When should you use Nothing changed since last time?
visual-diff · visual · recommended
When should you use this check?
Use this on pages whose appearance is a commitment: a signed-off landing page, a pricing table, a regulated disclosure, a campaign destination you are paying to send traffic to. Its value is highest on a schedule rather than on demand, because the changes it catches are the ones nobody announced — a CMS edit, an expired asset, an embed that updated itself overnight. It needs a previous capture of the same page at the same width to compare against, so the first run establishes the baseline and the second one starts earning.
When should you not?
Do not point it at pages that are supposed to change: a homepage with rotating hero content, a feed, a personalised dashboard, anything with a carousel or a live counter. On those, every run reports a difference and the check becomes noise you learn to dismiss. It is also the wrong tool for verifying an intended redesign — after a deliberate change, the comparison will fail by definition, and what you want then is a fresh baseline rather than a finding.
What does it inspect?
Optional monitoring check for pages that should not change. Compares the current screenshot against the previous capture of the same URL and viewport; fails when a meaningful difference is detected so a human can confirm whether the change was intended. Requires a prior capture — the first run always passes. Best used on a recurring schedule.
What does a failure mean?
A failure means this capture differs visibly from the last one taken of the same page at the same width, and the finding describes what moved. It is a neutral report, not a defect claim: the change may be exactly what someone intended and forgot to mention. That is the point — it converts a silent change into something you have to look at once, which is cheaper than discovering it from a customer.
What are the Standard defaults — and why?
One parameter carries the check: `comparisonCriteria`, a written instruction rather than a numeric tolerance. Its Standard default asks for any visually significant layout, branding or content change compared to the previous screenshot for this URL and viewport — significance is judged rather than counted, which is why there is no pixel-percentage dial to turn. That is deliberate: a percentage threshold treats a shifted legal disclaimer and a re-compressed background photo as the same size of event, and they are not.
How does an agent call it?
{
"type": "url",
"url": "https://example.com",
"validations": [
"visual-diff"
]
}
Who governs the Standard?
Schemafirst.org publishes community-governed standards for digital QA. ArbiterQA is a sponsor and commercial licensee of those standards; citation does not mean Schemafirst operates ArbiterQA.
Author ArbiterQA · Reviewed by ArbiterQA · 2026-08-14