashboard shows what changed, when, and what was done about it. I'm redesigning on purpose. Will everything get flagged? Once. Intentional changes go through the approval flow: you approve the new look, it becomes the new baseline, and future diffs compare against it. A redesign is not treated as a month-long emergency. Does it work for Next.js or static sites? Yes. Captures run from the public URL in a real browser, so anything a browser can render can be watched — Next.js, static sites, custom stacks. It's part of the monitoring-only coverage for non-WordPress sites, at the same plan pricing. free audit · send a url · no card ## Monitoring that actually looks at your site. Uptime checks watch your server. This watches what your visitors see. Connect a site in minutes, or send your URL for a free audit — we’ll tell you where a visual baseline would earn its keep. [Get started](/signup)[see plans & pricing →](/pricing) --- Get started: https://wpcloudlab.com/signup · Talk to an engineer: https://wpcloudlab.com/contact · Full agent index: https://wpcloudlab.com/llms.txt