Skip to content

Captures & Comparisons

A Capture is one screenshot, taken for one combination of environment, language, viewport, page, and (if configured) country profile and interaction state. A Comparison is a diff between two captures — usually a "before" and "after" of the same page.

Taking a manual capture

Use the Captures tab's capture builder to run captures on demand (as opposed to Monitors, which run automatically on a schedule).

Field What it means
Environments Which environment(s) to capture from.
Languages Which language(s) to capture.
Viewports Which viewport(s) to capture.
Pages Pick from your saved pages, or capture an ad-hoc URL you haven't saved as a page yet.
Country profiles Optional — apply your configured capture variables.

Capture builder: environments, languages, viewports, and page selection

Every combination you select multiplies together — 2 environments × 3 languages × 2 viewports produces 12 captures in one job, up to a limit of 500 per job (lower still if your plan caps languages/viewports/country-profiles per run). The builder shows a live count as you select, so you'll see the total before you submit.

If a page has interaction rules configured (see Projects), they're applied automatically — you don't choose them here.

The Chrome extension is required

Manual capture currently runs through the Chrome extension, not the server. When you click Start, the dashboard hands off to your installed extension, which does the actual screenshotting in your browser and uploads the results. If the extension isn't installed, or you're not using a Chromium-based browser (Chrome, Edge, Brave, etc.), you'll see a "Chrome extension not detected" or unsupported-browser message instead of captures running.

Because manual captures run in your own browser, only environments with No auth are available here — an environment that requires cookies, headers, or basic auth (see Environments) can't be captured this way. For those, and for anything you want to run automatically, use a Monitor instead — monitors always run on the server.

Reading a capture job

Each job shows a status: how many of its captures completed, failed, or are still running.

Status Meaning
Queued / Running In progress.
Completed Every capture succeeded.
Partially completed Some captures succeeded, some failed.
Failed None succeeded.
Cancelled You stopped it before it finished.

Stopping a job in progress cancels every capture that hasn't finished yet. Note that a cancelled capture still counts toward your usage for that period — it already occupied a capture slot even though it didn't complete.

From a finished job you can download every screenshot as a zip (all viewports, or one at a time), save the job's captures into a Collection, and mark individual captures to Keep permanently so they're exempt from your plan's normal retention cleanup (up to a quota).

Comparing two captures

Open any completed job and select two captures to compare directly — useful for a quick "did this change?" check outside of a monitor. (Monitors also produce comparisons automatically every run — see Monitors.)

The comparison viewer shows:

View What it shows
Side by side Both screenshots, with the changed regions outlined.
Overlay A draggable slider to reveal one image over the other.
Diff Just the newer screenshot, with changed regions outlined.

Alongside the images, a Change signals panel breaks down what actually changed:

Signal What it means
Visual mismatch The percentage of the page that changed pixel-for-pixel.
Text How much of the visible text changed, and roughly how many lines were added/removed.
Metadata Whether the page title, description, canonical URL, or similar tags changed.
Status / Redirect / Availability Whether the page's HTTP status, redirect target, or basic reachability changed — these count as a change even if the screenshot looks pixel-identical.

Change score

A single 0–100 number summarizing how much changed overall, combining all the signals above — a few extra points for a metadata change, more for a status or redirect change, and a fixed 100 whenever the page's availability changed (going down, or coming back up, is always treated as maximally significant regardless of what else changed).

Reviewing a comparison

Every comparison can be marked:

Status Meaning
Unreviewed Not looked at yet.
Approved Looked at, no action needed.
Dismissed Looked at, not relevant.
Ignored Looked at, and you don't want to see this specific region change again.

Choosing Ignored offers to create an ignore rule for the changed region so future runs stop flagging it — this always targets whichever region is currently highlighted first, scoped to the monitor you're reviewing. If you need a differently-scoped ignore rule, or want to review the whole queue at once instead of one comparison at a time, use the Review screen.

Only comparisons produced by a Monitor show up in the Review queue. A manual comparison you create from the Captures tab is reviewable too, but you'll find it back on that capture job's page — it won't appear in Review.