BackstopJS vs Argos

Where Argos wins

The approval becomes a real gate

With BackstopJS, "the visual suite passed" means whoever ran it looked at a local report — or didn't. Argos posts a status on the pull request and requires someone to approve the changed screenshots before it goes green. That's the difference between visual testing as a personal habit and visual testing as a process, and it's the reason most teams pay for anything in this category at all.

Published prices you can budget against

Free to 5,000 screenshots, Pro from $100 a month including 35,000, overage at $0.004 per screenshot and $0.0015 for Storybook screenshots. Compare that to Chromatic's $179 entry tier for the same 35,000, Happo's $149 for 50,000, and Percy and Applitools, neither of which publishes paid prices at all. Argos is the cheapest way to stop hosting baselines yourself without a sales call.

Billed on successful builds only

A small clause that matters more than it looks. Visual suites run on every push, and a chain of failed builds on a broken branch would otherwise burn quota for screenshots nobody will ever look at. This is the kind of pricing detail that separates "we metered the obvious thing" from "we've watched how CI actually behaves".

It doesn't care how you capture

Because Argos only ingests images, your capture step stays yours: Playwright, Cypress, Storybook or a script. Practically, that makes it composable with the migration on the Playwright page — move capture to Playwright for the pinned browser, send the results to Argos for the review workflow, and you've addressed both BackstopJS complaints without adopting a single monolithic product.

Where BackstopJS still wins

It captures, and Argos doesn't. This is easy to miss when comparing feature lists: replacing BackstopJS with Argos is not one swap but two, because you still need something to take the screenshots. If you have no Playwright or Cypress suite, the real project is "adopt a test framework and a review service", and the honest cost is higher than $100 a month.

Nothing leaves your infrastructure. BackstopJS runs on your machines and writes to your repo. Argos means uploading screenshots of unreleased pages — potentially with test data, internal dashboards or customer names rendered in them — to a third party. For some teams that's a procurement conversation that outlasts the migration itself, and there is no self-hosted Argos to fall back on.

Zero is a hard number to beat. A 30-scenario suite across 4 viewports on a repo with 200 builds a month is roughly 24,000 screenshots — inside the Pro tier, so $1,200 a year to review diffs you were reviewing for free. That's defensible if the review gate prevents one bad release. It's harder to defend if one person was going to look at the report either way.

Migrating BackstopJS → Argos

The path most teams take:

1. Inventory first. Count screenshots per run (scenarios × viewports) and multiply by builds per month — that number, not your scenario count, decides which tier you land on. Then check which of your scenarios render anything you'd rather not upload.
2. Move capture before you move review. Argos needs an upload step, so pick the capture tool first — usually Playwright — and get it producing the same screenshots BackstopJS was. Only then add the Argos SDK. Doing both at once means you can't tell which change caused a new diff.
3. Run both for one cycle. Keep the BackstopJS job alongside the Argos build for a release or two and compare what each one catches. Expect the first Argos baseline approval to be a large one — that's the cutover, not a bug.
4. What does not move. Your baseline history stays in git as bitmap commits; Argos starts from an empty baseline on its first build. Nothing is lost, but "what did this page look like six months ago?" will be answerable in two different places depending on the date, so decide whether to keep the old bitmaps in the repo or delete them and lose the pre-migration record entirely.