Visual regression testing with screenshots means capturing every important page at each screen size before a release, capturing the same pages at the same sizes after it, and comparing the two sets to catch anything that changed by accident. You do not need a test framework to start: a consistent URL list, fixed viewport widths, a strict naming pattern and a careful side-by-side review will catch most broken layouts, missing images and shifted buttons. This guide sets out a simple manual workflow you can run on a Windows PC.
Automated visual testing tools can do the comparison for you, but they take time to set up and maintain. For small teams, agencies and site owners, a repeatable manual process is often the right first step.
What visual regression testing catches
Functional tests check that a button works. Visual checks catch the things functional tests miss:
- A style sheet change that breaks the header on one template.
- A font that fails to load, so every heading falls back to a different typeface.
- An image path that changes, leaving empty boxes on product or blog pages.
- A layout that works at desktop width but overlaps on tablets or phones.
- Content that disappears because a plugin, widget or embed stopped rendering.
Most of these are obvious the moment you see the page. The problem is seeing every page at every size, which is why screenshots help.
Step 1: Choose the pages
You do not need to capture every URL on a large site. Pick pages that represent each template and each important journey:
- The home page and main landing pages.
- One or two examples of each template: a blog post, a category page, a product page, a contact form.
- Pages with complex components: pricing tables, sliders, maps, checkout steps.
- Any page that broke in a previous release.
Keep the list in a plain text file, one URL per line, and reuse the same file every time. For a smaller site, you may prefer to capture everything from its sitemap; our guide to screenshotting every page of a website explains how.
Step 2: Fix the screen sizes
Comparisons only work if both sets were captured at identical widths. Choose three or four and never change them between runs.
| Width | Represents |
|---|---|
| 1920 px | Full HD desktop |
| 1366 px | Common laptop |
| 768 px | Tablet in portrait |
| 390 px | Modern phone |
If your style sheet has breakpoints at other widths, add a capture just either side of each one, since that is where layouts tend to break. Our reference of common screen resolutions for testing lists more options.
Step 3: Capture the "before" set
- Load your URL list into the capture tool.
- Choose full page, so changes far down the page are included.
- Select your fixed widths.
- Set a load delay of a few seconds so fonts, images and scripts have finished.
- Choose PNG. Lossy formats add compression noise that makes small differences harder to judge.
- Set a naming pattern that includes the domain and a running counter, and send the run to a folder named
beforeor with the date and release number. Keep the list and the widths in the same order every time, so the counters line up between runs. - Run the batch and spot-check a few images.
Capture "before" against the same environment you will test "after" on. Comparing live production with a staging server mixes real regressions with environment differences such as test data and analytics banners.
Step 4: Capture the "after" set
Once the release is deployed, run exactly the same batch: same list, same widths, same delay, same format, same naming pattern, but a different output folder. Because the list, widths and pattern are identical, before\example.com_003.png and after\example.com_003.png are the same page at the same size, ready to compare.
Step 5: Compare the two sets
This is the part that takes care. A few ways to do it without special software:
- Two File Explorer windows side by side. Snap the before folder to the left (Windows+Left) and the after folder to the right (Windows+Right), switch both to Extra large icons and scroll together. Big layout changes stand out even at thumbnail size.
- Flip between images. Open the pair in the Windows Photos app or a browser, and switch back and forth quickly. Your eye notices movement, so a shifted button or changed spacing jumps out.
- Compare file sizes and heights. In File Explorer's Details view, add the Dimensions column (right-click a column header, then More). A page that is suddenly 2,000 pixels shorter or taller has changed, even if it looks similar at a glance.
- Use an image diff tool if you need pixel-level detail. Several free and open source tools highlight the differing pixels between two images.
Record what you find in a simple table: page, width, what changed, expected or not. Expected changes become the new baseline; unexpected ones become bugs. For writing those up, see our guide to taking screenshots for bug reports.
Keeping results trustworthy
False alarms waste time and train people to ignore real problems. Reduce them by controlling the things that change on their own:
- Rotating content: sliders, testimonials and "latest posts" change between runs. Note which pages have them and judge those areas by layout, not content.
- Cookie banners and pop-ups: they appear the same way in both runs if both runs start fresh, which is usually fine.
- Animations: a longer load delay lets entrance animations finish before the capture.
- Lazy-loaded images: if lower images are blank in one set, the delay was too short. See full page screenshots missing images.
- Dates and counters on the page: expected to differ, so ignore them.
Running the workflow with Bulk Captures
Bulk Captures handles the capturing side of this workflow on Windows. You load the URL list from a TXT or CSV file or a sitemap, choose full page or viewport, pick several widths such as 1920, 1366, 768 and 390 in the same run, set the load delay and naming pattern, and click Capture. It visits each page, waits for it to load, retries pages that fail and saves every screenshot to the folder you chose.
Be clear about what it does not do: Bulk Captures does not compare images or highlight differences, and it does not schedule runs for you. You start each run yourself, before and after the release, and do the comparison with the methods above or a separate diff tool. What it gives you is two complete, consistently named sets of screenshots, which is the hardest part to get right by hand.
A simple release checklist
- Update the URL list if new templates were added.
- Run the "before" batch and keep the folder.
- Deploy the release.
- Run the identical "after" batch to a new folder.
- Compare pairs, starting with the phone width, where breakages are most common.
- Log unexpected changes as bugs, and keep the "after" set as the next baseline.
Frequently asked questions
What is visual regression testing?
It means comparing screenshots of a site or app from before and after a change to spot layout, styling or content that broke unexpectedly.
Does Bulk Captures compare screenshots automatically?
No. It captures and names the before and after sets consistently, and you compare them yourself or with a separate image comparison tool.
How often should I take regression screenshots?
Before and after each release is the usual rhythm, plus after any change to shared templates, fonts or style sheets.
